<?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: MirrorCut</title>
    <description>The latest articles on DEV Community by MirrorCut (@mirrorcut).</description>
    <link>https://dev.to/mirrorcut</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%2F4202320%2F7a62d535-6989-430d-9783-c664ab3e175d.png</url>
      <title>DEV Community: MirrorCut</title>
      <link>https://dev.to/mirrorcut</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/mirrorcut"/>
    <language>en</language>
    <item>
      <title>Accept or Revert: Designing AI Edits as Reviewable Patches on a Real Video Timeline</title>
      <dc:creator>MirrorCut</dc:creator>
      <pubDate>Sun, 01 Nov 2026 14:10:45 +0000</pubDate>
      <link>https://dev.to/mirrorcut/accept-or-revert-designing-ai-edits-as-reviewable-patches-on-a-real-video-timeline-48jl</link>
      <guid>https://dev.to/mirrorcut/accept-or-revert-designing-ai-edits-as-reviewable-patches-on-a-real-video-timeline-48jl</guid>
      <description>&lt;p&gt;&lt;em&gt;By Sagar Dey, founder of MirrorCut&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Most AI features ship with an all-or-nothing contract. The model does a thing, and you either keep the result or throw it away.&lt;/p&gt;

&lt;p&gt;That works for a one-line autocomplete. It falls apart for a video edit, where a single "AI pass" might contain dozens of independent decisions: cut this pause, drop that filler word, keep the second take, caption this line.&lt;/p&gt;

&lt;p&gt;When I built MirrorCut, an AI video editor that runs in the browser, I made one rule early: &lt;strong&gt;every AI change is a patch on the timeline, and every patch can be accepted or reverted on its own.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;This post is about why, and the design constraints that rule creates. It's written for developers building any kind of AI-assisted editor: video, docs, code or design.&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Note:&lt;/strong&gt; the code below is a simplified illustration of the pattern, written for this post. It is not MirrorCut's production code.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h2&gt;
  
  
  The problem with "re-render and replace"
&lt;/h2&gt;

&lt;p&gt;The simplest way to add AI to an editor looks like this:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Send the current state to the model.&lt;/li&gt;
&lt;li&gt;Get back a new state.&lt;/li&gt;
&lt;li&gt;Replace the old state.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;It's easy to build. It's also hostile to the user, for three reasons:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;No granularity.&lt;/strong&gt; If the model made ten good decisions and one bad one, the user can't separate them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Lost work.&lt;/strong&gt; Anything the user did by hand can get silently overwritten.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;No explanation.&lt;/strong&gt; The user sees a different result, not &lt;em&gt;what changed&lt;/em&gt; or &lt;em&gt;why&lt;/em&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;In video, there's a fourth problem: re-rendering is slow and expensive. Throwing away a whole pass because of one bad cut costs real time.&lt;/p&gt;

&lt;h2&gt;
  
  
  The alternative: edits as small, named patches
&lt;/h2&gt;

&lt;p&gt;Instead of returning a new timeline, the agent proposes a list of small changes against the existing one. Conceptually, each patch carries:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;&lt;span class="c1"&gt;// Illustrative only&lt;/span&gt;
&lt;span class="kd"&gt;type&lt;/span&gt; &lt;span class="nx"&gt;EditPatch&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="na"&gt;id&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kr"&gt;string&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nl"&gt;kind&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;cut&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;caption&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;reframe&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;graphic&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nl"&gt;range&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="na"&gt;start&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kr"&gt;number&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="nl"&gt;end&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kr"&gt;number&lt;/span&gt; &lt;span class="p"&gt;};&lt;/span&gt; &lt;span class="c1"&gt;// seconds on the timeline&lt;/span&gt;
  &lt;span class="nl"&gt;summary&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kr"&gt;string&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="c1"&gt;// one human-readable line&lt;/span&gt;
  &lt;span class="nl"&gt;status&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;pending&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;accepted&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;reverted&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;};&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The &lt;code&gt;summary&lt;/code&gt; matters more than it looks. In MirrorCut, every AI change is explained in one line, for example &lt;em&gt;"Removing a 0.6 s pause at 0:54"&lt;/em&gt;. If you can't describe a change in one line, it's probably several changes and should be split.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 1: Show every patch where it happens
&lt;/h2&gt;

