<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: videocompress</title>
    <description>The latest articles on DEV Community by videocompress (@videocompress).</description>
    <link>https://dev.to/videocompress</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4121579%2F53243b52-1101-43e9-85d2-65f6320b2364.png</url>
      <title>DEV Community: videocompress</title>
      <link>https://dev.to/videocompress</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/videocompress"/>
    <language>en</language>
    <item>
      <title>25 of 35 Fails by One: Two Scripted Runs of the 2026–2030 FCC Technician Mock and What the Subelement Grid Shows</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Mon, 16 Nov 2026 02:09:01 +0000</pubDate>
      <link>https://dev.to/videocompress/25-of-35-fails-by-one-two-scripted-runs-of-the-2026-2030-fcc-technician-mock-and-what-the-2eoo</link>
      <guid>https://dev.to/videocompress/25-of-35-fails-by-one-two-scripted-runs-of-the-2026-2030-fcc-technician-mock-and-what-the-2eoo</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://dev.to/videocompress/25-of-35-fails-by-one-two-scripted-runs-of-the-2026-2030-fcc-technician-mock-and-what-the-2eoo"&gt;DEV.to&lt;/a&gt; — tested with &lt;a href="https://morsepass.com" rel="noopener noreferrer"&gt;MorsePass&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The FCC Technician exam has 35 questions and you need 26 right. So you can miss nine. Miss ten and you fail, and the score alone won't tell you which ten were the expensive ones.&lt;/p&gt;

&lt;p&gt;I wanted to see what a free practice exam actually shows you when that happens. So on Monday, November 16, 2026, I sat the &lt;a href="https://morsepass.com/fcc-technician-practice-test-2026" rel="noopener noreferrer"&gt;free FCC Technician practice test&lt;/a&gt; on MorsePass twice: once answering every question from the official answer key, and once missing ten questions on purpose. Both runs used the pool that has been in effect since July 1, 2026.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Disclosure:&lt;/strong&gt; I build MorsePass. A Playwright script in headless Chrome entered the answers, not my fingers, so the timings below are machine times and not human ones. Here is how the answers were picked. In run A, every answer came from the answer letters in the NCVEC's public pool document. In run B, the script picked the next letter after the correct one for every question in subelements T5 and T6, plus the first two T7 questions it met. Everything else in run B came from the official key.&lt;/p&gt;

&lt;h2&gt;
  
  
  First, the two numbers that matter (checked against the sources)
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Pass line:&lt;/strong&gt; 47 CFR &lt;a href="https://www.ecfr.gov/current/title-47/chapter-I/subchapter-D/part-97/subpart-F/section-97.503" rel="noopener noreferrer"&gt;§ 97.503(a)&lt;/a&gt; says Element 2 has "35 questions… The minimum passing score is 26 questions answered correctly" (eCFR, read November 16, 2026).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Pool:&lt;/strong&gt; NCVEC released the 2026–2030 Technician pool on December 18, 2025, and issued errata on February 19, 2026. The errata reworded four questions: T1C01, T5A05, T7A09 and T0A10. ARRL's write-up says the &lt;a href="https://www.arrl.org/news/ncvec-question-pool-committee-issues-revision-to-2026-2030-technician-pool" rel="noopener noreferrer"&gt;revised 2026–2030 Technician pool&lt;/a&gt; is valid for exams from July 1, 2026 through June 30, 2030.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;I parsed the Feb 19, 2026 pool PDF and counted &lt;strong&gt;409 questions&lt;/strong&gt; across ten subelements. MorsePass says the same thing at the bottom of the exam page:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0cp09rvbvgolgz60s91s.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0cp09rvbvgolgz60s91s.png" alt="MorsePass exam page footer: all 409 questions in the February 19, 2026 Technician pool release; 35 questions; at least 26 of 35 to pass; pool effective July 1, 2026 through June 30, 2030" width="800" height="417"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The site advertises "1,431 FCC questions." Its question sitemap adds up to that: 409 Technician + 423 General + 599 Extra = 1,431. Only the 409 Technician questions matter here. One more check: question T7A09 was in my run B exam, and the site shows the corrected February wording.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: open the exam page (no account)
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flc8pw7ofzpl45o9u3c07.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flc8pw7ofzpl45o9u3c07.png" alt="Free FCC Technician Practice Test 2026 page on MorsePass with the Start exam button and an official question preview (T1A01)" width="800" height="538"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;There's no sign-up and no email. The &lt;strong&gt;Start exam&lt;/strong&gt; button builds a 35-question set right away. The &lt;a href="https://morsepass.com/ham-radio-practice-test" rel="noopener noreferrer"&gt;ham radio practice test hub&lt;/a&gt; lists the three classes side by side: Technician 35/26 (pool 2026–2030), General 35/26 (pool 2023–2027) and Amateur Extra 50/37 (pool 2024–2028).&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: answer blind
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftac8nboc1p05jrhd7hb4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftac8nboc1p05jrhd7hb4.png" alt="Question 1 (T1C11) on the MorsePass Technician mock with choice D selected and no right/wrong feedback" width="800" height="491"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;When you click a choice, it just gets a highlight. There's no green tick and no red cross, which is how the real exam works. Each card shows the pool ID (T1C11 here), so you can look the question up later. In all 32 questions I could compare word for word against my pool parse, the A–D order matched the pool document. The script couldn't cleanly line up the other three.&lt;/p&gt;

&lt;p&gt;The "Auto-advance after each answer" box starts out &lt;strong&gt;checked&lt;/strong&gt;. My script unticked it and pressed &lt;strong&gt;Next question&lt;/strong&gt; itself, so the pace stayed predictable.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: move through the palette
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fa5b9dh2n1d9f8gh816jv.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fa5b9dh2n1d9f8gh816jv.png" alt="Palette view at question 18 (T8B04): answered numbers outlined, current question highlighted, Next/Previous/Score official exam buttons" width="800" height="621"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The 35-square palette lets you jump around and see what's still empty. &lt;strong&gt;Score official exam&lt;/strong&gt; is already clickable on question 1. Every set follows the official structure of one question from each of the 35 groups. That means the subelement weights below are fixed and never random.&lt;/p&gt;

&lt;h2&gt;
  
  
  Run A: the answer-key run, 35/35
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv89rwpyx6e89lmpqii8f.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv89rwpyx6e89lmpqii8f.png" alt="Official blind mock result: Passing score, 35 of 35 correct, every subelement tile full" width="800" height="382"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The result card said &lt;strong&gt;"Passing score · 35 of 35 correct · You cleared the 26 question passing line,"&lt;/strong&gt; with every tile full, from T0 3/3 to T9 2/2. Above the card: &lt;strong&gt;"Exam readiness: 100%."&lt;/strong&gt; Every one of the 35 answers came from the NCVEC key, and the site marked all 35 correct. So for that set, the site's key agreed with the official one on every question. The whole script finished in under 23 seconds, page load included.&lt;/p&gt;

&lt;h2&gt;
  
  
  Run B: ten deliberate misses, 25/35
&lt;/h2&gt;

&lt;p&gt;The script started run B at 10:02 (UTC+8). It was a fresh browser with a new random set, the same rules, and 10 wrong answers placed on purpose. From the click on Start to the result took 30.1 s, including four screenshots.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdsekk9zr94ivwy2lhm1q.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdsekk9zr94ivwy2lhm1q.png" alt="Official blind mock result: Keep drilling weak groups, 25 of 35 correct, You need 1 more correct answers to pass; T5 0/4, T6 0/4, T7 2/4" width="800" height="381"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Headline: &lt;strong&gt;"Keep drilling weak groups · 25 of 35 correct · You need 1 more correct answers to pass."&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;Grid: T0 3/3 · T1 6/6 · T2 3/3 · T3 3/3 · T4 2/2 · &lt;strong&gt;T5 0/4 · T6 0/4 · T7 2/4&lt;/strong&gt; · T8 4/4 · T9 2/2&lt;/li&gt;
&lt;li&gt;Readiness: &lt;strong&gt;71%&lt;/strong&gt;, with the note "Readiness turns green after three attempts at 74% or better."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;I also had the script compare all 35 questions in this set against the official key. There were 0 mismatches.&lt;/p&gt;

&lt;p&gt;Here is the same run laid out against the syllabus weights:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzjwvxuyje48te5ugki6h.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzjwvxuyje48te5ugki6h.png" alt="Chart: exam questions per subelement (NCVEC syllabus) vs run B correct; T5 0/4, T6 0/4, T7 2/4, all others full" width="800" height="520"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;This is why the grid matters more than the score. T5 Electrical Principles, T6 Components and T7 Practical Circuits are only three of the ten subelements, but together they put &lt;strong&gt;12 questions&lt;/strong&gt; on every exam. A learner who skips "the electronics part" has already spent more than the nine-miss budget.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: read "What to review next"
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fxb514bl6y8mdatt8o83x.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fxb514bl6y8mdatt8o83x.png" alt="What to review next: Equipment and circuits 6 missed, Calculations 3 missed, Schematics and figures 1 missed, with lesson buttons" width="800" height="509"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Under the grid, MorsePass sorts the misses into its own buckets: &lt;strong&gt;Equipment and circuits 6 · Calculations 3 · Schematics and figures 1&lt;/strong&gt;. Then it links three lessons: Components and schematics, Electricity basics, and &lt;a href="https://morsepass.com/lessons/technician/ohms-law-and-power-math" rel="noopener noreferrer"&gt;Ohm's law and power math&lt;/a&gt;. The page says outright that this is "guidance, not an AI-generated answer." These buckets are the site's own groupings. They aren't NCVEC subelements, so use the grid for "where" and the buckets for "what kind of thing."&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: open the review book
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fb59dgmyjav3f2n4pkx15.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fb59dgmyjav3f2n4pkx15.png" alt="Review missed Technician questions page listing T5A11, T5B09 and T5C03 with the correct answer for each" width="800" height="827"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Review missed questions&lt;/strong&gt; goes to &lt;a href="https://morsepass.com/review" rel="noopener noreferrer"&gt;the local review book&lt;/a&gt;. It listed all ten misses: T5A11, T5B09, T5C03, T5D14, T6A11, T6B06, T6C04, T6D02, T7A09 and T7B04. Each one shows its correct answer, for example "3 dB" for the 5 W → 10 W question (T5B09), "Carbon-zinc" for the non-rechargeable battery (T6A11), and "High SWR" for low RF output (T7B04). The page says that questions "leave this list after you answer them correctly in a later exam," and that the history stays on your device. To dig into one subelement, the &lt;a href="https://morsepass.com/subelement/T5" rel="noopener noreferrer"&gt;T5 question review&lt;/a&gt; shows all 50 T5 pool questions with their answers.&lt;/p&gt;

&lt;h2&gt;
  
  
  The free limit you'll hit
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd3hcdmg74so1cse0r5mj.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd3hcdmg74so1cse0r5mj.png" alt="You've used today's free simulated exam. Come back tomorrow for another complete exam or unlock unlimited practice." width="800" height="400"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The free tier on &lt;a href="https://morsepass.com/pricing" rel="noopener noreferrer"&gt;the pricing page&lt;/a&gt; says "1 complete 35-question simulated exam per day." I hit that card while I was setting up the script. A later start that morning still went through, which is how I got run B. Plan on one scored mock per day unless you pay. The question browser, the review book and the subelement pages kept loading normally.&lt;/p&gt;

&lt;h2&gt;
  
  
  How I'd use this as a learner
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Take one blind mock a day. Write down the grid, not just the score.&lt;/li&gt;
&lt;li&gt;Any subelement under half is your next study block. Under the syllabus T1 has 6 questions, T5 to T8 have 4 each, T0/T2/T3 have 3, and T4/T9 have 2.&lt;/li&gt;
&lt;li&gt;Clear the review book before the next mock. Misses only drop off when you get them right later.&lt;/li&gt;
&lt;li&gt;For calculation misses (dB, Ohm's law), drill the formulas with the &lt;a href="https://morsepass.com/ham-radio-ohms-law-calculator" rel="noopener noreferrer"&gt;Ohm's law calculator&lt;/a&gt; or the &lt;a href="https://morsepass.com/technician-exam-cheat-sheet" rel="noopener noreferrer"&gt;Technician cheat sheet&lt;/a&gt;. Don't keep re-reading answer letters.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Numbers at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;/th&gt;
&lt;th&gt;Run A (answer key)&lt;/th&gt;
&lt;th&gt;Run B (10 deliberate misses)&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Score&lt;/td&gt;
&lt;td&gt;35/35, pass&lt;/td&gt;
&lt;td&gt;25/35, fail by 1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Headline&lt;/td&gt;
&lt;td&gt;"Passing score"&lt;/td&gt;
&lt;td&gt;"Keep drilling weak groups"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Weak tiles&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;td&gt;T5 0/4, T6 0/4, T7 2/4&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Readiness&lt;/td&gt;
&lt;td&gt;100%&lt;/td&gt;
&lt;td&gt;71%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Site key vs NCVEC key&lt;/td&gt;
&lt;td&gt;35/35 agree&lt;/td&gt;
&lt;td&gt;35/35 agree&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Machine time&lt;/td&gt;
&lt;td&gt;&amp;lt; 23 s whole script&lt;/td&gt;
&lt;td&gt;30.1 s Start → result&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Account&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Pool: NCVEC 2026–2030 Technician, Feb 19, 2026 errata release, 409 questions, effective 2026-07-01 to 2030-06-30. Pass line: 26/35 under 47 CFR 97.503(a).&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Earlier in this series:&lt;/em&gt; &lt;a href="https://dev.to/videocompress/koch-morse-from-zero-in-a-browser-tab-100-on-k-and-m-88-when-r-arrives-and-what-farnsworth-3gjn"&gt;Koch Morse from zero in a browser tab&lt;/a&gt; covers the CW side of &lt;a href="https://morsepass.com/" rel="noopener noreferrer"&gt;MorsePass&lt;/a&gt; (Koch levels and Farnsworth spacing). If your other exam is a quant-trading mental-math test, &lt;a href="https://quantsprint.com/" rel="noopener noreferrer"&gt;QuantSprint&lt;/a&gt; runs the same kind of timed browser drill for that.&lt;/p&gt;

&lt;p&gt;— mengyuxuan&lt;/p&gt;

</description>
      <category>hamradio</category>
      <category>learning</category>
      <category>webdev</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>CRF 23 Gave Me 15 MB on One Clip and 30 MB on Another: "Compress by Quality" vs "Compress to a Size", Measured</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Wed, 11 Nov 2026 02:51:00 +0000</pubDate>
      <link>https://dev.to/videocompress/crf-23-gave-me-15-mb-on-one-clip-and-30-mb-on-another-compress-by-quality-vs-compress-to-a-143p</link>
      <guid>https://dev.to/videocompress/crf-23-gave-me-15-mb-on-one-clip-and-30-mb-on-another-compress-by-quality-vs-compress-to-a-143p</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://dev.to/videocompress/crf-23-gave-me-15-mb-on-one-clip-and-30-mb-on-another-compress-by-quality-vs-compress-to-a-143p"&gt;DEV.to&lt;/a&gt; — tested with &lt;a href="https://vidsmaller.com" rel="noopener noreferrer"&gt;VidSmaller&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Most video compressors offer two ways to ask for a smaller file. One is &lt;strong&gt;"keep this quality"&lt;/strong&gt; (a CRF number). The other is &lt;strong&gt;"make it this size"&lt;/strong&gt; (a target in MB). They sound like two routes to the same place. They aren't, and the difference only shows up once you run both on more than one clip.&lt;/p&gt;

&lt;p&gt;So I did: three 60-second 1080p clips from Blender's open movies, five real cloud compressions, every output downloaded and measured against the upload with ffmpeg's SSIM and PSNR filters.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The short version:&lt;/strong&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Clip (upload)&lt;/th&gt;
&lt;th&gt;CRF 23 ("quality")&lt;/th&gt;
&lt;th&gt;Target 20 MB ("size")&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Tears of Steel, 101 MB&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;15.1 MB&lt;/strong&gt; · SSIM 0.9897&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;18.7 MB&lt;/strong&gt; · SSIM 0.9909&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Sintel, 77.3 MB&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;30.2 MB&lt;/strong&gt; · SSIM 0.9844&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;18.7 MB&lt;/strong&gt; · SSIM 0.9764&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Big Buck Bunny, 17.4 MB&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;16.3 MB&lt;/strong&gt; (−6.1%) · SSIM 0.9914&lt;/td&gt;
&lt;td&gt;not run&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Same CRF, &lt;strong&gt;2× different file sizes&lt;/strong&gt;. Same target, &lt;strong&gt;identical sizes, different quality&lt;/strong&gt;. That's the whole trade-off, and the rest of this post shows how to pick.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I build &lt;a href="https://vidsmaller.com" rel="noopener noreferrer"&gt;VidSmaller&lt;/a&gt;, the cloud compressor used for every run here. All numbers are from these runs; nothing is estimated unless I say so.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The setup
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Clips&lt;/strong&gt; (all CC BY 3.0, © Blender Foundation): a 60 s cut of &lt;a href="https://studio.blender.org/films/tears-of-steel/" rel="noopener noreferrer"&gt;Tears of Steel&lt;/a&gt; (live action + VFX, 1920×800, 106,098,955 bytes), a 60 s cut of &lt;a href="https://studio.blender.org/films/sintel/" rel="noopener noreferrer"&gt;Sintel&lt;/a&gt; (animation, 1920×818, 81,101,968 bytes) and a 60 s cut of &lt;a href="https://studio.blender.org/films/big-buck-bunny/" rel="noopener noreferrer"&gt;Big Buck Bunny&lt;/a&gt; (animation, 1920×1080, 18,210,139 bytes — already a lean 2.4 Mbps file).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Tool:&lt;/strong&gt; vidsmaller.com → &lt;strong&gt;Advanced Compression&lt;/strong&gt;, H.264, speed "Faster (default)", MP4. Driven by a headless Chrome, signed in to my own account.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Measurement:&lt;/strong&gt; &lt;code&gt;stat&lt;/code&gt; for bytes, &lt;code&gt;ffprobe&lt;/code&gt; for bitrate, the x264 settings string inside each output (it records &lt;code&gt;rc=crf crf=23.0&lt;/code&gt; or &lt;code&gt;rc=2pass bitrate=…&lt;/code&gt;), and ffmpeg &lt;code&gt;ssim&lt;/code&gt; + &lt;code&gt;psnr&lt;/code&gt; against the exact file I uploaded.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;One note on units before the numbers: the app shows sizes in binary megabytes. "18.7 MB" on the card is 19,624,758 bytes (18.7 MiB, 19.6 decimal MB). Either way it's under 20.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Find the two modes
&lt;/h2&gt;