&lt;p&gt;Each pending change is outlined on the timeline itself, not buried in a side panel. The user's eyes are already on the timeline, so that's where the review should happen.&lt;/p&gt;

&lt;p&gt;This also forces a useful discipline on the agent: every change has to map to a visible location. "Improved the pacing" isn't a patch. "Removed 2.1 s of dead air at 3:12" is.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 2: Accept and Revert are per patch
&lt;/h2&gt;

&lt;p&gt;The whole point is granularity. A pass with ten patches gives the user ten decisions, not one.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;&lt;span class="c1"&gt;// Illustrative only&lt;/span&gt;
&lt;span class="kd"&gt;function&lt;/span&gt; &lt;span class="nf"&gt;revert&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;patchId&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kr"&gt;string&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="nx"&gt;patch&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nx"&gt;timeline&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;patches&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;patchId&lt;/span&gt;&lt;span class="p"&gt;);&lt;/span&gt;
  &lt;span class="nx"&gt;timeline&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;apply&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nf"&gt;inverse&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;patch&lt;/span&gt;&lt;span class="p"&gt;));&lt;/span&gt;
  &lt;span class="nx"&gt;patch&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;status&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;reverted&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The hard part isn't the button. It's making sure patches are &lt;strong&gt;independent enough&lt;/strong&gt; that reverting one doesn't break the others. Cutting 0.6 s at 0:54 shifts everything after it, so later patches need to be expressed in a way that survives earlier ones being reverted.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 3: One undo history for humans and AI
&lt;/h2&gt;

&lt;p&gt;This is the rule people find most surprising.&lt;/p&gt;

&lt;p&gt;In MirrorCut, AI edits and manual edits share one undo history. There's no separate "AI layer" or "AI mode." If you trim a clip by hand, then the agent removes a pause, then you press undo, you undo the pause removal, exactly like you'd expect.&lt;/p&gt;

&lt;p&gt;Why not keep them separate? Because two histories create two mental models. Users end up asking "which undo do I need?" And they lose trust when the AI's changes behave differently from their own.&lt;/p&gt;

&lt;p&gt;A cut is a cut, whoever made it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 4: Follow-ups patch, they don't restart
&lt;/h2&gt;

&lt;p&gt;When the user asks for a follow-up, like &lt;em&gt;"make the intro faster and add captions"&lt;/em&gt;, the agent edits the existing timeline instead of regenerating from scratch.&lt;/p&gt;

&lt;p&gt;This protects everything the user already accepted or adjusted by hand. A restart would quietly throw that away.&lt;/p&gt;

&lt;p&gt;It also keeps responses small and reviewable. The site shows this flow: "Done. Intro trimmed from 4s to 2s and captions added." That's two patches, not a new video.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 5: Expensive operations ask first
&lt;/h2&gt;

&lt;p&gt;Some patches are cheap to preview. Others, like a full-resolution render, take real compute.&lt;/p&gt;

&lt;p&gt;In MirrorCut, expensive steps ask before running. The general pattern: &lt;strong&gt;separate proposing a change from paying for it.&lt;/strong&gt; Let the user review the plan before the costly part kicks off.&lt;/p&gt;

&lt;p&gt;There's a pricing angle too. On MirrorCut's plans, each request to the agent uses AI credits based on the work it takes, while editing by hand on the timeline uses no credits. That makes "do it yourself" a real, free option for small fixes, and it keeps the agent honest about when it's worth calling.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where this pattern applies beyond video
&lt;/h2&gt;

&lt;p&gt;Strip out the video specifics and you get a reusable checklist for any AI-assisted editor:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Return &lt;strong&gt;patches&lt;/strong&gt;, not replacement states.&lt;/li&gt;
&lt;li&gt;Give each patch a &lt;strong&gt;one-line explanation&lt;/strong&gt; and a &lt;strong&gt;visible location&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Make &lt;strong&gt;accept/revert granular&lt;/strong&gt;, and design patches to survive each other's reversal.&lt;/li&gt;
&lt;li&gt;Use &lt;strong&gt;one undo history&lt;/strong&gt; for human and AI changes.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Patch on follow-up&lt;/strong&gt;; never silently restart.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Ask before expensive work.&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Try it on your own footage
&lt;/h2&gt;

&lt;p&gt;MirrorCut, AI video editor (mirrorcut.ai), runs in the browser. It cuts dead air, filler words and retakes, adds word-timed captions, reframes to 16:9, 9:16, 1:1 and 4:5, finds Shorts, and adds motion graphics, all as reviewable patches on a real timeline. Exports are MP4 up to 4K, plus SRT and VTT captions.&lt;/p&gt;

&lt;p&gt;Every account starts with 7 days of the Creator plan free (card required). After that, it's Creator $10, Pro $30 or Studio $60 a month. MirrorCut launches publicly in November 2026: &lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;https://www.mirrorcut.ai&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;I'd love to hear how you handle review flows for AI output in your own tools. Drop it in the comments.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>webdev</category>
      <category>ux</category>
      <category>architecture</category>
    </item>
    <item>
      <title>7 Best Tools for Developer Launch Videos in 2026 (Code vs Camera)</title>
      <dc:creator>MirrorCut</dc:creator>
      <pubDate>Sat, 31 Oct 2026 14:08:38 +0000</pubDate>
      <link>https://dev.to/mirrorcut/7-best-tools-for-developer-launch-videos-in-2026-code-vs-camera-1ge1</link>
      <guid>https://dev.to/mirrorcut/7-best-tools-for-developer-launch-videos-in-2026-code-vs-camera-1ge1</guid>
      <description>&lt;p&gt;&lt;em&gt;Every other tool here is described from its own site, README or help docs, linked at the end and checked Oct 30, 2026.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;If you've scrolled DEV lately, you've seen the pattern. Developers are making launch videos without opening a video editor:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;One writer renders a trailer from a single web page where every frame is "a pure function of time", screenshots each frame with Playwright, and glues them with ffmpeg (&lt;a href="https://dev.to/susanacharya/i-make-my-launch-videos-with-a-web-page-not-a-video-editor-5f32"&gt;post&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;Another argues that "text-to-video is the wrong primitive" and HTML is the right one, because an agent can write HTML &lt;em&gt;exactly&lt;/em&gt; (&lt;a href="https://dev.to/wx4766/your-html-page-is-already-a-video-renderer-it-just-needs-one-global-function-k31"&gt;post&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;Others turn release notes into a product video with Claude Code (&lt;a href="https://dev.to/sheen_417f0f3a7f4/how-i-triage-release-notes-before-claude-code-turns-them-into-a-product-video-1h3"&gt;post&lt;/a&gt;).&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;I think this is a genuinely good idea. I also think it's being stretched to cover a job it can't do. Anthropic's own help page says Claude Motion "doesn't use a video generation model, so it doesn't create realistic footage or people." Code is great at rendering your &lt;strong&gt;product&lt;/strong&gt;. It can't render &lt;strong&gt;you&lt;/strong&gt;: your face when you say why you built this, the pause before the line you actually believe. For indie launches, that human part often carries the trust.&lt;/p&gt;

&lt;p&gt;So a good launch video usually has two halves: a coded product segment and a recorded founder segment. Here are seven tools across both halves, and how to pick.&lt;/p&gt;

&lt;h2&gt;
  