&lt;p&gt;On the &lt;a href="https://vidsmaller.com/" rel="noopener noreferrer"&gt;homepage compressor&lt;/a&gt;, switch to &lt;strong&gt;Advanced Compression&lt;/strong&gt; and open &lt;strong&gt;Compression Method&lt;/strong&gt;. "Target a video quality" is CRF mode; "Target a file size" is the size mode. (The other three — percentage, resolution, max bitrate — are out of scope here.)&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9gyw8ti0bad397s5799m.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9gyw8ti0bad397s5799m.png" alt="VidSmaller Advanced Compression: the Compression Method menu with Target a file size, Target a video quality, percentage, resolution and max bitrate" width="800" height="549"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The hint under the menu is the one-line version of this article: &lt;em&gt;"Pick a file size for an exact ceiling, or a video quality to let the size follow."&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: CRF 23 on the live-action clip
&lt;/h2&gt;

&lt;p&gt;I loaded the 101 MB Tears of Steel cut, chose &lt;strong&gt;Target a video quality&lt;/strong&gt;, and left &lt;strong&gt;CRF 23 · Recommended (default)&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F2h8hl621di5rykbyy3fc.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F2h8hl621di5rykbyy3fc.png" alt="Tears of Steel 101 MB loaded, Target a video quality, CRF 23 selected, Estimated result 90.1 MB" width="800" height="589"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Look at the line under the file name: &lt;strong&gt;Estimated result: 90.1 MB&lt;/strong&gt;. Keep that number in mind.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7dc4921gyv57j2p1us68.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7dc4921gyv57j2p1us68.png" alt="Result card: 101 MB → 15.1 MB, −85.1%" width="800" height="326"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;101 MB → 15.1 MB (−85.1%)&lt;/strong&gt;. Video bitrate 1.97 Mbps, SSIM &lt;strong&gt;0.9897&lt;/strong&gt;, PSNR 46.26 dB. The x264 string confirms &lt;code&gt;rc=crf crf=23.0&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;The estimate was off by 6×. In quality mode nobody can know the size before encoding — that's the point of CRF — so treat the pre-run estimate as a placeholder. (In my three CRF runs it was 90.1, 68.9 and 15.5 MB, about 89% of each upload every time.)&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Same CRF 23 on the animated clip
&lt;/h2&gt;

&lt;p&gt;Same settings, Sintel (77.3 MB on the card).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6we3cyxlozy5yp3lp0c9.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6we3cyxlozy5yp3lp0c9.png" alt="Sintel 77.3 MB loaded with CRF 23, Estimated result 68.9 MB" width="800" height="589"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5igpg41zueg3ye4ifp0u.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5igpg41zueg3ye4ifp0u.png" alt="Result card: 77.3 MB → 30.2 MB, −60.9%" width="800" height="326"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;77.3 MB → 30.2 MB (−60.9%)&lt;/strong&gt;. Video bitrate &lt;strong&gt;4.09 Mbps&lt;/strong&gt; — more than double Tears of Steel's 1.97 Mbps, from a &lt;em&gt;smaller&lt;/em&gt; upload. SSIM 0.9844, PSNR 43.77 dB.&lt;/p&gt;

&lt;p&gt;This is what CRF means: "spend whatever bits this content needs to look like CRF 23". Sintel's sand, gravel and fast camera moves need more bits than Tears of Steel's dark interiors, so the file came out twice as big. Quality across the two clips stayed close (SSIM within 0.0053); size didn't.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: CRF 23 on a file that's already compressed
&lt;/h2&gt;

&lt;p&gt;Big Buck Bunny's 60 s cut is only 17.4 MB (2.4 Mbps for 1080p30 — already lean).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3m1pekob9tgp7tbz8ky7.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3m1pekob9tgp7tbz8ky7.png" alt="Big Buck Bunny 17.4 MB loaded with CRF 23, Estimated result 15.5 MB" width="800" height="589"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F28moodoufg2rfnh30bci.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F28moodoufg2rfnh30bci.png" alt="Recent compressions: big-buck-bunny 17.4 MB → 16.3 MB −6.1%, tears-of-steel 101 MB → 15.1 MB −85.1%" width="800" height="120"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;17.4 MB → 16.3 MB, just −6.1%.&lt;/strong&gt; SSIM 0.9914. CRF has no idea what size you need; if the source is already near "CRF 23 quality", you get almost nothing back. If you need a phone clip under a hard limit, CRF can leave you above it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Target 20 MB on both big clips
&lt;/h2&gt;

&lt;p&gt;Now the other mode. &lt;strong&gt;Target a file size&lt;/strong&gt;, typed &lt;strong&gt;20&lt;/strong&gt; in the MB box.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fe5v12x6vzm7fxtdk99v6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fe5v12x6vzm7fxtdk99v6.png" alt="Tears of Steel with Target a file size, 20 MB typed, Estimated result 20.0 MB" width="800" height="589"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Frnp5j7fh5sppa86cburz.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Frnp5j7fh5sppa86cburz.png" alt="Result card: 101 MB → 18.7 MB, −81.5%" width="800" height="326"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Tears of Steel: &lt;strong&gt;101 MB → 18.7 MB (−81.5%)&lt;/strong&gt;, &lt;code&gt;rc=2pass bitrate=2463&lt;/code&gt;. SSIM &lt;strong&gt;0.9909&lt;/strong&gt;, PSNR 47.00 dB.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flsehdg13wkplra2xw254.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flsehdg13wkplra2xw254.png" alt="Sintel with Target a file size, 20 MB" width="800" height="589"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzz8fc4aq5ej5wtakwqww.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzz8fc4aq5ej5wtakwqww.png" alt="Result card: 77.3 MB → 18.7 MB, −75.8%" width="800" height="326"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Sintel: &lt;strong&gt;77.3 MB → 18.7 MB (−75.8%)&lt;/strong&gt;, &lt;code&gt;rc=2pass bitrate=2466&lt;/code&gt;. SSIM &lt;strong&gt;0.9764&lt;/strong&gt;, PSNR 41.38 dB.&lt;/p&gt;

&lt;p&gt;Both landed at &lt;strong&gt;18.7 MB&lt;/strong&gt; (19,624,758 and 19,612,900 bytes) — a two-pass encode at ~2.46 Mbps video either way. The size held; the quality moved.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Put the four numbers side by side
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7mt5p4whscdrpu2geas5.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7mt5p4whscdrpu2geas5.png" alt="Chart: Tears of Steel CRF 23 15.1 MB SSIM 0.9897, target 20 18.7 MB SSIM 0.9909; Sintel CRF 23 30.2 MB SSIM 0.9844, target 20 18.7 MB SSIM 0.9764" width="800" height="300"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Two things fall out:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;On Tears of Steel, the 20 MB target was &lt;em&gt;more&lt;/em&gt; than the clip needed.&lt;/strong&gt; It spent 24% more bytes than CRF 23 (18.7 vs 15.1 MB) for +0.0012 SSIM. If your limit is 20 MB, CRF 23 already fit with room to spare.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;On Sintel, the 20 MB target cost real quality.&lt;/strong&gt; It's 38% smaller than CRF 23 (18.7 vs 30.2 MB), and SSIM drops from 0.9844 to 0.9764 (PSNR −2.4 dB).&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;What does that drop look like? Here's the worst frame of the Sintel size run (frame 1397, 0:58.2), cropped at 100%:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5rqyyhvwporwid8vzmal.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5rqyyhvwporwid8vzmal.png" alt="100% crop of Sintel frame 1397: upload, CRF 23 30.2 MB (frame SSIM 0.965), target 20 MB 18.7 MB (frame SSIM 0.934); gravel texture smoother in the 20 MB version" width="800" height="228"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The gravel in the 20 MB version is visibly smoother and the shadow edges are softer. Fine on a phone in a chat, noticeable on a monitor. Most frames are much closer than this one.&lt;/p&gt;

&lt;h2&gt;
  
  
  So which one should you use?
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Use "target a file size" when there's a hard number:&lt;/strong&gt; a chat app's upload cap, an email attachment limit, a form that rejects anything over X MB. You'll always fit; the price is that busy footage gets softer. The &lt;a href="https://vidsmaller.com/compress-video-for-telegram" rel="noopener noreferrer"&gt;Telegram&lt;/a&gt;, &lt;a href="https://vidsmaller.com/compress-video-for-slack" rel="noopener noreferrer"&gt;Slack&lt;/a&gt; and &lt;a href="https://vidsmaller.com/compress-video-for-twitter" rel="noopener noreferrer"&gt;X/Twitter&lt;/a&gt; pages walk through each app's limits.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Use "target a video quality" (CRF) when there's no hard limit&lt;/strong&gt; — archiving, a site you host yourself, a long &lt;a href="https://vidsmaller.com/compress-zoom-recording" rel="noopener noreferrer"&gt;meeting recording&lt;/a&gt; you just want lighter. You get consistent quality and the smallest file that delivers it. On easy content that can be much smaller than any target you'd have guessed (Tears of Steel: 15.1 MB without asking).&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Check two things before trusting CRF on a deadline:&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Is the source already lean? Big Buck Bunny only dropped 6.1%. If a 17 MB file must go under 15, CRF 23 won't do it — a size target will.&lt;/li&gt;
&lt;li&gt;Does the result happen to fit? Run CRF first, and only switch to a size target if it comes out too big. That's what I'd do with a &lt;a href="https://vidsmaller.com/compress-4k-video" rel="noopener noreferrer"&gt;4K phone clip&lt;/a&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you want to go past the default, VidSmaller's &lt;a href="https://vidsmaller.com/blog/crf-explained-what-number-to-use" rel="noopener noreferrer"&gt;CRF explainer&lt;/a&gt; covers what 18, 26 or 28 trade away.&lt;/p&gt;

&lt;h2&gt;
  
  
  Practical notes from the runs
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Each run here was 1 credit&lt;/strong&gt; (the button label said "Cost 1 credit"). A signed-out guest gets one compression (up to 200 MB / 3 min) — enough to try one mode on one clip. A free account adds one welcome credit (500 MB). Re-running this five-job comparison needs a pack (&lt;a href="https://vidsmaller.com/pricing" rel="noopener noreferrer"&gt;pricing&lt;/a&gt;: $3 for 50 credits).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Timing:&lt;/strong&gt; click to "Your video is ready", including the upload, took 30–45 s for the runs I timed.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Surround audio:&lt;/strong&gt; my first Sintel cut kept the film's 6-channel audio and was rejected with "This is not a readable video file". A stereo downmix of the same video went through, and the failed attempt's credit was refunded automatically. If you hit that message with a movie-style file, try stereo audio first.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Sintel was re-exported&lt;/strong&gt; to fix that (x264 CRF 16, 10.8 Mbps), so its SSIM numbers are measured against that upload, not the original film master.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you'd rather not upload at all, my other tool, &lt;a href="https://videocompress.dev" rel="noopener noreferrer"&gt;videocompress.dev&lt;/a&gt;, does target-size compression inside the browser tab.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Credits: Tears of Steel (mango.blender.org), Sintel (durian.blender.org) and Big Buck Bunny (peach.blender.org) © Blender Foundation, &lt;a href="https://creativecommons.org/licenses/by/3.0/" rel="noopener noreferrer"&gt;CC BY 3.0&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Related:&lt;/strong&gt; &lt;a href="https://videocompress-guides.hashnode.dev/h265-vs-h264-measured-before-you-pay" rel="noopener noreferrer"&gt;Before you pay for H.265: measured against a real H.264 encode&lt;/a&gt; (Hashnode) · &lt;a href="https://videocompress-guides.hashnode.dev/discord-20mb-one-chip-tap-178mb-to-1868mb" rel="noopener noreferrer"&gt;Discord's 20 MB limit with one preset tap&lt;/a&gt; (Hashnode) · &lt;a href="https://medium.com/@videocompress/i-typed-48-18-and-9-mb-into-a-video-compressor-heres-where-a-227-mb-clip-actually-landed-acb7f8f85466" rel="noopener noreferrer"&gt;I typed 48, 18 and 9 MB into a video compressor&lt;/a&gt; (Medium)&lt;/p&gt;

&lt;p&gt;&lt;em&gt;— mengyuxuan&lt;/em&gt;&lt;/p&gt;

</description>
      <category>video</category>
      <category>ffmpeg</category>
      <category>webdev</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>Koch Morse From Zero in a Browser Tab: 100% on K and M, 88% When R Arrives, and What Farnsworth Spacing Really Changes</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Sun, 08 Nov 2026 02:17:32 +0000</pubDate>
      <link>https://dev.to/videocompress/koch-morse-from-zero-in-a-browser-tab-100-on-k-and-m-88-when-r-arrives-and-what-farnsworth-3gjn</link>
      <guid>https://dev.to/videocompress/koch-morse-from-zero-in-a-browser-tab-100-on-k-and-m-88-when-r-arrives-and-what-farnsworth-3gjn</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://dev.to/videocompress/koch-morse-from-zero-in-a-browser-tab-100-on-k-and-m-88-when-r-arrives-and-what-farnsworth-3gjn"&gt;DEV Community&lt;/a&gt; and tested with &lt;a href="https://morsepass.com/" rel="noopener noreferrer"&gt;MorsePass&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Most "learn Morse code" advice either starts with a printed chart (which teaches you to count dits) or says "use the Koch method" without showing what a session looks like. So I sat down with a free Koch trainer, no account, and logged every number: which characters you start with, the speeds, the accuracy the trainer reports, and what it tells you to do next.&lt;/p&gt;

&lt;p&gt;Three 50-character drills on the &lt;a href="https://morsepass.com/koch-method-trainer" rel="noopener noreferrer"&gt;MorsePass Koch method trainer&lt;/a&gt;:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Run&lt;/th&gt;
&lt;th&gt;Level and set&lt;/th&gt;
&lt;th&gt;Speed (character/effective)&lt;/th&gt;
&lt;th&gt;Result&lt;/th&gt;
&lt;th&gt;What the trainer said next&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;Level 2: K M&lt;/td&gt;
&lt;td&gt;20/12 WPM&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;100%&lt;/strong&gt;, 50 of 50&lt;/td&gt;
&lt;td&gt;"Level 3 unlocked."&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;Level 3: K M R&lt;/td&gt;
&lt;td&gt;20/12 WPM&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;88%&lt;/strong&gt;, 44 of 50&lt;/td&gt;
&lt;td&gt;"Repeat this level to reach 90%." Focus next: "R: 6 of 16 missed"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;3&lt;/td&gt;
&lt;td&gt;Level 3: K M R&lt;/td&gt;
&lt;td&gt;20/20 WPM (no Farnsworth)&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;100%&lt;/strong&gt;, 50 of 50&lt;/td&gt;
&lt;td&gt;"Level 4 unlocked."&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;How I got repeatable numbers: I captured the tone schedule the page builds with the Web Audio API and typed each five-letter group only after that group had finished playing, the way a human copies. In run 2 I &lt;strong&gt;deliberately&lt;/strong&gt; typed K for five R's and skipped one R to see how the trainer handles the classic K/R confusion. Run 1 and run 3 are straight copies. (Disclosure: I build MorsePass.)&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Open the trainer and look at what "from zero" means
&lt;/h2&gt;

&lt;p&gt;There's no sign-up step. The trainer opens on &lt;strong&gt;Level 2&lt;/strong&gt;, which means your unlocked set is just &lt;strong&gt;K and M&lt;/strong&gt;. Level 1 is K alone. Each level adds one character, in this order on MorsePass: K, M, R, S, U, A, P, T, L, O, and so on up to 40. Levels 1–5 are free; the padlocks from level 6 onward are the paid part (see &lt;a href="https://morsepass.com/pricing" rel="noopener noreferrer"&gt;the plan table&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3gmmzs7qagkiyh7j010d.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3gmmzs7qagkiyh7j010d.png" alt="Koch trainer ready state: Level 2, active character M, 60 s, 0 / 50, levels 1–5 open and 6+ locked" width="800" height="358"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The defaults are what matter for a beginner:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Character speed 20 WPM&lt;/strong&gt;: each letter is sent fast enough that you can't count its dits and dahs. You hear K as one rhythm, not "dah-dit-dah".&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Effective speed 12 WPM&lt;/strong&gt;: extra silence goes between letters and groups, so you have time to decide. That is Farnsworth spacing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Target 90%&lt;/strong&gt;: the bar for moving to the next character.&lt;/li&gt;
&lt;li&gt;650 Hz sidetone, 70% volume, HF static off.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;A drill is 10 groups of 5 random characters from your unlocked set, so 50 characters.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Run a Level 2 drill and copy group by group
&lt;/h2&gt;