  
  What matters for a developer launch video
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Exact text.&lt;/strong&gt; Feature names, prices and UI labels have to be spelled right. Code renders the words you typed; video models often don't.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Cheap iteration.&lt;/strong&gt; Before a launch, the copy changes daily. A change should be a re-render or a quick re-edit, not a rebuild.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The human part.&lt;/strong&gt; Someone, ideally you, explaining the problem on camera. That needs recording and editing, not rendering.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Reviewability.&lt;/strong&gt; Whether an agent writes your markup or cuts your recording, you want to see what it changed and undo just that part.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Licence and cost.&lt;/strong&gt; Some "free" frameworks have company-size limits. Check before you put them in a pipeline.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  1. MirrorCut: edit the recorded half
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;MirrorCut&lt;/a&gt; is an AI agent that edits your recordings into videos, in the browser. You upload the founder take and the agent makes the first cut: it removes awkward pauses, dead air, filler words, retakes and false starts. It can add captions (48 styles that stay in sync after cuts), reframe to 16:9, 9:16, 1:1 or 4:5 following the speaker, and build an ad around the recording (hook, captions, motion graphics from 200+ templates, an offer end card, music) in the &lt;a href="https://www.mirrorcut.ai/ad-video-editor" rel="noopener noreferrer"&gt;ad editor&lt;/a&gt;.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; every AI change is outlined on the timeline with a one-line reason and its own Accept and Revert. AI edits and your hand edits share one undo history.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Output:&lt;/strong&gt; MP4 up to 4K, plus SRT and VTT caption files.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Creator $10/month, Pro $30, Studio $60. Every plan has every feature; plans differ by how much you can make each month (&lt;a href="https://www.mirrorcut.ai/pricing" rel="noopener noreferrer"&gt;pricing&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; the recorded half of a launch video, and small brand ads built from a founder take.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  2. Remotion: React for video
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://www.remotion.dev/" rel="noopener noreferrer"&gt;Remotion&lt;/a&gt; is the established "write video in React" framework. Each frame is a component rendered at a given time, and you render to MP4 from the command line.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Good at:&lt;/strong&gt; exact text, reusable components, and copy changes that are a re-render. Your video lives in git, so a change is a diff.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Watch out for:&lt;/strong&gt; you're writing the motion design yourself (or prompting an agent to). It renders the product, not you.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Licence:&lt;/strong&gt; free for individuals and for-profit companies with up to 3 employees; larger for-profit companies need a company licence (&lt;a href="https://github.com/remotion-dev/remotion/blob/main/LICENSE.md" rel="noopener noreferrer"&gt;licence&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; teams who already write React and will make more than one video.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  3. HyperFrames: HTML to video, built for agents
&lt;/h2&gt;

&lt;p&gt;HeyGen open-sourced &lt;a href="https://github.com/heygen-com/hyperframes" rel="noopener noreferrer"&gt;HyperFrames&lt;/a&gt; ("Write HTML. Render video. Built for agents.").&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Good at:&lt;/strong&gt; plain HTML/CSS as the source, which is what LLM agents write most reliably.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Watch out for:&lt;/strong&gt; the same judgment problems as any agent-written video (see #4). Someone still has to watch it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Licence:&lt;/strong&gt; Apache 2.0.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; agent-driven pipelines where Claude Code or another agent writes the scenes.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  4. Claude Motion: short animations from a chat
&lt;/h2&gt;

&lt;p&gt;Anthropic's &lt;a href="https://support.claude.com/en/articles/17454997-get-started-with-claude-motion" rel="noopener noreferrer"&gt;Claude Motion&lt;/a&gt; builds short animations as code from a conversation.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Good at:&lt;/strong&gt; quick animated explainers and feature callouts without setting up a framework.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Watch out for:&lt;/strong&gt; by design it doesn't create realistic footage or people. A write-up of one solo founder's Claude-made promo describes six iterations over two days: full-page screenshots nobody can read, sound effects on every UI interaction that ended up "too dense, too harsh", and a first cut that toured features instead of arguing why they matter (&lt;a href="https://dev.to/zaradevto/what-breaks-when-claude-makes-your-product-video-and-how-to-fix-it-before-version-6-26no"&gt;post&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; see Anthropic's help page for plan availability.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; a fast coded product segment when you don't want to own a rendering stack.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  5. The DIY web page + Playwright + ffmpeg
&lt;/h2&gt;

&lt;p&gt;The approach from the &lt;a href="https://dev.to/susanacharya/i-make-my-launch-videos-with-a-web-page-not-a-video-editor-5f32"&gt;web-page trailer post&lt;/a&gt;: one page, seek to time &lt;code&gt;t&lt;/code&gt;, screenshot each frame with Playwright, encode with ffmpeg.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Good at:&lt;/strong&gt; total control, zero licence questions, deterministic renders (a slow CI box makes the same frames as your laptop).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Watch out for:&lt;/strong&gt; you own everything, including audio. The author admits they can't hear what the box renders, so they check loudness with &lt;code&gt;ebur128&lt;/code&gt; and get a human to listen before posting.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; free, apart from your time.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; developers who want to understand every frame.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  6. Descript: the transcript-first editor
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://www.descript.com" rel="noopener noreferrer"&gt;Descript&lt;/a&gt; edits video like a doc, and its chat co-editor Underlord handles Remove Filler Words, Remove Retakes, captions and clips.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; Underlord creates a checkpoint you can revert to; per Descript's help docs, checkpoints last only for the current session (&lt;a href="https://help.descript.com/ai-assistant/revert" rel="noopener noreferrer"&gt;help&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hand-off:&lt;/strong&gt; timeline export to Premiere, Final Cut Pro and others.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Free (60 media minutes a month, 720p export); Hobbyist $24/month or $16 billed annually (&lt;a href="https://www.descript.com/pricing" rel="noopener noreferrer"&gt;pricing&lt;/a&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; editing the recorded half if you finish in an NLE.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  7. video-use: a coding-agent skill for your footage
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/browser-use/video-use" rel="noopener noreferrer"&gt;video-use&lt;/a&gt; is an open-source skill that transcribes footage with word timestamps, cuts filler and dead space, and proposes a strategy, then waits for your OK before cutting.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Good at:&lt;/strong&gt; fitting into an agent workflow you already use for the coded half.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Watch out for:&lt;/strong&gt; setup (ffmpeg, a transcription API key, your agent's usage limits) and no timeline to nudge a cut by hand.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; open source; you pay for API and agent usage.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; developers who want the whole launch video, both halves, inside one agent workflow.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Side-by-side
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Tool&lt;/th&gt;
&lt;th&gt;Half it covers&lt;/th&gt;
&lt;th&gt;Source of truth&lt;/th&gt;
&lt;th&gt;How you review it&lt;/th&gt;
&lt;th&gt;Cost to start*&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;MirrorCut&lt;/td&gt;
&lt;td&gt;Recorded&lt;/td&gt;
&lt;td&gt;Your recording + timeline&lt;/td&gt;
&lt;td&gt;Accept/Revert per change&lt;/td&gt;
&lt;td&gt;$10/mo&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Remotion&lt;/td&gt;
&lt;td&gt;Coded&lt;/td&gt;
&lt;td&gt;React components&lt;/td&gt;
&lt;td&gt;Code diff + re-render&lt;/td&gt;
&lt;td&gt;Free (≤3 employees)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;HyperFrames&lt;/td&gt;
&lt;td&gt;Coded&lt;/td&gt;
&lt;td&gt;HTML&lt;/td&gt;
&lt;td&gt;Code diff + re-render&lt;/td&gt;
&lt;td&gt;Free (Apache 2.0)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Claude Motion&lt;/td&gt;
&lt;td&gt;Coded&lt;/td&gt;
&lt;td&gt;Chat-written code&lt;/td&gt;