&lt;p&gt;Press &lt;strong&gt;Start 60 second drill&lt;/strong&gt; and type what you hear. Spaces between groups are optional (the trainer says they're formatting only).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd9bmemtqz71g1uw7li1l.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd9bmemtqz71g1uw7li1l.png" alt="Mid-drill: NOW SENDING Group 5 of 10, 37 s left, copy progress 20 / 50, live accuracy 40%" width="800" height="262"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;One thing that confused me at first: at this point I had copied 20 of 20 characters correctly, yet &lt;strong&gt;Live accuracy said 40%&lt;/strong&gt;. It counts correct characters against all 50 in the drill, not against what has been sent so far. It climbs to the real figure as you go, so don't panic at 40% in the middle.&lt;/p&gt;

&lt;p&gt;At 20/12 WPM a group arrived roughly every 5.3 seconds. The last tone of the 50th character played &lt;strong&gt;52.8 s&lt;/strong&gt; after the first.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Read the score and the character grid
&lt;/h2&gt;

&lt;p&gt;When the drill ends, press &lt;strong&gt;Score and review copy&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjqhex1n6gdha8fmib8an.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjqhex1n6gdha8fmib8an.png" alt="Run 1 result: 100% accurate, 50 of 50 characters correct, Level 3 unlocked, Focus next: no character confusions" width="800" height="480"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Run 1 (K and M only): &lt;strong&gt;100%, 50 of 50, "Level 3 unlocked."&lt;/strong&gt; The grid shows each sent character on top and your copy below, and "Focus next" reports no confusions.&lt;/p&gt;

&lt;p&gt;Don't over-read one clean run. The trainer's own guide says to move on when "at least three separate sets stay near or above 90 percent". A two-letter set is easy; the next letter is where it gets real.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Add R and watch the confusion show up
&lt;/h2&gt;

&lt;p&gt;Click level 3. The set becomes &lt;strong&gt;K M R&lt;/strong&gt;. K is &lt;code&gt;-.-&lt;/code&gt; and R is &lt;code&gt;.-.&lt;/code&gt;, the same length with mirrored rhythm, which is why they're a famous beginner pair.&lt;/p&gt;

&lt;p&gt;For this run I typed K instead of R five times and dropped one R near the end:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffb9u7bkupvvmvyxstazr.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffb9u7bkupvvmvyxstazr.png" alt="Run 2 result: 88% accurate, 44 of 50 correct, Repeat this level to reach 90%; red cells show R copied as K and one missing R; Focus next: R: 6 of 16 missed" width="800" height="480"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;88%, 44 of 50&lt;/strong&gt;: below the 90% target, so no unlock. The message is "Repeat this level to reach 90%."&lt;/li&gt;
&lt;li&gt;The red cells show &lt;strong&gt;R sent, K copied&lt;/strong&gt; five times and one &lt;strong&gt;R sent, nothing copied&lt;/strong&gt; (the dash).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Focus next: "R: 6 of 16 missed"&lt;/strong&gt;. It counts the misses per sent character, so you can see it's an R problem, not a general one.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That's the most useful thing a Koch trainer can tell you. A score of 88% from one K/R mix-up needs a different fix than 88% from random late misses. The guide's advice: drill the two confusable rhythms against each other at the same character speed, then go back to mixed groups. Don't slow the letters down.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: See what Farnsworth spacing actually does
&lt;/h2&gt;

&lt;p&gt;"Farnsworth" gets explained in a lot of different ways, so I measured it. I captured the Web Audio schedule for the first group of a 20/12 drill and a 20/20 drill:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo6ppvfd1uxzycrkp64wg.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo6ppvfd1uxzycrkp64wg.png" alt="Timeline of the first group: at 20/12 the gap between letters is 0.501 s and the group ends at 4.36 s; at 20/20 the gap is 0.185 s and the group ends at 3.22 s" width="800" height="320"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;/th&gt;
&lt;th&gt;20/12 WPM (default)&lt;/th&gt;
&lt;th&gt;20/20 WPM&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Dit / dah&lt;/td&gt;
&lt;td&gt;55 ms / 175 ms&lt;/td&gt;
&lt;td&gt;55 ms / 175 ms&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Gap inside a letter&lt;/td&gt;
&lt;td&gt;65 ms&lt;/td&gt;
&lt;td&gt;65 ms&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Gap between letters&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;0.501 s&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;0.185 s&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Gap between groups&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;1.162 s&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;0.425 s&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;50 characters, first to last tone&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;52.8 s&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;34.1 s&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;(Tone times are measured between the 5 ms on/off ramps, which is why a 60 ms dit reads as 55 ms and the gaps read ~5 ms long.)&lt;/p&gt;

&lt;p&gt;The letters themselves are &lt;strong&gt;identical&lt;/strong&gt;. Only the silence changes. That matches the ARRL timing standard (&lt;a href="https://www.arrl.org/files/file/Technology/x9004008.pdf" rel="noopener noreferrer"&gt;Jon Bloom, KE3Z, "A Standard for Morse Timing Using the Farnsworth Technique"&lt;/a&gt;): for character speed &lt;em&gt;c&lt;/em&gt; = 20 and overall speed &lt;em&gt;s&lt;/em&gt; = 12, the added delay is (60c − 37.2s)/(s·c) = 3.14 s per word, which works out to &lt;strong&gt;0.496 s&lt;/strong&gt; between characters and &lt;strong&gt;1.157 s&lt;/strong&gt; between words. The browser scheduled 0.501 s and 1.162 s.&lt;/p&gt;

&lt;p&gt;To try it, drag &lt;strong&gt;Effective speed&lt;/strong&gt; up to match character speed:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjzem2n3lc6rgvh7lz2b0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjzem2n3lc6rgvh7lz2b0.png" alt="Settings: Level 3/40, unlocked set K M R, character speed 20 WPM, effective speed 20 WPM" width="798" height="145"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Run 3 (K M R at 20/20, clean copy): &lt;strong&gt;100%, 50 of 50, "Level 4 unlocked."&lt;/strong&gt; But the whole drill is done in about 34 seconds instead of 53. The "60 s" timer is the limit, not the length:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5poi2k6f5253wn4ey8v6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5poi2k6f5253wn4ey8v6.png" alt="Bar chart: 50-character drill lasts 52.8 s at 20/12 versus 34.1 s at 20/20" width="798" height="171"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftnjymlqp8b2p04vp2262.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftnjymlqp8b2p04vp2262.png" alt="Run 3: 100% accurate, 50 of 50 characters correct, Level 4 unlocked" width="798" height="116"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;So the order I'd suggest for a real beginner is:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Keep character speed at 20 (or 18) from day one and never lower it.&lt;/li&gt;
&lt;li&gt;Start with effective speed around 12 so you have about half a second to decide on each letter.&lt;/li&gt;
&lt;li&gt;Raise effective speed only after three sets in a row at 90% or higher on the current level.&lt;/li&gt;
&lt;li&gt;Add the next character only when "Focus next" stops naming the same letter.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Step 6: What stays in your browser
&lt;/h2&gt;

&lt;p&gt;During the three drills I logged every network request. The tones are generated on the page with Web Audio, so no audio file was downloaded (the only GET to morsepass.com during a drill was the touch icon). Scoring made no request to a MorsePass API. The only uploads were the usual analytics beacons (Microsoft Clarity and Google Analytics). On the free plan, progress is stored locally; the paid plans add cloud sync and levels 6–40.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where to go after level 5
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Once K M R S U feel automatic, copy real call signs at the same pitch with the &lt;a href="https://morsepass.com/cw-callsign-trainer" rel="noopener noreferrer"&gt;callsign trainer&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;Use the &lt;a href="https://morsepass.com/morse-code-translator" rel="noopener noreferrer"&gt;Morse code translator&lt;/a&gt; to hear any word or phrase at your current speed and export it as a WAV for offline listening.&lt;/li&gt;
&lt;li&gt;The broader &lt;a href="https://morsepass.com/morse-code-trainer" rel="noopener noreferrer"&gt;CW training path&lt;/a&gt; also has a straight key and paddle timing lab for sending.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you're also drilling timed mental math for trading interviews, &lt;a href="https://quantsprint.com/" rel="noopener noreferrer"&gt;QuantSprint&lt;/a&gt; uses the same "short timed sets, then look at the first mistake" approach.&lt;/p&gt;

</description>
      <category>hamradio</category>
      <category>learning</category>
      <category>webdev</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>Amazon Wants Pure White RGB 255. A Centered Square Crop Deletes a Third of This Product Photo</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Thu, 05 Nov 2026 02:27:22 +0000</pubDate>
      <link>https://dev.to/videocompress/amazon-wants-pure-white-rgb-255-a-centered-square-crop-deletes-a-third-of-this-product-photo-1no4</link>
      <guid>https://dev.to/videocompress/amazon-wants-pure-white-rgb-255-a-centered-square-crop-deletes-a-third-of-this-product-photo-1no4</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://dev.to/videocompress/amazon-wants-pure-white-rgb-255-a-centered-square-crop-deletes-a-third-of-this-product-photo-1no4"&gt;DEV.to&lt;/a&gt; — tested with &lt;a href="https://squareimage.dev" rel="noopener noreferrer"&gt;SquareImage&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;I build SquareImage, a small set of in-browser square-image tools. This week I ran its &lt;a href="https://squareimage.dev/make-image-square-without-cropping/" rel="noopener noreferrer"&gt;make-an-image-square-without-cropping&lt;/a&gt; page against a marketplace question sellers keep hitting: the phone photo is neither square nor on pure white, Amazon MAIN wants both, Shopify collections look jagged when aspect ratios drift, and a centered 1:1 crop quietly deletes a third of a 3:2 product shot.&lt;/p&gt;

&lt;p&gt;I used four large CC0 product photos (camera, running shoe, watch, leather duffel), exported real 2048×2048 JPEGs from the live tool, and checked every file in Python — including corner RGB samples for the Amazon white rule and a product-bounding-box fill % for the 85% rule.&lt;/p&gt;

&lt;p&gt;Short version: &lt;strong&gt;"Square with color" set to &lt;code&gt;#FFFFFF&lt;/code&gt; really does give corner pixels of RGB 255,255,255&lt;/strong&gt;, keeps the whole photo, and lands at Shopify's usual 2048 square. A centered &lt;strong&gt;Square Crop&lt;/strong&gt; on the same 3:2 sources keeps only &lt;strong&gt;66.7%&lt;/strong&gt; of the width. Blur pad looks nicer for social, but its corners are warm brown — it fails Amazon MAIN. And white-pad alone on a full-bleed 3:2 photo only fills about &lt;strong&gt;67%&lt;/strong&gt; of the square frame, so the Amazon 85% fill rule is a separate problem from the white-background rule.&lt;/p&gt;

&lt;h2&gt;
  
  
  Marketplace rules I actually checked (2026-11-05)
&lt;/h2&gt;

&lt;p&gt;I only used each platform's own help pages (or, for Amazon, a Seller Central manager restatement, because the Seller Central help article itself is login-walled).&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Marketplace&lt;/th&gt;
&lt;th&gt;What matters for this job&lt;/th&gt;
&lt;th&gt;Source&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Shopify&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Best types PNG then JPEG; max 5000×5000 or 25 MP; file under 20 MB; &lt;strong&gt;square product images 2048×2048 usually display best&lt;/strong&gt;; keep a &lt;strong&gt;consistent aspect ratio&lt;/strong&gt; on featured/main images so collection grids align. White background is recommended in the photography guide (setup tip), not a hard RGB lock like Amazon.&lt;/td&gt;
&lt;td&gt;
&lt;a href="https://help.shopify.com/en/manual/products/product-media/product-media-types" rel="noopener noreferrer"&gt;Product media types&lt;/a&gt;, &lt;a href="https://help.shopify.com/en/manual/products/product-media/product-photography" rel="noopener noreferrer"&gt;Product photography&lt;/a&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Amazon MAIN&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Pure white background &lt;strong&gt;RGB 255,255,255&lt;/strong&gt;; product fills &lt;strong&gt;85–100%&lt;/strong&gt; of the frame; show only the product (no props/text/logos/watermarks on MAIN); ≥1000 px on the longest side for zoom; JPEG preferred.&lt;/td&gt;
&lt;td&gt;Seller Central community post by Dougal_Amazon restating the standards (&lt;a href="https://sellercentral.amazon.com/seller-forums/discussions/t/0149bdb3-2056-42ce-b0bb-9eef94e3d2b8" rel="noopener noreferrer"&gt;discussion&lt;/a&gt;) — honest note: I could not quote a public G1881 help URL because it redirects behind login&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Etsy&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Listing photos recommended &lt;strong&gt;≥2000×2000&lt;/strong&gt;; first photo &lt;strong&gt;≥635×635&lt;/strong&gt; to avoid lower search placement; types &lt;code&gt;.jpg .gif .png .svg .heic&lt;/code&gt;; animated GIF and transparent PNG not supported (transparency becomes black). &lt;strong&gt;At least one unedited photo of the actual item against a real background is required&lt;/strong&gt; — white-pad is for other slots / other marketplaces, not a replacement for that required real photo.&lt;/td&gt;
&lt;td&gt;
&lt;a href="https://help.etsy.com/hc/en-us/articles/115015663347-Requirements-and-Best-Practices-for-Images-in-Your-Etsy-Shop" rel="noopener noreferrer"&gt;Image requirements&lt;/a&gt;, &lt;a href="https://help.etsy.com/hc/en-us/articles/44297885802647-How-Do-Etsy-s-Listing-Image-Requirements-Apply-to-My-Listings" rel="noopener noreferrer"&gt;Listing image requirements / authenticity&lt;/a&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  The four samples
&lt;/h2&gt;

&lt;p&gt;All CC0 from Wikimedia Commons. Non-white, busy backgrounds on purpose so the white pad is obvious:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Photo&lt;/th&gt;
&lt;th&gt;Size&lt;/th&gt;
&lt;th&gt;Bytes&lt;/th&gt;
&lt;th&gt;Aspect&lt;/th&gt;
&lt;th&gt;Author&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Polaroid Land Camera Pronto flat-lay&lt;/td&gt;
&lt;td&gt;6000×4000&lt;/td&gt;
&lt;td&gt;6,942,650&lt;/td&gt;
&lt;td&gt;3:2 landscape&lt;/td&gt;
&lt;td&gt;WI-Photos&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Hoka running shoe (on foot, patio)&lt;/td&gt;
&lt;td&gt;3024×4032&lt;/td&gt;
&lt;td&gt;1,493,123&lt;/td&gt;
&lt;td&gt;3:4 portrait&lt;/td&gt;
&lt;td&gt;Mimimccraney&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Festina chronograph on wood&lt;/td&gt;
&lt;td&gt;5616×3744&lt;/td&gt;
&lt;td&gt;10,543,643&lt;/td&gt;
&lt;td&gt;3:2 landscape&lt;/td&gt;
&lt;td&gt;
&lt;a href="http://www.Pixel.la" rel="noopener noreferrer"&gt;www.Pixel.la&lt;/a&gt; Free Stock Photos&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Leather duffel on forest floor&lt;/td&gt;
&lt;td&gt;5472×3648&lt;/td&gt;
&lt;td&gt;1,813,326&lt;/td&gt;
&lt;td&gt;3:2 landscape&lt;/td&gt;
&lt;td&gt;Harsh Jadav&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Total source weight: &lt;strong&gt;20,792,742 bytes&lt;/strong&gt;. These are lifestyle product photos, not Amazon-ready MAIN shots (props, outdoor ground, a foot in the shoe). I am measuring the padding math, not claiming the content rules are already satisfied.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: open the tool
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fch3k0k66v4i2hqlv2zy4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fch3k0k66v4i2hqlv2zy4.png" alt="Empty Make Image Square Without Cropping page" width="800" height="273"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The converter runs in the browser. Modes are radio buttons: &lt;strong&gt;Square with blur&lt;/strong&gt;, &lt;strong&gt;Square with color&lt;/strong&gt;, &lt;strong&gt;Square by resizing&lt;/strong&gt;, &lt;strong&gt;Square Crop&lt;/strong&gt;. The color picker is &lt;code&gt;input#background-color&lt;/code&gt; (default &lt;code&gt;#ff0000&lt;/code&gt;). Size menu: 512 / 1080 / 1200 / 3000 / Original / &lt;strong&gt;Custom&lt;/strong&gt; — I set Custom to &lt;strong&gt;2048&lt;/strong&gt; to match Shopify's usual square.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fry9l8475q58vx4505ar4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fry9l8475q58vx4505ar4.png" alt="Size menu open on Custom 2048" width="800" height="461"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: what a centered Square Crop deletes
&lt;/h2&gt;

&lt;p&gt;For any landscape W×H with W&amp;gt;H, a centered 1:1 crop keeps a H×H window and throws away &lt;code&gt;(W−H)/W&lt;/code&gt; of the width. On every 3:2 sample here that is &lt;strong&gt;exactly 33.33% of the width gone, 66.67% kept&lt;/strong&gt;. The portrait Hoka (3024×4032) keeps &lt;strong&gt;75%&lt;/strong&gt; of the height and loses the top and bottom 12.5% each.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbal7d2pzylhc9pp20rqd.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbal7d2pzylhc9pp20rqd.png" alt="Crop-loss overlay on the Polaroid flat-lay — red = deleted by a centered 1:1 crop" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;On the Polaroid file that means the film canister on the left and the modern lens on the right are at risk. On the watch, the strap ends go. On the duffel, the handles or the side pocket.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8447tbwyru3p2h3ne49f.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8447tbwyru3p2h3ne49f.png" alt="Polaroid loaded, Square Crop selected" width="800" height="461"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Exported &lt;strong&gt;Square Crop → Custom 2048 → JPEG&lt;/strong&gt;:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Photo&lt;/th&gt;
&lt;th&gt;Crop 2048 JPEG&lt;/th&gt;
&lt;th&gt;What the crop kept&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Polaroid&lt;/td&gt;
&lt;td&gt;697,723 B&lt;/td&gt;
&lt;td&gt;66.67% of width&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Hoka&lt;/td&gt;
&lt;td&gt;838,339 B&lt;/td&gt;
&lt;td&gt;75.00% of height&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Watch&lt;/td&gt;
&lt;td&gt;688,727 B&lt;/td&gt;
&lt;td&gt;66.67% of width&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Duffel&lt;/td&gt;
&lt;td&gt;615,860 B&lt;/td&gt;
&lt;td&gt;66.67% of width&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Step 3: Square with color &lt;code&gt;#FFFFFF&lt;/code&gt;
&lt;/h2&gt;

&lt;p&gt;Switch to &lt;strong&gt;Square with color&lt;/strong&gt;, set the picker to &lt;code&gt;#ffffff&lt;/code&gt;, keep Custom 2048 JPEG, download.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ft94w4ptcqan23nfw6ti6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ft94w4ptcqan23nfw6ti6.png" alt="Polaroid with Square with color and #FFFFFF selected" width="800" height="461"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Corner pixels sampled at (5,5), (W−6,5), (5,H−6), (W−6,H−6) on every white export:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Photo&lt;/th&gt;
&lt;th&gt;White-pad 2048 JPEG&lt;/th&gt;
&lt;th&gt;Corner RGB (4 corners)&lt;/th&gt;
&lt;th&gt;Non-white bbox fill of frame&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Polaroid&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;545,560 B&lt;/strong&gt; (7.9% of the 6.94 MB source, &lt;strong&gt;−92.1%&lt;/strong&gt;)&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;(255,255,255)&lt;/strong&gt; ×4&lt;/td&gt;
&lt;td&gt;67.19%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Hoka&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;715,871 B&lt;/strong&gt; (47.9% of source, &lt;strong&gt;−52.1%&lt;/strong&gt;)&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;(255,255,255)&lt;/strong&gt; ×4&lt;/td&gt;
&lt;td&gt;75.10%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Watch&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;402,352 B&lt;/strong&gt; (3.8% of the 10.54 MB source, &lt;strong&gt;−96.2%&lt;/strong&gt;)&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;(255,255,255)&lt;/strong&gt; ×4&lt;/td&gt;
&lt;td&gt;66.89%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Duffel&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;409,071 B&lt;/strong&gt; (22.6% of source, &lt;strong&gt;−77.4%&lt;/strong&gt;)&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;(255,255,255)&lt;/strong&gt; ×4&lt;/td&gt;
&lt;td&gt;67.09%&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;All four white exports are &lt;strong&gt;2048×2048&lt;/strong&gt; JPEGs. All four have pure-white corners. The whole source photo is visible inside the square — nothing was cut.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpytxlthhln5e163gysif.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpytxlthhln5e163gysif.png" alt="All four products white-padded to 2048" width="800" height="248"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  The 85% fill catch (measured, not guessed)
&lt;/h3&gt;

&lt;p&gt;Amazon MAIN wants the product to fill &lt;strong&gt;85–100%&lt;/strong&gt; of the frame. After a no-crop pad, the source photo occupies &lt;code&gt;min(W,H)/max(W,H)&lt;/code&gt; of the square. For 3:2 that is &lt;strong&gt;66.7%&lt;/strong&gt;; for the Hoka 3:4 portrait it is &lt;strong&gt;75%&lt;/strong&gt;. None of these four lifestyle shots cleared 85% after white-pad alone.&lt;/p&gt;

&lt;p&gt;So white-pad fixes the &lt;strong&gt;RGB 255&lt;/strong&gt; rule and the &lt;strong&gt;square&lt;/strong&gt; rule, but it does &lt;strong&gt;not&lt;/strong&gt; automatically fix the &lt;strong&gt;85% fill&lt;/strong&gt; rule. If your product is already small inside a loose phone frame, pad first then tighten with a &lt;a href="https://squareimage.dev/squarecrop/" rel="noopener noreferrer"&gt;square crop&lt;/a&gt; (or reshoot tighter). If the product already fills the short edge edge-to-edge, you still only get ~67% of a square from a 3:2 original — you need a nearer-to-square source or a deliberate zoom.&lt;/p&gt;

&lt;p&gt;Portrait crop vs pad on the Hoka shoe:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fe1c996835zi24ui823lu.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fe1c996835zi24ui823lu.png" alt="Hoka Square Crop vs white pad" width="686" height="398"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Crop fills the square but cuts the toe/heel region (25% of height gone). White pad keeps the whole shoe and ankle, with white bars left and right, at 75% fill.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: one brand solid (Shopify / Etsy catalog)
&lt;/h2&gt;

&lt;p&gt;Same Polaroid file, &lt;strong&gt;Square with color&lt;/strong&gt; &lt;code&gt;#f5f0e8&lt;/code&gt; (warm paper), Custom 2048 JPEG: &lt;strong&gt;545,466 B&lt;/strong&gt;. Corner samples after JPEG encode: &lt;strong&gt;(245,241,232)&lt;/strong&gt; — within a couple of levels of &lt;code&gt;#f5f0e8&lt;/code&gt; = (245,240,232) because JPEG is lossy. On the watch I also tried &lt;code&gt;#111111&lt;/code&gt; for a dark catalog card: &lt;strong&gt;399,654 B&lt;/strong&gt;, corners &lt;strong&gt;(17,17,17)&lt;/strong&gt; after JPEG.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbccytqe6dvmlo9psd6ur.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbccytqe6dvmlo9psd6ur.png" alt="Polaroid with brand cream pad #f5f0e8" width="800" height="461"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Useful for Shopify/Etsy lifestyle grids where you want a consistent brand matte instead of pure white. &lt;strong&gt;Do not use a brand color for Amazon MAIN.&lt;/strong&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: why blur pad fails Amazon MAIN
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fqnh83uyev7wml9b6da03.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fqnh83uyev7wml9b6da03.png" alt="Polaroid with Square with blur" width="800" height="461"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Same Polaroid, &lt;strong&gt;Square with blur&lt;/strong&gt;, 2048 JPEG: &lt;strong&gt;593,054 B&lt;/strong&gt;. Corner samples: &lt;strong&gt;(172,125,35), (7,3,0), (39,13,0), (52,19,0)&lt;/strong&gt; — warm fairy-light brown and near-black wood, not RGB 255. Fill of non-white content is essentially 100% of the frame (the blur covers everything), which is the opposite of a white MAIN plate.&lt;/p&gt;

&lt;p&gt;Pretty for Instagram. Automatic reject material for Amazon MAIN.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: crop vs white vs brand vs blur, side by side
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdf04goxcnitm5db78r7j.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdf04goxcnitm5db78r7j.png" alt="Polaroid four-way: Crop, White, Brand cream, Blur" width="798" height="244"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Mode&lt;/th&gt;
&lt;th&gt;Bytes&lt;/th&gt;
&lt;th&gt;Keeps full photo?&lt;/th&gt;
&lt;th&gt;Corner RGB&lt;/th&gt;
&lt;th&gt;Amazon MAIN white?&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Square Crop&lt;/td&gt;
&lt;td&gt;697,723&lt;/td&gt;
&lt;td&gt;No (66.7% of width)&lt;/td&gt;
&lt;td&gt;scene colors&lt;/td&gt;
&lt;td&gt;No (and product is cropped)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Square with color &lt;code&gt;#FFFFFF&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;545,560&lt;/td&gt;
&lt;td&gt;Yes&lt;/td&gt;
&lt;td&gt;255,255,255&lt;/td&gt;
&lt;td&gt;Yes (white rule)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Square with color &lt;code&gt;#f5f0e8&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;545,466&lt;/td&gt;
&lt;td&gt;Yes&lt;/td&gt;
&lt;td&gt;~245,241,232&lt;/td&gt;
&lt;td&gt;No&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Square with blur&lt;/td&gt;
&lt;td&gt;593,054&lt;/td&gt;
&lt;td&gt;Yes&lt;/td&gt;
&lt;td&gt;warm brown / black&lt;/td&gt;
&lt;td&gt;No&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Step 7: simulated Shopify strip and Amazon card
&lt;/h2&gt;

&lt;p&gt;Four white-padded 2048 thumbs in a simulated collection row — same 1:1 ratio, so the grid lines up the way Shopify's media docs describe when featured images share an aspect ratio:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwyw19pjews09j1f5ck95.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwyw19pjews09j1f5ck95.png" alt="Simulated Shopify collection strip of the four white pads" width="800" height="259"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;And a mock Amazon search card built from the real Polaroid white export, with the four corner swatches sampled from the file (all white):&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fee6kdwwakr8u8xg0qymm.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fee6kdwwakr8u8xg0qymm.png" alt="Simulated Amazon card from the white-padded Polaroid" width="360" height="420"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Blur version of the same card — the swatches are brown/black, which is the whole point:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftk16vmquhez3y6u58a8w.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftk16vmquhez3y6u58a8w.png" alt="Simulated Amazon card from the blur pad — corners fail white" width="360" height="420"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Checklist I would actually use
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Amazon MAIN:&lt;/strong&gt; white pad &lt;code&gt;#FFFFFF&lt;/code&gt; + corners really read 255,255,255 + product bbox ≥85% of the square + no props/text/logos on MAIN + ≥1000 px longest side (2048 is safely above that). White-pad a loose 3:2 lifestyle shot alone is not enough for the 85% rule.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Shopify:&lt;/strong&gt; consistent 1:1 (2048×2048 is the size their media docs call out as usually best), PNG or JPEG, under 20 MB / 5000 px. Brand solid pads are fine when you want a catalog color instead of white.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Etsy:&lt;/strong&gt; keep &lt;strong&gt;one unedited real-background photo&lt;/strong&gt; of the actual item for the authenticity rule; use padded squares for other listing slots or for channels that want white/square. First photo ≥635×635; prefer ≥2000×2000.&lt;/li&gt;
&lt;li&gt;Prefer &lt;a href="https://squareimage.dev/make-image-square-without-cropping/" rel="noopener noreferrer"&gt;Square with color&lt;/a&gt; over &lt;a href="https://squareimage.dev/squarecrop/" rel="noopener noreferrer"&gt;Square Crop&lt;/a&gt; when the product touches the short edge and you cannot afford to lose the sides. Use &lt;a href="https://squareimage.dev/blurredframe/" rel="noopener noreferrer"&gt;blurred frame&lt;/a&gt; only when Amazon MAIN is not the destination.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you also need the JPEG/PNG under a shop's upload cap after padding, a related browser compressor is &lt;a href="https://pixsip.com/" rel="noopener noreferrer"&gt;PixSip&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Credits
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Polaroid-Camera.jpg" rel="noopener noreferrer"&gt;Polaroid-Camera.jpg&lt;/a&gt; — WI-Photos, CC0&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Hoka_Running_Shoe.jpg" rel="noopener noreferrer"&gt;Hoka Running Shoe.jpg&lt;/a&gt; — Mimimccraney, CC0&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Fashion-wristwatch-time-watch_(24217032812).jpg" rel="noopener noreferrer"&gt;Fashion-wristwatch-time-watch (24217032812).jpg&lt;/a&gt; — &lt;a href="http://www.Pixel.la" rel="noopener noreferrer"&gt;www.Pixel.la&lt;/a&gt; Free Stock Photos, CC0&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Leather_duffel_bag_on_the_ground.jpg" rel="noopener noreferrer"&gt;Leather duffel bag on the ground.jpg&lt;/a&gt; — Harsh Jadav, CC0&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Related SquareImage writes on other platforms: &lt;a href="https://videocompress-guides.hashnode.dev/instagram-carousel-mixed-shapes-square-1080" rel="noopener noreferrer"&gt;Hashnode — Instagram carousel, mixed shapes to square 1080&lt;/a&gt; · &lt;a href="https://medium.com/@videocompress/keep-the-whole-subject-when-you-force-a-photo-to-square-75d5ad1f72db" rel="noopener noreferrer"&gt;Medium — keep the whole subject when you force a photo to square&lt;/a&gt; · &lt;a href="https://videocompress-guides.blogspot.com/2026/10/a-11-crop-deletes-44-of-this-wide-photo.html" rel="noopener noreferrer"&gt;Blogger — a 1:1 crop deletes 44% of this wide photo&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Also useful on the product: the &lt;a href="https://squareimage.dev/square-image-guide-en.html" rel="noopener noreferrer"&gt;square image size guide&lt;/a&gt; and the homepage &lt;a href="https://squareimage.dev/" rel="noopener noreferrer"&gt;square image converter&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>ecommerce</category>
      <category>images</category>
      <category>tutorial</category>
      <category>shopify</category>
    </item>
    <item>
      <title>A 14 MB Panorama to a Seamless 3-Slide Instagram Carousel (and the 3x3 Grid Trap on the Tall Profile)</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Tue, 03 Nov 2026 02:13:32 +0000</pubDate>
      <link>https://dev.to/videocompress/a-14-mb-panorama-a-seamless-3-slide-instagram-carousel-and-the-3x3-grid-trap-on-the-tall-profile-29po</link>
      <guid>https://dev.to/videocompress/a-14-mb-panorama-a-seamless-3-slide-instagram-carousel-and-the-3x3-grid-trap-on-the-tall-profile-29po</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://dev.to/videocompress/a-14-mb-panorama-a-seamless-3-slide-instagram-carousel-and-the-3x3-grid-trap-on-the-tall-profile-29po"&gt;DEV.to&lt;/a&gt; — tested with &lt;a href="https://squareimage.dev" rel="noopener noreferrer"&gt;SquareImage&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;I build SquareImage, a small set of in-browser square-image tools. This week I put its &lt;a href="https://squareimage.dev/instagram-grid-maker/" rel="noopener noreferrer"&gt;Instagram grid maker&lt;/a&gt; through two jobs people keep asking about: turning a wide panorama into a carousel you can swipe through without a visible break, and splitting one big photo into a 3×3 profile mosaic. I ran both on large public-domain photos from Wikimedia Commons and checked every output file in Python.&lt;/p&gt;

&lt;p&gt;Short version: the panorama trick works, and the three slides come out &lt;strong&gt;90% smaller than the source&lt;/strong&gt;. The 3×3 mosaic also works, but the tall profile grid Instagram rolled out in 2025 will clip the edges of every tile, so where your subject sits matters more than it used to.&lt;/p&gt;

&lt;h2&gt;
  
  
  What Instagram actually says (checked 2026-11-03)
&lt;/h2&gt;

&lt;p&gt;I only used Instagram's own pages for the facts below:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Carousels:&lt;/strong&gt; up to &lt;strong&gt;20&lt;/strong&gt; photos and videos per post. "The orientation you choose (square, portrait or landscape) affects all of the photos or videos in your post. You can't select a different orientation for each." Your profile shows the &lt;strong&gt;first&lt;/strong&gt; item. After sharing you can reorder or remove items, but you can't add new ones. (&lt;a href="https://help.instagram.com/269314186824048" rel="noopener noreferrer"&gt;Help Center: multiple photos&lt;/a&gt;)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Resolution:&lt;/strong&gt; photos are shared at up to &lt;strong&gt;1080 px wide&lt;/strong&gt;. The original resolution is kept for widths of 320–1080 px with an aspect ratio between &lt;strong&gt;1.91:1 and 3:4&lt;/strong&gt;. Other ratios get cropped. (&lt;a href="https://help.instagram.com/1631821640426723" rel="noopener noreferrer"&gt;Help Center: image resolution&lt;/a&gt;)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Profile grid:&lt;/strong&gt; Adam Mosseri announced the "tall grid" in January 2025, saying rectangles do a better job than squares now that most uploads are vertical (&lt;a href="https://www.threads.com/@mosseri/post/DFBmsT0SKO6" rel="noopener noreferrer"&gt;Threads post&lt;/a&gt;). The Help Center doesn't give an exact thumbnail ratio. Third-party guides usually measure it as &lt;strong&gt;3:4&lt;/strong&gt;, so that's the ratio I simulate below. Treat it as a measurement, not an official spec.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So a 1:1 carousel of 1080×1080 slides is fully supported. A square 3×3 mosaic still posts fine, but on the profile each square tile is shown through a taller window.&lt;/p&gt;

&lt;h2&gt;
  
  
  Job 1: a 3:1 panorama → seamless 3-slide carousel
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Sample:&lt;/strong&gt; &lt;em&gt;Coibita Island Panorama&lt;/em&gt; by Nosferattus (CC0), &lt;strong&gt;11,722 × 3,902 px, 13,992,807 bytes&lt;/strong&gt;. The width is almost exactly 3× the height, which is what makes this trick work.&lt;/p&gt;

&lt;p&gt;The grid maker has no dedicated "panorama" button. It splits into 2×2, 3×3 or 4×4 squares. But a 3:1 panorama padded to a square takes up exactly the &lt;strong&gt;middle row&lt;/strong&gt; of a 3×3 grid, so the middle three tiles are your carousel.&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 1: open the tool and drop the photo
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fxgs42h2t7nozf9xk2bkr.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fxgs42h2t7nozf9xk2bkr.png" alt="Empty Instagram Grid Maker page" width="800" height="386"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Everything runs in the browser. My request log showed no POST of the 14 MB file; the only network traffic was the page assets themselves.&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 2: notice what "Cropping" does to a panorama
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9p9pwr06fo34wp9uk1dj.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9p9pwr06fo34wp9uk1dj.png" alt="Panorama loaded with the default Cropping mode — only the middle third survives" width="800" height="579"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The default is &lt;strong&gt;Cropping&lt;/strong&gt;, which fills a square from the centre. For this photo that keeps 3,902 of 11,722 px of width, which is &lt;strong&gt;33.3%&lt;/strong&gt;. Two thirds of the beach is gone.&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 3: switch to Padding, keep 3×3, JPEG, and download
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd0jlqe57aia1kpoeg2iw.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd0jlqe57aia1kpoeg2iw.png" alt="Padding mode: the panorama sits exactly in the middle row" width="800" height="579"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Padding&lt;/strong&gt; keeps the whole photo and adds white above and below. With a 3:1 source, that white fills the top and bottom rows exactly. "Download all tiles" saved nine files:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhflanen9tjf42mxiovt8.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhflanen9tjf42mxiovt8.png" alt="The nine downloaded tiles with sizes" width="680" height="790"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Tiles&lt;/th&gt;
&lt;th&gt;What they are&lt;/th&gt;
&lt;th&gt;Size each&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;grid-01–03, 07–09&lt;/td&gt;
&lt;td&gt;pure white (bin these)&lt;/td&gt;
&lt;td&gt;7,694 B&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;grid-06&lt;/td&gt;
&lt;td&gt;slide 1 (left)&lt;/td&gt;
&lt;td&gt;488,500 B&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;grid-05&lt;/td&gt;
&lt;td&gt;slide 2 (middle)&lt;/td&gt;
&lt;td&gt;313,384 B&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;grid-04&lt;/td&gt;
&lt;td&gt;slide 3 (right)&lt;/td&gt;
&lt;td&gt;556,299 B&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;All are &lt;strong&gt;1080 × 1080 px&lt;/strong&gt; JPEGs. The three real slides total &lt;strong&gt;1,358,183 bytes, 9.7% of the 13,992,807-byte source (−90.3%)&lt;/strong&gt;, and they're already at Instagram's 1080 px width, so nothing gets rescaled on upload.&lt;/p&gt;

&lt;p&gt;One gotcha: the filenames are numbered for &lt;strong&gt;profile-grid posting order&lt;/strong&gt; (newest post shows top-left), so they run right-to-left. For a carousel, pick them &lt;strong&gt;06 → 05 → 04&lt;/strong&gt;.&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 4: check the seams
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkjnbapnxf6b8l5hc7c0l.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkjnbapnxf6b8l5hc7c0l.png" alt="The three slides in carousel order" width="799" height="297"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;I compared the last pixel column of each slide with the first column of the next (mean absolute difference, 0–255 scale):&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;slide 1 → 2: &lt;strong&gt;5.75&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;slide 2 → 3: &lt;strong&gt;4.60&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;two neighbouring columns &lt;em&gt;inside&lt;/em&gt; one slide: &lt;strong&gt;4.50&lt;/strong&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The jump at each cut is about the same as the change between any two neighbouring columns, so there's no visible break when you swipe. Putting them in the wrong order (04 → 05) gives &lt;strong&gt;72.8&lt;/strong&gt;, a hard edge you'd notice. None of the slides has a white band at its top or bottom edge either: 3,902 × 3 is within a few pixels of 11,722, so the padding falls almost entirely in the discarded rows.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F20e0m7j7n65m7fa6p3r4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F20e0m7j7n65m7fa6p3r4.png" alt="Rebuilt strip from the three slides" width="800" height="267"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Rule of thumb:&lt;/strong&gt; crop your panorama to exactly 3:1 before splitting, and use 3×3. With an odd grid, an N:1 panorama lands neatly in the middle row. With 2×2 or 4×4, the padded panorama straddles a row boundary, so the slides come out half white.&lt;/p&gt;

&lt;h2&gt;
  
  
  Job 2: one photo → 3×3 profile mosaic
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Sample:&lt;/strong&gt; &lt;em&gt;Podersdorf Lighthouse 1&lt;/em&gt; by Dimitry Anikin (CC0), &lt;strong&gt;10,266 × 7,700 px, 24,607,534 bytes&lt;/strong&gt;.&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 1: 3×3, Cropping, JPEG
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fcv44lha5j9nf3pr3knyh.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fcv44lha5j9nf3pr3knyh.png" alt="Lighthouse photo split 3×3 with Cropping" width="800" height="579"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Cropping takes the centre square (x = 1,283 → 8,983) and drops 25% of the width. The nine 1080 × 1080 tiles total &lt;strong&gt;805,987 bytes, 3.3% of the source (−96.7%)&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5wzuh0ea5lp3iah8vesk.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5wzuh0ea5lp3iah8vesk.png" alt="Nine tiles reassembled in profile order" width="800" height="800"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 2: the problem, a seam in the subject
&lt;/h3&gt;

&lt;p&gt;The tower's red-and-white body spans x = 3,344–3,836 in the original. The first column cut is at x ≈ 3,850, so the whole tower is squeezed against the right edge of the left column and the lantern gallery pokes across the cut.&lt;/p&gt;

&lt;p&gt;That's fine on the old square grid. Now simulate the tall profile: each square tile is shown through a 3:4 window, which keeps the centre &lt;strong&gt;75%&lt;/strong&gt; of its width and hides 12.5% on each side.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fnnxt7is1zbit6xs4cucb.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fnnxt7is1zbit6xs4cucb.png" alt="Square grid vs simulated tall 3:4 grid" width="800" height="473"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;In the tall view, &lt;strong&gt;only 38% of the tower's width is still visible&lt;/strong&gt;. The rest falls in the hidden strip at the tile's edge.&lt;/p&gt;

&lt;h3&gt;
  
  
  Step 3: the fix, put the subject in the middle column first
&lt;/h3&gt;

&lt;p&gt;The grid maker always crops from the centre and has no drag-to-reposition. So I cropped the photo to its &lt;strong&gt;left&lt;/strong&gt; 7,700 × 7,700 square first (Pillow, quality 95, 9,477,077 B) and ran the same 3×3 split again. The tower now sits at ~47% across, inside the middle column, and &lt;strong&gt;100% of it stays visible&lt;/strong&gt; in the tall view. Nine tiles, 754,701 bytes total.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5d2h3w3rbn0ra3typzbo.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5d2h3w3rbn0ra3typzbo.png" alt="Full photo vs left-square, in square and tall grids" width="800" height="234"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;If you'd rather not touch Python, any square cropper where you choose the frame position will do, then feed the result to the grid maker. The &lt;a href="https://squareimage.dev/instagram-square-guide.html" rel="noopener noreferrer"&gt;Instagram square guide&lt;/a&gt; covers framing for single posts.&lt;/p&gt;

&lt;h2&gt;
  
  
  Checklist
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Panorama carousel:&lt;/strong&gt; crop to 3:1, then choose 3×3 + &lt;strong&gt;Padding&lt;/strong&gt;. Keep tiles &lt;strong&gt;06, 05, 04&lt;/strong&gt; and post them in that order as one 1:1 carousel (Instagram applies one orientation to all slides).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Profile mosaic:&lt;/strong&gt; keep the subject away from the outer 12.5% of every tile column, because the tall grid hides it. Better still, put it fully inside the middle column.&lt;/li&gt;
&lt;li&gt;Use a big source (the tool page suggests at least 2,000 px on the short edge) so every tile actually reaches 1080 px.&lt;/li&gt;
&lt;li&gt;Post grid tiles in the numbered order (01 first) so the image assembles correctly on your profile.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you only need one photo to fit a square without losing the edges, that's what the &lt;a href="https://squareimage.dev/make-image-square-without-cropping/" rel="noopener noreferrer"&gt;no-crop square maker&lt;/a&gt; and the &lt;a href="https://squareimage.dev/" rel="noopener noreferrer"&gt;SquareImage homepage&lt;/a&gt; are for. If the tiles feel heavy for a website gallery, I also run &lt;a href="https://pixsip.com/" rel="noopener noreferrer"&gt;PixSip&lt;/a&gt;, a browser-side image compressor.&lt;/p&gt;

&lt;h2&gt;
  
  
  Things I noticed in my own tool
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;No 1×N "panorama" mode: you need the 3×3 + Padding detour above.&lt;/li&gt;
&lt;li&gt;Tiles are square only, and the page doesn't mention the tall profile grid yet.&lt;/li&gt;
&lt;li&gt;Padding is always white, and there's no reposition control for Cropping.&lt;/li&gt;
&lt;li&gt;The download is nine separate files (your browser may ask once to allow multiple downloads), named for grid order rather than carousel order.&lt;/li&gt;
&lt;/ul&gt;




&lt;p&gt;&lt;strong&gt;Photo credits:&lt;/strong&gt; &lt;a href="https://commons.wikimedia.org/wiki/File:Coibita_Island_Panorama.jpg" rel="noopener noreferrer"&gt;Coibita Island Panorama&lt;/a&gt; by Nosferattus and &lt;a href="https://commons.wikimedia.org/wiki/File:Podersdorf_Lighthouse_1.jpg" rel="noopener noreferrer"&gt;Podersdorf Lighthouse 1&lt;/a&gt; by Dimitry Anikin, both via Wikimedia Commons under &lt;a href="https://creativecommons.org/publicdomain/zero/1.0/" rel="noopener noreferrer"&gt;CC0 1.0&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Related SquareImage posts:&lt;/strong&gt; &lt;a href="https://videocompress-guides.hashnode.dev/instagram-carousel-mixed-shapes-square-1080" rel="noopener noreferrer"&gt;Mixed-shape carousel to 1080 squares (Hashnode)&lt;/a&gt; · &lt;a href="https://medium.com/@videocompress/keep-the-whole-subject-when-you-force-a-photo-to-square-75d5ad1f72db" rel="noopener noreferrer"&gt;Keep the whole subject when forcing a photo square (Medium)&lt;/a&gt; · &lt;a href="https://dev.to/videocompress/make-an-instagram-ready-square-without-cropping-faces-1b1l"&gt;Instagram-ready square without cropping faces (DEV)&lt;/a&gt;&lt;/p&gt;

</description>
      <category>instagram</category>
      <category>images</category>
      <category>design</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>Lighthouse Said My Hero Images Were Too Heavy. I Compressed the Phone JPGs in the Browser and Re-Measured</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Sat, 24 Oct 2026 02:12:33 +0000</pubDate>
      <link>https://dev.to/videocompress/lighthouse-said-my-hero-images-were-too-heavy-i-compressed-the-phone-jpgs-in-the-browser-and-40lc</link>
      <guid>https://dev.to/videocompress/lighthouse-said-my-hero-images-were-too-heavy-i-compressed-the-phone-jpgs-in-the-browser-and-40lc</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://dev.to/videocompress/lighthouse-said-my-hero-images-were-too-heavy-i-compressed-the-phone-jpgs-in-the-browser-and-40lc"&gt;DEV.to&lt;/a&gt; by the &lt;a href="https://pixsip.com" rel="noopener noreferrer"&gt;PixSip&lt;/a&gt; team.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Five real phone photos (35.0 MB) → 7.9 MB on &lt;a href="https://pixsip.com/compress-jpg" rel="noopener noreferrer"&gt;pixsip.com/compress-jpg&lt;/a&gt;. DevTools Network transfer size and load timing on a local test page, before and after. No upload, no target-KB slider.&lt;/p&gt;

&lt;p&gt;I build PixSip, so I'll say that up front. The numbers below are from a fresh headless run on the live page, not a marketing slide.&lt;/p&gt;

&lt;h2&gt;
  
  
  The inventory (Network → Img first)
&lt;/h2&gt;

&lt;p&gt;Skip the compressor for a minute. Open the page that feels slow, DevTools → &lt;strong&gt;Network&lt;/strong&gt;, filter &lt;strong&gt;Img&lt;/strong&gt;, disable cache, hard-reload.&lt;/p&gt;

&lt;p&gt;On a local test gallery that dropped five full-size phone JPGs into one hero strip, the panel looked like this:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;File&lt;/th&gt;
&lt;th&gt;Camera&lt;/th&gt;
&lt;th&gt;Pixels&lt;/th&gt;
&lt;th&gt;Bytes&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;metro-invalidovna.jpg&lt;/td&gt;
&lt;td&gt;Pixel 8 Pro&lt;/td&gt;
&lt;td&gt;8160×6144&lt;/td&gt;
&lt;td&gt;8.6 MB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ayup-market.jpg&lt;/td&gt;
&lt;td&gt;Samsung SM-A336B&lt;/td&gt;
&lt;td&gt;4000×3000&lt;/td&gt;
&lt;td&gt;7.5 MB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;sturgeons-mill.jpg&lt;/td&gt;
&lt;td&gt;iPhone 15 Pro Max&lt;/td&gt;
&lt;td&gt;4032×3024&lt;/td&gt;
&lt;td&gt;6.5 MB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;mudbank-adur.jpg&lt;/td&gt;
&lt;td&gt;Pixel 8 Pro&lt;/td&gt;
&lt;td&gt;8160×6144&lt;/td&gt;
&lt;td&gt;6.9 MB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ad-column.jpg&lt;/td&gt;
&lt;td&gt;Pixel 4a&lt;/td&gt;
&lt;td&gt;4032×3024&lt;/td&gt;
&lt;td&gt;5.4 MB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Total&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;/td&gt;
&lt;td&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;35.0 MB&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8yiofz26ere57yqrw0f0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8yiofz26ere57yqrw0f0.png" alt="DevTools-style inventory of five heavy phone JPGs totaling 35 MB" width="800" height="281"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Lighthouse's "Improve image delivery" opportunity is pointing at the same files. You don't need every JPG on the site — only the ones that dominate transfer size. Compress those, leave the rest alone.&lt;/p&gt;

&lt;p&gt;Same five files on a local page, emulated ~10 Mbps, cache off:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8ahqbw8to5dnt6fjzxmm.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8ahqbw8to5dnt6fjzxmm.png" alt="Local test page before compression: 35.0 MB transfer, 28.1 s load" width="800" height="451"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Img requests: &lt;strong&gt;5&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;Transfer size: &lt;strong&gt;35.0 MB&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;Largest: &lt;strong&gt;8.6 MB&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;Load event: &lt;strong&gt;28.1 s&lt;/strong&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That's the baseline. Everything after this is "make those five smaller without leaving the browser."&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Open /compress-jpg and wait for the lock
&lt;/h2&gt;

&lt;p&gt;Go to &lt;a href="https://pixsip.com/compress-jpg" rel="noopener noreferrer"&gt;pixsip.com/compress-jpg&lt;/a&gt;. The drop zone starts disabled while the worker (MozJPEG wasm) loads — usually about a second. Wait until the small &lt;strong&gt;🔒&lt;/strong&gt; appears under "Drop images here" and the file input is enabled. Dropping too early hangs the job.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fj2fk6ewlxbhk9emuu2jg.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fj2fk6ewlxbhk9emuu2jg.png" alt="PixSip compress-jpg ready state with lock icon under the drop zone" width="800" height="500"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Free plan: up to &lt;strong&gt;20 images × 10 MB&lt;/strong&gt; each, up to 3 output formats at once. There is &lt;strong&gt;no target-KB setting and no quality slider&lt;/strong&gt; — MozJPEG runs at one calibrated quality. Don't expect "compress to exactly 200 KB."&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Drop the batch, read the results table
&lt;/h2&gt;

&lt;p&gt;I dropped the five originals as JPG → JPG. Worker ready in ~1.0 s; the batch finished in ~11 s. Nothing over a few KB left the machine (clarity.ms analytics pings only — no image upload).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5b0dvzbmbxpzzae5n90e.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5b0dvzbmbxpzzae5n90e.png" alt="Results table: 5/5 done, Saved 27.1 MB total · 77% smaller" width="800" height="423"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;File&lt;/th&gt;
&lt;th&gt;Before&lt;/th&gt;
&lt;th&gt;After&lt;/th&gt;
&lt;th&gt;Δ&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;metro-invalidovna.jpg&lt;/td&gt;
&lt;td&gt;8.6 MB&lt;/td&gt;
&lt;td&gt;2.0 MB&lt;/td&gt;
&lt;td&gt;−76%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ayup-market.jpg&lt;/td&gt;
&lt;td&gt;7.5 MB&lt;/td&gt;
&lt;td&gt;963 KB&lt;/td&gt;
&lt;td&gt;−87%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;sturgeons-mill.jpg&lt;/td&gt;
&lt;td&gt;6.5 MB&lt;/td&gt;
&lt;td&gt;1.7 MB&lt;/td&gt;
&lt;td&gt;−74%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;mudbank-adur.jpg&lt;/td&gt;
&lt;td&gt;6.9 MB&lt;/td&gt;
&lt;td&gt;1.6 MB&lt;/td&gt;
&lt;td&gt;−76%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ad-column.jpg&lt;/td&gt;
&lt;td&gt;5.4 MB&lt;/td&gt;
&lt;td&gt;1.5 MB&lt;/td&gt;
&lt;td&gt;−71%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Total&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;35.0 MB&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;7.9 MB&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;−77%&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Card text on the page: &lt;strong&gt;"Saved 27.1 MB total · 77% smaller."&lt;/strong&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Verify pixels, EXIF, GPS
&lt;/h2&gt;

&lt;p&gt;Download the ZIP and check what actually changed. Dimensions must stay the same — PixSip does not resize. EXIF and GPS should be gone (four of the five originals carried GPS; none of the outputs do).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6d6yz8f6myixnyf235no.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6d6yz8f6myixnyf235no.png" alt="Terminal-style check: dimensions unchanged, EXIF/GPS stripped, 35.0 → 7.9 MB" width="799" height="388"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Summary from the real ZIP:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Pixel sizes changed: &lt;strong&gt;0 of 5&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;GPS left: &lt;strong&gt;0 of 5&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;EXIF tags left: &lt;strong&gt;none&lt;/strong&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That last point matters for WordPress: if you upload the phone original, WordPress keeps a publicly reachable copy of it (and often its GPS) even after it builds &lt;code&gt;-scaled&lt;/code&gt; thumbnails. Strip first, then upload.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Compare at 1:1, then your own 100% crop
&lt;/h2&gt;

&lt;p&gt;Hit &lt;strong&gt;Compare&lt;/strong&gt; on any row, then &lt;strong&gt;Zoom 1:1&lt;/strong&gt;. Drag the divider across fine detail — tiles, foliage, sign lettering — not a downscaled preview.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fi9icjzvqnfd3xlv0bro3.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fi9icjzvqnfd3xlv0bro3.png" alt="PixSip Compare view with Zoom 1:1 on the mill photo" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;And a 100% crop from the real files (metro platform tiles + tactile paving), original on the left, PixSip on the right:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Faml2oq72sr1vnbwfnvw3.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Faml2oq72sr1vnbwfnvw3.png" alt="100% crop: metro tiles original 8.6 MB vs PixSip 2.0 MB" width="799" height="297"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;At this distance the difference is hard to spot. If a crop looks soft, that photo may need a lighter pass elsewhere — or simply a smaller display size in CSS, not a quality slider PixSip doesn't have.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Swap the files and re-weigh
&lt;/h2&gt;

&lt;p&gt;Replace the five &lt;code&gt;&amp;lt;img&amp;gt;&lt;/code&gt; sources with the compressed JPGs. Same local page, same emulated ~10 Mbps, cache still off:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fs0uuvbmzqi7wnrvg1ioz.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fs0uuvbmzqi7wnrvg1ioz.png" alt="Local test page after compression: 7.9 MB transfer, 6.4 s load" width="800" height="451"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Before&lt;/th&gt;
&lt;th&gt;After&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Transfer size&lt;/td&gt;
&lt;td&gt;35.0 MB&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;7.9 MB&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Largest image&lt;/td&gt;
&lt;td&gt;8.6 MB&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;2.0 MB&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Load event (~10 Mbps)&lt;/td&gt;
&lt;td&gt;28.1 s&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;6.4 s&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;That's the Lighthouse/Network loop closed: inventory → compress the offenders → re-measure. LCP on a real WordPress theme will still depend on font/CSS/JS, but the image bytes are no longer the obvious bottleneck.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short checklist
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Network → Img (or Lighthouse → Improve image delivery) — list the heaviest JPGs only.&lt;/li&gt;
&lt;li&gt;Open &lt;a href="https://pixsip.com/compress-jpg" rel="noopener noreferrer"&gt;/compress-jpg&lt;/a&gt;, wait for the &lt;strong&gt;🔒&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Drop the batch. Read the results table; download the ZIP.&lt;/li&gt;
&lt;li&gt;Confirm dimensions unchanged and EXIF/GPS gone before you upload to WordPress.&lt;/li&gt;
&lt;li&gt;Compare + Zoom 1:1 (and a 100% crop of your own) on at least one hero image.&lt;/li&gt;
&lt;li&gt;Swap files, re-weigh transfer size. Repeat for the next heavy set if needed.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Free plan caps at 20 × 10 MB; bigger batches are on &lt;a href="https://pixsip.com/pricing" rel="noopener noreferrer"&gt;pricing&lt;/a&gt;. For how MozJPEG lands vs other compressors, see the &lt;a href="https://pixsip.com/blog/pixsip-benchmark" rel="noopener noreferrer"&gt;PixSip benchmark&lt;/a&gt;. Nothing is uploaded — details in &lt;a href="https://pixsip.com/blog/privacy-by-default" rel="noopener noreferrer"&gt;privacy by default&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;If the heavy asset is a &lt;strong&gt;video&lt;/strong&gt; hero instead of a photo strip, I also build &lt;a href="https://vidsmaller.com/" rel="noopener noreferrer"&gt;vidsmaller.com&lt;/a&gt; for target-size cloud compression.&lt;/p&gt;

&lt;h2&gt;
  
  
  Photo credits (CC0)
&lt;/h2&gt;

&lt;p&gt;All five samples are CC0 on Wikimedia Commons:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Invalidovna_Metro_Station.jpg" rel="noopener noreferrer"&gt;Invalidovna Metro Station&lt;/a&gt; — AVDLCZ, Pixel 8 Pro&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Ay-Up_Market,_Nottingham_-_29_March_2025.jpg" rel="noopener noreferrer"&gt;Ay-Up Market, Nottingham - 29 March 2025&lt;/a&gt; — Aethonatic, Samsung SM-A336B&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Sturgeon%27s_Mill_-_August_2025_-_Sarah_Stierch_31.jpg" rel="noopener noreferrer"&gt;Sturgeon's Mill - August 2025 - Sarah Stierch 31&lt;/a&gt; — Missvain, iPhone 15 Pro Max&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Mudbank_in_the_River_Adur_2026-05-05.jpg" rel="noopener noreferrer"&gt;Mudbank in the River Adur 2026-05-05&lt;/a&gt; — Andy Li, Pixel 8 Pro&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://commons.wikimedia.org/wiki/File:Advertising_column_with_torn_up_posters_2.jpg" rel="noopener noreferrer"&gt;Advertising column with torn up posters 2&lt;/a&gt; — Profpcde, Pixel 4a&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Related
&lt;/h2&gt;

&lt;p&gt;Related: &lt;a href="https://medium.com/@videocompress/your-wordpress-images-are-too-heavy-compress-the-jpgs-in-your-browser-before-you-upload-2da9127005a6" rel="noopener noreferrer"&gt;Medium — Your WordPress images are too heavy&lt;/a&gt; · &lt;a href="https://videocompress-guides.blogspot.com/2026/10/before-you-hit-upload-in-wordpress.html" rel="noopener noreferrer"&gt;Blogger — Before You Hit Upload in WordPress&lt;/a&gt; · &lt;a href="https://dev.to/videocompress/convert-images-to-webp-or-avif-for-faster-pages-and-check-which-one-is-actually-smaller-5aed"&gt;DEV.to — Convert images to WebP or AVIF&lt;/a&gt;&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>performance</category>
      <category>wordpress</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>Your .webm or .mov keeps getting rejected — convert it to a smaller MP4 in the browser</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Wed, 14 Oct 2026 02:35:56 +0000</pubDate>
      <link>https://dev.to/videocompress/your-webm-or-mov-keeps-getting-rejected-convert-it-to-a-smaller-mp4-in-the-browser-3hn7</link>
      <guid>https://dev.to/videocompress/your-webm-or-mov-keeps-getting-rejected-convert-it-to-a-smaller-mp4-in-the-browser-3hn7</guid>
      <description>&lt;p&gt;A screen recording lands as &lt;code&gt;.webm&lt;/code&gt;. An iPhone clip lands as &lt;code&gt;.mov&lt;/code&gt;. Both play fine on &lt;em&gt;your&lt;/em&gt; machine — then Slack, an email form, a Windows laptop, or a CMS says the file type isn't supported, or the upload is simply too big.&lt;/p&gt;

&lt;p&gt;The fix that usually works is boring and universal: turn the file into an &lt;strong&gt;MP4 with H.264 video and AAC audio&lt;/strong&gt;, and shrink it while you're at it. That is what most uploaders and players actually expect.&lt;/p&gt;

&lt;p&gt;I ran two real clips through &lt;a href="https://videocompress.dev/" rel="noopener noreferrer"&gt;videocompress.dev&lt;/a&gt; in Chrome — nothing uploaded off the machine — and measured the downloads with &lt;code&gt;ffprobe&lt;/code&gt; afterwards.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Source&lt;/th&gt;
&lt;th&gt;Container / codecs&lt;/th&gt;
&lt;th&gt;Size (tool)&lt;/th&gt;
&lt;th&gt;Target&lt;/th&gt;
&lt;th&gt;Output&lt;/th&gt;
&lt;th&gt;Reduction&lt;/th&gt;
&lt;th&gt;Encode time&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Tears of Steel clip → WebM&lt;/td&gt;
&lt;td&gt;WebM · VP8 + Vorbis · 1920×800 · 1:15&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;91.4 MB&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;15 MB&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;13.9 MB&lt;/strong&gt; MP4 H.264+AAC&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;85% smaller&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;~4 min 19 s&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Same clip → MOV&lt;/td&gt;
&lt;td&gt;MOV · H.264 + AAC · 1920×800 · 1:15&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;113.1 MB&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;20 MB&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;15.0 MB&lt;/strong&gt; MP4 H.264+AAC&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;87% smaller&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;~13 s&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Both outputs stayed &lt;strong&gt;1920×800&lt;/strong&gt;. The footage is a 75-second cut of &lt;em&gt;Tears of Steel&lt;/em&gt; (Blender Foundation open movie), re-encoded with ffmpeg into a VP8/Vorbis &lt;code&gt;.webm&lt;/code&gt; and an iPhone-style H.264/AAC &lt;code&gt;.mov&lt;/code&gt; so the demo matches the formats people actually send around.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why .webm and .mov trip up uploads
&lt;/h2&gt;

&lt;p&gt;They fail for different reasons:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;WebM&lt;/strong&gt; (often VP8/VP9 + Opus/Vorbis) is great in Chrome and for browser screen recordings, but plenty of desktop apps, email gateways, and older players still want MP4.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;MOV&lt;/strong&gt; is Apple's QuickTime container. iPhones write H.264 or HEVC into it. Many Windows tools and web forms accept the &lt;em&gt;codecs&lt;/em&gt; but reject the &lt;code&gt;.mov&lt;/code&gt; extension — or they accept MOV but choke on the size.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you only need a container swap and the video is already H.264+AAC, the dedicated &lt;a href="https://videocompress.dev/mov-to-mp4" rel="noopener noreferrer"&gt;MOV to MP4&lt;/a&gt; page can rewrap without re-encoding. When the file is also too large, or the codec isn't MP4-friendly (VP8/VP9 WebM, ProRes, PCM audio), you want the compressor on the homepage: it re-encodes to H.264+AAC and can aim at a megabyte budget. Background on that path: &lt;a href="https://videocompress.dev/video-compressor-no-upload" rel="noopener noreferrer"&gt;video compressor with no upload&lt;/a&gt; and &lt;a href="https://videocompress.dev/webcodecs-vs-ffmpeg-wasm" rel="noopener noreferrer"&gt;WebCodecs vs ffmpeg.wasm&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Open the compressor and drop the file
&lt;/h2&gt;

&lt;p&gt;Go to &lt;a href="https://videocompress.dev/" rel="noopener noreferrer"&gt;videocompress.dev&lt;/a&gt;. The drop zone accepts &lt;strong&gt;MP4, MOV, WebM, MKV, AVI, WMV, FLV and more&lt;/strong&gt;. No account, no watermark.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffgvhflafy8qe0y1zy6rw.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffgvhflafy8qe0y1zy6rw.png" alt="Empty drop zone on videocompress.dev" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Drop the &lt;code&gt;.webm&lt;/code&gt; or &lt;code&gt;.mov&lt;/code&gt; (or click &lt;strong&gt;Choose file&lt;/strong&gt;). When it is ready you should see the filename, size, resolution, and duration, plus &lt;strong&gt;Quality&lt;/strong&gt; / &lt;strong&gt;Target size&lt;/strong&gt; controls and a &lt;strong&gt;Compress video&lt;/strong&gt; button. The file stays in the tab — the lock line under the controls is literal.&lt;/p&gt;

&lt;p&gt;Here is the WebM after load — &lt;strong&gt;91.4 MB · 1920×800 · 1:15&lt;/strong&gt;:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fgjfhh4kefao9morfi1y3.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fgjfhh4kefao9morfi1y3.png" alt="WebM loaded in the local workspace" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;And the MOV — &lt;strong&gt;113.1 MB · 1920×800 · 1:15&lt;/strong&gt;:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fat088khf1pvitu9s6mp4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fat088khf1pvitu9s6mp4.png" alt="MOV loaded in the local workspace" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Switch to Target size (skip the quality presets for a hard cap)
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Quality&lt;/strong&gt; mode (High / Balanced / Small file) is fine when you only care about “smaller.” When Slack, Discord, email, or a form has a hard megabyte limit, click &lt;strong&gt;Target size&lt;/strong&gt; instead and type the budget in MB.&lt;/p&gt;

&lt;p&gt;For the WebM I set &lt;strong&gt;15&lt;/strong&gt;. The estimate under the button read &lt;strong&gt;≈ 14.1 MB&lt;/strong&gt; before I pressed Compress:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9h2tna44u0a8b6jpexw5.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9h2tna44u0a8b6jpexw5.png" alt="WebM with Target size set to 15 MB" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;For the MOV I set &lt;strong&gt;20&lt;/strong&gt; (estimate &lt;strong&gt;≈ 18.8 MB&lt;/strong&gt;):&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwqzypaxyveczyktfzr7l.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwqzypaxyveczyktfzr7l.png" alt="MOV with Target size set to 20 MB" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Leave &lt;strong&gt;Advanced settings → Resolution&lt;/strong&gt; on &lt;strong&gt;Original&lt;/strong&gt; unless the encode is starving and you truly need more headroom — dropping resolution on a clip that already fits the target just throws pixels away. More on picking a number: &lt;a href="https://videocompress.dev/compress-video-to-target-size" rel="noopener noreferrer"&gt;compress video to a target size&lt;/a&gt;. Screen recordings with UI text are the hard case; see &lt;a href="https://videocompress.dev/compress-screen-recording" rel="noopener noreferrer"&gt;compress a screen recording&lt;/a&gt; if that is your source.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Compress the WebM → MP4
&lt;/h2&gt;

&lt;p&gt;Press &lt;strong&gt;Compress video&lt;/strong&gt;. Keep the tab in the foreground; browsers may pause the encode if you switch away.&lt;/p&gt;

&lt;p&gt;On this machine the VP8 WebM took about &lt;strong&gt;4 minutes 19 seconds&lt;/strong&gt;. The result card:&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;91.4 MB → 13.9 MB · 85% smaller&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8gd9lht4lp4htrdfk74k.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8gd9lht4lp4htrdfk74k.png" alt="WebM compression finished — 13.9 MB MP4 ready" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;code&gt;ffprobe&lt;/code&gt; on the download:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Container: MP4
&lt;/li&gt;
&lt;li&gt;Video: &lt;strong&gt;H.264&lt;/strong&gt; 1920×800 ~1.4 Mbps
&lt;/li&gt;
&lt;li&gt;Audio: &lt;strong&gt;AAC&lt;/strong&gt; stereo 44.1 kHz
&lt;/li&gt;
&lt;li&gt;Duration: still ~75 s
&lt;/li&gt;
&lt;li&gt;Bytes: 14,552,715 (13.9 MB as the tool reports)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So the awkward WebM is now a normal MP4 under a 15 MB cap.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Compress the MOV the same way
&lt;/h2&gt;

&lt;p&gt;Same page, &lt;strong&gt;Change video&lt;/strong&gt;, drop the &lt;code&gt;.mov&lt;/code&gt;, &lt;strong&gt;Target size → 20&lt;/strong&gt;, &lt;strong&gt;Compress video&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;This one finished in about &lt;strong&gt;13 seconds&lt;/strong&gt; (H.264 source + WebCodecs path is much faster than decoding VP8 in the tab):&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;113.1 MB → 15.0 MB · 87% smaller&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkanvhnlm52asyezezbjx.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkanvhnlm52asyezezbjx.png" alt="MOV compression finished — 15.0 MB MP4 ready" width="799" height="562"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;code&gt;ffprobe&lt;/code&gt; again: &lt;strong&gt;H.264 + AAC&lt;/strong&gt;, 1920×800, ~75 s, 15,713,740 bytes. Same codecs the tool always aims for — which is why &lt;a href="https://videocompress.dev/compress-mov" rel="noopener noreferrer"&gt;compress MOV files&lt;/a&gt; and a plain “make it MP4 and smaller” goal are the same button once you need a size cap.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Download and spot-check
&lt;/h2&gt;

&lt;p&gt;Click &lt;strong&gt;Download compressed video&lt;/strong&gt;. Quick checks before you re-upload anywhere:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Extension is &lt;code&gt;.mp4&lt;/code&gt;.
&lt;/li&gt;
&lt;li&gt;Size is under your destination limit (for Discord free uploads that means ≤ 20 MB; server boosts raise it — see &lt;a href="https://videocompress.dev/compress-mp4-for-discord" rel="noopener noreferrer"&gt;compress MP4 for Discord&lt;/a&gt;).
&lt;/li&gt;
&lt;li&gt;It plays on a second device or in a private window.
&lt;/li&gt;
&lt;li&gt;If it is still over the limit, lower the target a few MB and run again — or drop resolution one step in Advanced only if the picture is already soft.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Bitrate vs resolution tradeoffs are covered in the &lt;a href="https://videocompress.dev/video-compression-guide" rel="noopener noreferrer"&gt;video compression guide&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;If the file is so large the tab runs out of memory, use &lt;a href="https://vidsmaller.com/" rel="noopener noreferrer"&gt;vidsmaller.com&lt;/a&gt; (cloud exact-size compressor from the same maker) instead of forcing the browser job.&lt;/p&gt;

&lt;h2&gt;
  
  
  Checklist
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;[ ] Source is WebM, MOV, or another browser-decodable file
&lt;/li&gt;
&lt;li&gt;[ ] Opened &lt;a href="https://videocompress.dev/" rel="noopener noreferrer"&gt;videocompress.dev&lt;/a&gt; and saw the file listed locally
&lt;/li&gt;
&lt;li&gt;[ ] Chose &lt;strong&gt;Target size&lt;/strong&gt; with a real destination budget
&lt;/li&gt;
&lt;li&gt;[ ] Downloaded an MP4 that &lt;code&gt;ffprobe&lt;/code&gt; (or any media info tool) shows as H.264 + AAC
&lt;/li&gt;
&lt;li&gt;[ ] Confirmed the byte size against the place you are sending it
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Two formats in, one boring MP4 out — and in these runs, 85–87% smaller without the file ever leaving the browser.&lt;/p&gt;

&lt;p&gt;Related: &lt;a href="https://medium.com/@videocompress/discord-rejected-your-video-shrink-it-to-10mb-in-the-browser-f57904c4a3e9" rel="noopener noreferrer"&gt;Discord rejected your video? Shrink it under 20MB in the browser (Medium)&lt;/a&gt; · &lt;a href="https://medium.com/@videocompress/the-25mb-email-limit-is-smaller-than-it-looks-fitting-a-video-under-it-without-uploading-it-b301e2a0734d" rel="noopener noreferrer"&gt;The 25MB email limit is smaller than it looks (Medium)&lt;/a&gt; · &lt;a href="https://videocompress-guides.hashnode.dev/your-bug-repro-is-89-mb-and-github-takes-10-shrinking-screen-recordings-without-uploading-them" rel="noopener noreferrer"&gt;Your bug repro is 89 MB and GitHub takes 10 (Hashnode)&lt;/a&gt;&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>video</category>
      <category>tutorial</category>
      <category>beginners</category>
    </item>
    <item>
      <title>Fit a tall photo into a WhatsApp or LinkedIn circle without losing your head</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Thu, 08 Oct 2026 02:09:07 +0000</pubDate>
      <link>https://dev.to/videocompress/fit-a-tall-photo-into-a-whatsapp-or-linkedin-circle-without-losing-your-head-557h</link>
      <guid>https://dev.to/videocompress/fit-a-tall-photo-into-a-whatsapp-or-linkedin-circle-without-losing-your-head-557h</guid>
      <description>&lt;p&gt;Most good photos of a person are tall. Phones shoot 3:4, photographers like 2:3, and a half-body shot with crossed arms almost never comes out square. Then you set it as your WhatsApp DP or LinkedIn photo, and the app hands you a fixed circle to drag around. Either the top of your hair goes or your arms do.&lt;/p&gt;

&lt;p&gt;I wanted to see exactly what gets lost and whether padding the photo first is worth it, so I ran one real portrait through the whole thing and checked the result in a circle before uploading anywhere.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the app actually does with your photo
&lt;/h2&gt;

&lt;p&gt;Profile pictures are stored as squares and displayed as circles. SquareImage's &lt;a href="https://squareimage.dev/full-dp/" rel="noopener noreferrer"&gt;Full DP Maker page&lt;/a&gt; lists the stored sizes it designs for: 640 × 640 on WhatsApp, 512 × 512 on Telegram, 320 × 320 on Instagram and Facebook. LinkedIn shows your photo as a circle as well. Whatever you send gets downscaled, so a 6,000-pixel upload buys you nothing except a slower upload.&lt;/p&gt;

&lt;p&gt;That gives you two separate losses:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;The square crop.&lt;/strong&gt; A 2:3 photo forced into 1:1 loses a third of its height. On my test photo (3760 × 5640) that's 1,880 pixels of hair, forehead or arms.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The circle mask.&lt;/strong&gt; A circle inside a square covers π/4 of it, about 78.5%. The four corners are always gone, on every platform that shows round avatars.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;You can't avoid the second one, but you can avoid the first: make the photo square yourself, without cropping, and then keep the subject away from the corners.&lt;/p&gt;

&lt;h2&gt;
  
  
  Squaring the photo with the Full DP Maker
&lt;/h2&gt;

&lt;p&gt;The test file was a CC0 headshot from Wikimedia Commons: 3760 × 5640, 11.6 MB, a man in a blue shirt with crossed arms against a dark backdrop. That's a typical "photographer sent me the full-res file" situation.&lt;/p&gt;

&lt;h3&gt;
  
  
  1. Add the photo
&lt;/h3&gt;

&lt;p&gt;Open the &lt;a href="https://squareimage.dev/full-dp/" rel="noopener noreferrer"&gt;Full DP Maker&lt;/a&gt; and drop the photo in, pick it with the button, or paste it from the clipboard. The square is drawn on a canvas in your browser, so the 11.6 MB original never leaves your machine.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0yiiuw0vgpueqlknri1y.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0yiiuw0vgpueqlknri1y.png" alt="Full DP Maker drop zone on squareimage.dev, before adding a photo" width="800" height="212"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  2. Choose what fills the empty sides
&lt;/h3&gt;

&lt;p&gt;You get three options: &lt;strong&gt;Blur&lt;/strong&gt; fills the sides with a blurred copy of the same photo, &lt;strong&gt;Color&lt;/strong&gt; fills them with one flat color, and &lt;strong&gt;Crop&lt;/strong&gt; is there if you decide you want to fill the frame after all.&lt;/p&gt;

&lt;p&gt;For a person against a busy or textured background, Blur usually looks most natural because the colors already match. If the backdrop is plain, or the photo is a logo, Color looks cleaner. Here's the photo with Blur at the default photo size of 100%:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbwvduuhelswzzwfm7ox5.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbwvduuhelswzzwfm7ox5.png" alt="Portrait squared with the Blur fill at 100% photo size, showing 3760 × 5640 px source" width="800" height="485"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;At 100% the photo touches the top and bottom of the square. That's fine for a square post, but for a circle the head is right where the mask starts curving in.&lt;/p&gt;

&lt;h3&gt;
  
  
  3. Pull the subject in with Photo size
&lt;/h3&gt;

&lt;p&gt;This is the setting that makes the difference for avatars. I moved &lt;strong&gt;Photo size&lt;/strong&gt; from 100% down to 80%. The whole photo shrinks toward the center and the blurred fill grows around it, so the head and elbows sit inside the circle instead of in the corners.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3abl7nux6jqtrfmpvt89.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3abl7nux6jqtrfmpvt89.png" alt="Photo size slider set to 80%, subject pulled away from the edges with a blurred surround" width="799" height="449"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Go lower if the photo has important stuff near its edges. The tool's own advice is to keep text in the middle 70% of the square.&lt;/p&gt;

&lt;h3&gt;
  
  
  4. Download, then check it in a circle before uploading
&lt;/h3&gt;

&lt;p&gt;Leave the download size at 1024 × 1024 and the format at JPEG, and click &lt;strong&gt;Download DP&lt;/strong&gt;. My 11.6 MB original came out as a 1024 × 1024 JPEG of about 133 KB, which any app will accept without complaining.&lt;/p&gt;

&lt;p&gt;Before uploading, I masked both versions with a circle to see roughly what the app will show. On the left is the original with its center square taken (what you get if you just accept the app's crop), and on the right is the Full DP export:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffgy5op1jritkyd3avpxw.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffgy5op1jritkyd3avpxw.png" alt="Circle mask comparison: original 2:3 photo cropped to its center square versus the Full DP export at 80% photo size" width="799" height="444"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The cropped version cuts into the hair and loses most of the arms. The padded version keeps the whole person. The trade-off is that he's smaller in the circle, which matters at 40 pixels in a chat list.&lt;/p&gt;

&lt;h2&gt;
  
  
  When padding is the wrong call
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Tight headshots.&lt;/strong&gt; If the face already fills most of the frame, padding just makes it smaller. Choose the crop yourself with &lt;a href="https://squareimage.dev/squarecrop/" rel="noopener noreferrer"&gt;Square Crop&lt;/a&gt; and move the box so the face sits in the middle.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Very wide photos.&lt;/strong&gt; A 16:9 landscape padded to a square leaves big bands above and below, and the person turns into a speck. Crop it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Busy backgrounds.&lt;/strong&gt; A blurred copy of a cluttered background looks muddy. Switch to Color, or tune the blur strength in the &lt;a href="https://squareimage.dev/blurredframe/" rel="noopener noreferrer"&gt;Blurred Frame editor&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Low-resolution originals.&lt;/strong&gt; Fitting a 300-pixel photo into a 1024 square doesn't add detail.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you need a square for a post rather than an avatar, the &lt;a href="https://squareimage.dev/" rel="noopener noreferrer"&gt;main square converter on squareimage.dev&lt;/a&gt; does the same blur, color, resize and crop modes and defaults to 1200 px.&lt;/p&gt;

&lt;p&gt;Related on Medium: &lt;a href="https://medium.com/@videocompress/linkedin-and-whatsapp-avatars-that-keep-the-whole-face-0f674ae10cd9" rel="noopener noreferrer"&gt;an earlier take on LinkedIn and WhatsApp avatars that pads with blur first&lt;/a&gt; and &lt;a href="https://medium.com/@videocompress/instagram-1-1-image-without-cutting-off-the-subject-f3d060eec2a9" rel="noopener noreferrer"&gt;the Instagram 1:1 version of the same padding idea&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Sample photo: "Rob Headshot HQ", Wikimedia Commons, CC0.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>tutorial</category>
      <category>productivity</category>
      <category>design</category>
      <category>beginners</category>
    </item>
    <item>
      <title>Convert images to WebP or AVIF for faster pages (and check which one is actually smaller)</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Tue, 06 Oct 2026 11:57:17 +0000</pubDate>
      <link>https://dev.to/videocompress/convert-images-to-webp-or-avif-for-faster-pages-and-check-which-one-is-actually-smaller-5aed</link>
      <guid>https://dev.to/videocompress/convert-images-to-webp-or-avif-for-faster-pages-and-check-which-one-is-actually-smaller-5aed</guid>
      <description>&lt;p&gt;PageSpeed Insights flags your images ("Improve image delivery" now, "Serve images in next-gen formats" in older reports), and the standard advice is to convert everything to WebP, or to AVIF if you want the smallest files. I wanted to see whether that holds up, so I converted one batch to WebP, AVIF and JPG at the same time and compared the sizes.&lt;/p&gt;

&lt;p&gt;AVIF won once out of four. For three big camera photos, the plain JPG beat both newer formats. So this post is less "use format X" and more "measure your own images before you touch your markup", which takes about a minute with the right tool.&lt;/p&gt;

&lt;h2&gt;
  
  
  The scoreboard
&lt;/h2&gt;

&lt;p&gt;The batch was four files, picked to look like what a typical site serves: three 4032×3024 phone photos (a pond, a land-art park, a railroad park; lots of grass and leaves, 6.5–6.7 MB each) and one 2880×1628 PNG screenshot of a browser window (1.9 MB). All four are CC0 from Wikimedia Commons. Here's what came out:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;File&lt;/th&gt;
&lt;th&gt;Original&lt;/th&gt;
&lt;th&gt;WebP&lt;/th&gt;
&lt;th&gt;AVIF&lt;/th&gt;
&lt;th&gt;JPG&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;lake-pond.jpg&lt;/td&gt;
&lt;td&gt;6.7 MB&lt;/td&gt;
&lt;td&gt;2.5 MB (−62%)&lt;/td&gt;
&lt;td&gt;3.5 MB (−48%)&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;2.1 MB (−69%)&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;earthworks.jpg&lt;/td&gt;
&lt;td&gt;6.5 MB&lt;/td&gt;
&lt;td&gt;2.9 MB (−55%)&lt;/td&gt;
&lt;td&gt;3.8 MB (−41%)&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;2.2 MB (−66%)&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;rail-park.jpg&lt;/td&gt;
&lt;td&gt;6.5 MB&lt;/td&gt;
&lt;td&gt;2.6 MB (−61%)&lt;/td&gt;
&lt;td&gt;3.5 MB (−47%)&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;2.1 MB (−68%)&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;browser-screenshot.png&lt;/td&gt;
&lt;td&gt;1.9 MB&lt;/td&gt;
&lt;td&gt;219.2 KB (−88%)&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;170.0 KB (−91%)&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;253.8 KB (−86%)&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Taking the smallest version of each file, the batch went from 21.6 MB to about 6.6 MB ("Saved 15.0 MB total · 70% smaller"). Every output kept its original dimensions.&lt;/p&gt;

&lt;p&gt;I wondered whether the photo result was a quirk of full camera resolution, so I resized the three photos to 1600×1200 with a script and ran them again. The order didn't change: JPG 381–446 KB, WebP 538–597 KB, AVIF 672–749 KB.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why "just use AVIF" didn't hold here
&lt;/h2&gt;

&lt;p&gt;PixSip's &lt;a href="https://pixsip.com/blog/image-format-guide" rel="noopener noreferrer"&gt;image format guide&lt;/a&gt; gives the usual rules of thumb: JPG or WebP for photos, PNG for screenshots and UI, WebP or AVIF when you want the smallest website images. All current major browsers decode both WebP and AVIF, so browser support isn't really the obstacle anymore. Whether the file is actually smaller is.&lt;/p&gt;

&lt;p&gt;That depends on the encoder settings as much as on the format. PixSip has no quality slider. Each format gets one fixed quality picked by its developers, so this table isn't a lab comparison at matched quality. It shows what you'd actually get from the tool. With those settings, MozJPEG was the most aggressive on detailed outdoor photos, while the screenshot's flat colors and sharp text played to AVIF's and WebP's strengths.&lt;/p&gt;

&lt;h2&gt;
  
  
  Run the same test on your own images
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Pick the formats before you drop anything
&lt;/h3&gt;

&lt;p&gt;Open &lt;a href="https://pixsip.com/webp-converter" rel="noopener noreferrer"&gt;PixSip's WebP converter&lt;/a&gt;. It takes JPG, PNG, HEIC, WebP and AVIF, up to 20 files per batch and 10 MB per file on the free plan. Nothing gets uploaded: libwebp, libavif and MozJPEG run as WebAssembly inside the tab.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F1oz4bgai6nywrafj6si6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F1oz4bgai6nywrafj6si6.png" alt="PixSip WebP converter page with the drop area" width="800" height="450"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Before dropping files, look under the drop area. "Convert my images automatically" is already on, with WEBP selected. Click &lt;strong&gt;AVIF&lt;/strong&gt; and &lt;strong&gt;JPG&lt;/strong&gt; too. The free plan writes up to three formats per image, which is exactly the comparison you need: two next-gen candidates plus a JPG that doubles as the fallback. The same switch sits on the &lt;a href="https://pixsip.com/" rel="noopener noreferrer"&gt;PixSip homepage&lt;/a&gt;, where JPEG XL is also an option (all five formats at once needs the Ultra plan).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F52l13kwoqyyjysp70dg7.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F52l13kwoqyyjysp70dg7.png" alt="Convert switch on with AVIF, WEBP and JPG selected" width="798" height="136"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Drop the batch and let the check marks pick
&lt;/h3&gt;

&lt;p&gt;Now drop the files. My four took about 30 seconds; AVIF is the slow one. Each file gets one box per format with its size and saving, and a check mark lands on the smallest. You don't have to do any arithmetic.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fl0apt3fghcaw4d42jsas.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fl0apt3fghcaw4d42jsas.png" alt="Results: JPG smallest for the three photos, AVIF smallest for the PNG screenshot" width="799" height="345"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Zoom in on the winner
&lt;/h3&gt;

&lt;p&gt;Click &lt;strong&gt;Compare&lt;/strong&gt; on a row to see the original next to the smallest output (here the 1.9 MB PNG against the 170 KB AVIF), with a drag divider and a &lt;strong&gt;Zoom 1:1&lt;/strong&gt; button. Small text in screenshots is the first thing to go soft, so check it at 1:1 before you commit.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F2we92hoxvfsjtqbt0sew.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F2we92hoxvfsjtqbt0sew.png" alt="Compare view: 1.9 MB PNG original against the 170.0 KB AVIF" width="800" height="575"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Download 4 as ZIP&lt;/strong&gt; gives you every format of every file, named like &lt;code&gt;browser-screenshot-compressed.avif&lt;/code&gt;, so you can keep the winners and the fallbacks.&lt;/p&gt;

&lt;h2&gt;
  
  
  Ship only the winners
&lt;/h2&gt;

&lt;p&gt;Use &lt;code&gt;&amp;lt;picture&amp;gt;&lt;/code&gt; only where a next-gen format actually won. For the screenshot, that's AVIF first, WebP second, and the JPG as the fallback every browser can show:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight html"&gt;&lt;code&gt;&lt;span class="nt"&gt;&amp;lt;picture&amp;gt;&lt;/span&gt;
  &lt;span class="nt"&gt;&amp;lt;source&lt;/span&gt; &lt;span class="na"&gt;srcset=&lt;/span&gt;&lt;span class="s"&gt;"/img/browser-screenshot.avif"&lt;/span&gt; &lt;span class="na"&gt;type=&lt;/span&gt;&lt;span class="s"&gt;"image/avif"&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&lt;/span&gt;
  &lt;span class="nt"&gt;&amp;lt;source&lt;/span&gt; &lt;span class="na"&gt;srcset=&lt;/span&gt;&lt;span class="s"&gt;"/img/browser-screenshot.webp"&lt;/span&gt; &lt;span class="na"&gt;type=&lt;/span&gt;&lt;span class="s"&gt;"image/webp"&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&lt;/span&gt;
  &lt;span class="nt"&gt;&amp;lt;img&lt;/span&gt; &lt;span class="na"&gt;src=&lt;/span&gt;&lt;span class="s"&gt;"/img/browser-screenshot.jpg"&lt;/span&gt; &lt;span class="na"&gt;alt=&lt;/span&gt;&lt;span class="s"&gt;"Arc browser showing a Wikipedia article"&lt;/span&gt;
       &lt;span class="na"&gt;width=&lt;/span&gt;&lt;span class="s"&gt;"2880"&lt;/span&gt; &lt;span class="na"&gt;height=&lt;/span&gt;&lt;span class="s"&gt;"1628"&lt;/span&gt; &lt;span class="na"&gt;loading=&lt;/span&gt;&lt;span class="s"&gt;"lazy"&lt;/span&gt; &lt;span class="na"&gt;decoding=&lt;/span&gt;&lt;span class="s"&gt;"async"&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&lt;/span&gt;
&lt;span class="nt"&gt;&amp;lt;/picture&amp;gt;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The browser uses the first &lt;code&gt;&amp;lt;source&amp;gt;&lt;/code&gt; whose &lt;code&gt;type&lt;/code&gt; it supports and skips the rest, so put the smallest one first. Keep &lt;code&gt;width&lt;/code&gt; and &lt;code&gt;height&lt;/code&gt; on the &lt;code&gt;&amp;lt;img&amp;gt;&lt;/code&gt; so the layout doesn't jump while the image loads.&lt;/p&gt;

&lt;p&gt;For the three photos, I'd skip &lt;code&gt;&amp;lt;picture&amp;gt;&lt;/code&gt; and serve the compressed JPG. It was the smallest file, and it's one less format to keep in sync. Resizing before you compress matters more anyway, because a 4032 px photo in a 1200 px column is the bigger waste.&lt;/p&gt;

&lt;p&gt;If JPG keeps winning for your photos too, PixSip's &lt;a href="https://pixsip.com/compress-jpg" rel="noopener noreferrer"&gt;JPG compressor&lt;/a&gt; gets you there with one less click.&lt;/p&gt;

&lt;p&gt;Related: &lt;a href="https://medium.com/@videocompress/your-wordpress-images-are-too-heavy-compress-the-jpgs-in-your-browser-before-you-upload-2da9127005a6" rel="noopener noreferrer"&gt;the WordPress JPG compression walkthrough on Medium&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>performance</category>
      <category>images</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>How to convert iPhone HEIC photos to JPG without uploading them</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Thu, 01 Oct 2026 13:01:49 +0000</pubDate>
      <link>https://dev.to/videocompress/how-to-convert-iphone-heic-photos-to-jpg-without-uploading-them-af</link>
      <guid>https://dev.to/videocompress/how-to-convert-iphone-heic-photos-to-jpg-without-uploading-them-af</guid>
      <description>&lt;p&gt;You AirDrop or email a few iPhone photos to a Windows laptop, and the files end in &lt;code&gt;.heic&lt;/code&gt;. Photos won't open them, the upload form on some website rejects them, and the colleague you forwarded them to asks what format that is.&lt;/p&gt;

&lt;p&gt;HEIC is Apple's default camera format. It's efficient, but outside Apple's own apps support is patchy. The usual fix is an "online HEIC converter," which means sending your personal photos to someone's server. You don't need to. Here's how to do it in the browser with the files staying on your machine, and how to come out with much smaller files while you're at it.&lt;/p&gt;

&lt;p&gt;For this walkthrough I used three full-resolution iPhone photos (shot on an iPhone X, 12 Pro Max and 13 Pro, CC0 on Wikimedia Commons), saved as HEIC: 4032×3024 pixels, 3.1 MB, 4.1 MB and 4.7 MB.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Open the converter and drop in your HEIC files
&lt;/h2&gt;

&lt;p&gt;Go to the &lt;a href="https://pixsip.com/heic-to-jpg" rel="noopener noreferrer"&gt;HEIC to JPG converter on PixSip&lt;/a&gt;. No account needed. Drag your &lt;code&gt;.heic&lt;/code&gt; files onto the drop area, or click it to pick them. The free plan takes up to 20 files at a time, each up to 10 MB, so full-size iPhone photos fit.&lt;/p&gt;

&lt;p&gt;The decoding happens in the tab (WebAssembly), so nothing gets uploaded. If you want to check that yourself, load the page, switch off Wi-Fi, and then drop the files. It still works.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4ta8lfg707hlrjnhbjt9.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4ta8lfg707hlrjnhbjt9.png" alt="PixSip HEIC to JPG page with the drop area" width="800" height="450"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Pick your output formats: JPG, plus WebP and AVIF if you can use them
&lt;/h2&gt;

&lt;p&gt;Below the drop area there's a "Convert my images automatically" switch with a row of formats. On this page JPG is already selected. There's no quality slider: PixSip encodes at one quality setting that it tuned against TinyPNG, and that setting is what does the shrinking.&lt;/p&gt;

&lt;p&gt;I also ticked &lt;strong&gt;WebP&lt;/strong&gt; and &lt;strong&gt;AVIF&lt;/strong&gt;. The free plan writes up to three formats side by side, so you get all three versions of each photo from one drop and can keep whichever is smallest. Use JPG if the file is going somewhere picky (an old upload form, a printer, an email to someone on an old PC). WebP is fine for most websites and chat apps.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn6pp06arka0aluay7uyz.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn6pp06arka0aluay7uyz.png" alt="Convert switch on, with JPG, WebP and AVIF selected" width="798" height="136"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Check the results
&lt;/h2&gt;

&lt;p&gt;Each photo gets a row with one box per format: the new size, the percentage saved, and a download button. A check mark sits on the smallest version of each photo. My three photos came out like this:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Photo&lt;/th&gt;
&lt;th&gt;HEIC&lt;/th&gt;
&lt;th&gt;JPG&lt;/th&gt;
&lt;th&gt;WebP&lt;/th&gt;
&lt;th&gt;AVIF&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;IMG_7364 (4032×3024)&lt;/td&gt;
&lt;td&gt;3.1 MB&lt;/td&gt;
&lt;td&gt;806.5 KB (−74%)&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;684.2 KB (−78%)&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;896.5 KB (−71%)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IMG_8097 (3024×4032)&lt;/td&gt;
&lt;td&gt;4.1 MB&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;786.6 KB (−81%)&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;879.5 KB (−79%)&lt;/td&gt;
&lt;td&gt;1.4 MB (−66%)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IMG_MILAN (3024×4032)&lt;/td&gt;
&lt;td&gt;4.7 MB&lt;/td&gt;
&lt;td&gt;645.5 KB (−86%)&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;548.7 KB (−88%)&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;728.6 KB (−85%)&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Taking the smallest file for each photo, the batch went from 11.9 MB to about 2 MB (PixSip's summary: "Saved 9.9 MB total · 83% smaller"). I didn't resize anything, and PixSip doesn't have a resize option. The outputs are still 4032×3024. All of the saving comes from re-encoding at that calibrated quality.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0ln4bnz09jw7vgty3xro.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0ln4bnz09jw7vgty3xro.png" alt="Three HEIC photos converted to JPG, WebP and AVIF, each 66–88% smaller" width="799" height="286"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;A caveat so you know what to expect: how much you save depends on how heavily the HEIC was compressed to begin with. In an earlier run with small (1440×960) HEIC samples that were already tightly compressed, the JPGs came out only 1–42% smaller. A full-size iPhone HEIC can also come out &lt;em&gt;bigger&lt;/em&gt; as a JPG, because HEIC is the more efficient format. If that happens, try the WebP version.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Compare before you download (optional)
&lt;/h2&gt;

&lt;p&gt;Click &lt;strong&gt;Compare&lt;/strong&gt; on any row to get a slider view: the decoded original on one side, the smallest output on the other (here the 684 KB WebP against the 3.1 MB original). Drag the divider, or use &lt;strong&gt;Zoom 1:1&lt;/strong&gt;, to check fine detail like text, skies and skin tones before you send the file to anyone.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5k19t6vljonyzeovwukv.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5k19t6vljonyzeovwukv.png" alt="Before/after slider: 3.1 MB original vs 684.2 KB WebP" width="800" height="509"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Download one file or the whole batch
&lt;/h2&gt;

&lt;p&gt;Click the button in any format box to save just that file, or &lt;strong&gt;Download 3 as ZIP&lt;/strong&gt; (the number matches your batch) to grab everything. The ZIP is put together in the browser too, with every format you picked, named like &lt;code&gt;IMG_7364-compressed.jpg&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Metadata such as GPS location is stripped automatically. That's usually what you want before posting a photo publicly, but keep the original HEIC if you need the date and location later.&lt;/p&gt;

&lt;h2&gt;
  
  
  If you only need to &lt;em&gt;view&lt;/em&gt; HEIC on Windows
&lt;/h2&gt;

&lt;p&gt;Converting isn't always the answer. If you just want to open the photos on your own PC, Microsoft's HEIF/HEVC extensions or an iPhone camera setting might suit you better. This &lt;a href="https://pixsip.com/blog/how-to-open-heic-on-windows" rel="noopener noreferrer"&gt;walkthrough of the three ways to open HEIC files on Windows&lt;/a&gt; compares them. And if you're curious why the tool runs locally instead of on a server, the team explains it in &lt;a href="https://pixsip.com/blog/privacy-by-default" rel="noopener noreferrer"&gt;why PixSip doesn't upload your photos&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Related: &lt;a href="https://medium.com/@videocompress/your-wordpress-images-are-too-heavy-compress-the-jpgs-in-your-browser-before-you-upload-2da9127005a6" rel="noopener noreferrer"&gt;compressing JPGs for a WordPress site before you upload them&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>iphone</category>
      <category>tutorial</category>
      <category>images</category>
      <category>privacy</category>
    </item>
    <item>
      <title>H.264 vs H.265: which codec when the size target still fails</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Mon, 21 Sep 2026 02:23:31 +0000</pubDate>
      <link>https://dev.to/videocompress/h264-vs-h265-which-codec-when-the-size-target-still-fails-4l1l</link>
      <guid>https://dev.to/videocompress/h264-vs-h265-which-codec-when-the-size-target-still-fails-4l1l</guid>
      <description>

&lt;p&gt;You already set a megabyte cap and the clip is still too big—or too slow to finish in a browser tab. That is when codec choice matters. Use H.264 when recipients must play it everywhere; switch to H.265 only when both ends support it and you need another size cut.&lt;/p&gt;

&lt;p&gt;This walkthrough uses &lt;a href="https://vidsmaller.com" rel="noopener noreferrer"&gt;VidSmaller&lt;/a&gt;, a cloud compressor that hits an exact output size (Advanced mode unlocks H.265 on paid plans).&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Upload once, keep the laptop cool
&lt;/h2&gt;

&lt;p&gt;Open &lt;a href="https://vidsmaller.com" rel="noopener noreferrer"&gt;vidsmaller.com&lt;/a&gt;. Drop MP4, MOV, MKV, WebM, or another supported format. Guest accounts have per-file limits and limited &lt;a href="https://vidsmaller.com/pricing" rel="noopener noreferrer"&gt;credits&lt;/a&gt;. There is no watermark; uploads are deleted after processing.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7eqwoy1vu665aa3i11my.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7eqwoy1vu665aa3i11my.png" alt="VidSmaller cloud upload drop zone" width="800" height="410"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Long 4K timelines belong here—not in a browser tab that can OOM mid-encode.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Hit the size first, then choose the codec
&lt;/h2&gt;

&lt;p&gt;In &lt;strong&gt;Basic&lt;/strong&gt;, set &lt;strong&gt;Output file size&lt;/strong&gt; or tap Email / WhatsApp / Discord / Reddit so the two-pass job aims at a real number.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv598irgha6bh9xq2h1yo.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv598irgha6bh9xq2h1yo.png" alt="Size presets and Advanced codec options" width="800" height="410"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Then open &lt;strong&gt;Advanced&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;H.264 + MP4&lt;/strong&gt; — default. Widest phone, chat app, and browser playback.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;H.265&lt;/strong&gt; — smaller file at the same quality, slower encode, and older devices may refuse it.&lt;/li&gt;
&lt;li&gt;Drop resolution (4K → 1080p or 720p) &lt;strong&gt;before&lt;/strong&gt; you flip codecs if the source is huge.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For hard caps (forms, WhatsApp comfort), target about 1MB under the published limit so overhead does not push you over.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Download and verify playback
&lt;/h2&gt;

&lt;p&gt;Start compression, wait for the cloud job, download. You can close the tab; the result link stays until it expires.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdzhu4p3adg6v861hzusw.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdzhu4p3adg6v861hzusw.png" alt="Compressed result ready to download" width="800" height="410"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Spot-check on the device that will receive the file. If H.265 fails to play, re-run the same size target with H.264 (or one resolution step lower).&lt;/p&gt;

&lt;p&gt;Related: &lt;a href="https://medium.com/@videocompress/when-the-browser-runs-out-of-memory-cloud-compress-instead-ae5220f4d777" rel="noopener noreferrer"&gt;why long 4K encodes belong in the cloud, not in a browser tab&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Sensitive short clips that fit in device memory can stay local: &lt;a href="https://videocompress.dev" rel="noopener noreferrer"&gt;videocompress.dev&lt;/a&gt;.
&lt;/h2&gt;

</description>
      <category>video</category>
      <category>compression</category>
      <category>tutorial</category>
      <category>webdev</category>
    </item>
    <item>
      <title>Shrink a 4K phone video for WhatsApp or email</title>
      <dc:creator>videocompress</dc:creator>
      <pubDate>Sun, 20 Sep 2026 10:42:36 +0000</pubDate>
      <link>https://dev.to/videocompress/shrink-a-4k-phone-video-for-whatsapp-or-email-4ig3</link>
      <guid>https://dev.to/videocompress/shrink-a-4k-phone-video-for-whatsapp-or-email-4ig3</guid>
      <description>&lt;p&gt;4K clips blow past WhatsApp and mailbox limits fast. Knock resolution down first, then set an exact megabyte target in the cloud so a long encode does not kill the browser tab.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Upload the 4K source
&lt;/h2&gt;

&lt;p&gt;Open &lt;a href="https://vidsmaller.com" rel="noopener noreferrer"&gt;vidsmaller.com&lt;/a&gt;. Drop MP4, MOV, or another supported file. Guests get large per-file limits and limited &lt;a href="https://vidsmaller.com/pricing" rel="noopener noreferrer"&gt;credits&lt;/a&gt;. No watermark; uploads are removed after the job.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fy45gpwrr4qtq9sor35tn.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fy45gpwrr4qtq9sor35tn.png" alt="VidSmaller upload drop zone" width="800" height="410"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Cap the size, then fix resolution
&lt;/h2&gt;

&lt;p&gt;In &lt;strong&gt;Basic&lt;/strong&gt;, tap WhatsApp or Email, or drag &lt;strong&gt;Output file size&lt;/strong&gt; (often ~16MB for WhatsApp comfort, ~18–20MB toward a 25MB mailbox).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fk8e9pc7gdikm19zzuxkf.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fk8e9pc7gdikm19zzuxkf.png" alt="Size presets and slider" width="800" height="410"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;In &lt;strong&gt;Advanced&lt;/strong&gt;, set 1080p (or 720p) before chasing codecs. Prefer H.264 + MP4 for playback. H.265 can go smaller but older phones may fail.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Download and send
&lt;/h2&gt;

&lt;p&gt;Start the job, wait, download. You can leave the page; the result link stays until it expires.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fny5m5ct4ny6q5axe3z5k.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fny5m5ct4ny6q5axe3z5k.png" alt="Compressed result ready" width="800" height="410"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Still too big? Drop one more resolution step and re-run before switching codecs.&lt;/p&gt;

&lt;p&gt;Short private clips that fit in memory can stay local: &lt;a href="https://videocompress.dev" rel="noopener noreferrer"&gt;videocompress.dev&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Related: &lt;a href="https://medium.com/@videocompress/when-the-browser-runs-out-of-memory-cloud-compress-instead-ae5220f4d777" rel="noopener noreferrer"&gt;moving a long encode to the cloud when the tab crashes&lt;/a&gt; and &lt;a href="https://medium.com/@videocompress/h-264-or-h-265-when-the-file-still-will-not-fit-daf82b2e093e" rel="noopener noreferrer"&gt;a quick H.264 vs H.265 decision guide&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>video</category>
      <category>webdev</category>
      <category>productivity</category>
      <category>tutorial</category>
    </item>
  </channel>
</rss>