&lt;td&gt;Re-prompt + re-render&lt;/td&gt;
&lt;td&gt;See Anthropic plans&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;DIY page + ffmpeg&lt;/td&gt;
&lt;td&gt;Coded&lt;/td&gt;
&lt;td&gt;HTML + scripts&lt;/td&gt;
&lt;td&gt;Code diff + re-render&lt;/td&gt;
&lt;td&gt;Free&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Descript&lt;/td&gt;
&lt;td&gt;Recorded&lt;/td&gt;
&lt;td&gt;Transcript + timeline&lt;/td&gt;
&lt;td&gt;Revert per checkpoint (session only)&lt;/td&gt;
&lt;td&gt;Free; $24/mo ($16 annual)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;video-use&lt;/td&gt;
&lt;td&gt;Recorded&lt;/td&gt;
&lt;td&gt;Agent strategy&lt;/td&gt;
&lt;td&gt;Approve the strategy before cutting&lt;/td&gt;
&lt;td&gt;API + agent usage&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;*Checked Oct 30, 2026, from each project's own pages. Prices and licences change, so click through first.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which one should you pick?
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;You need the founder part edited fast and you finish in the browser:&lt;/strong&gt; MirrorCut.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You write React and will make many videos:&lt;/strong&gt; Remotion (check the licence for your headcount).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;An agent writes your scenes:&lt;/strong&gt; HyperFrames.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One quick animated segment, no setup:&lt;/strong&gt; Claude Motion.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You want to own every frame:&lt;/strong&gt; the DIY page + ffmpeg route.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You finish in Premiere or Final Cut:&lt;/strong&gt; Descript.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You want everything inside your coding agent:&lt;/strong&gt; video-use.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  A split that works: code for the product, a recording for the person
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Segment&lt;/th&gt;
&lt;th&gt;Length&lt;/th&gt;
&lt;th&gt;Made with&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Hook: you, on camera, one sentence&lt;/td&gt;
&lt;td&gt;3–5 s&lt;/td&gt;
&lt;td&gt;Recording&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Problem: you, explaining it&lt;/td&gt;
&lt;td&gt;8–12 s&lt;/td&gt;
&lt;td&gt;Recording&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Product: UI flow, feature callouts&lt;/td&gt;
&lt;td&gt;10–25 s&lt;/td&gt;
&lt;td&gt;Code (#2–#5)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Offer and CTA&lt;/td&gt;
&lt;td&gt;5–8 s&lt;/td&gt;
&lt;td&gt;Recording or a code end card&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Script the recorded parts, shot-list the coded parts.&lt;/strong&gt; One screen, one claim, one piece of proof per coded shot.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Record in one take, and allow retakes.&lt;/strong&gt; When you fluff a line, pause and say it again. Don't stop recording.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Edit the recording&lt;/strong&gt; with #1, #6 or #7.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Stitch the two halves.&lt;/strong&gt; If both clips match resolution and frame rate, this works:
&lt;/li&gt;
&lt;/ol&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="c"&gt;# founder.mp4 = your edited recording, product.mp4 = your rendered segment&lt;/span&gt;
&lt;span class="c"&gt;# If product.mp4 has no audio track, add one first:&lt;/span&gt;
&lt;span class="c"&gt;#   ffmpeg -i product.mp4 -f lavfi -i anullsrc=r=48000:cl=stereo -shortest -c:v copy -c:a aac product_a.mp4&lt;/span&gt;
ffmpeg &lt;span class="nt"&gt;-i&lt;/span&gt; founder.mp4 &lt;span class="nt"&gt;-i&lt;/span&gt; product.mp4 &lt;span class="nt"&gt;-filter_complex&lt;/span&gt; &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="s2"&gt;"[0:v]scale=1080:1920,setsar=1,fps=30[v0];[1:v]scale=1080:1920,setsar=1,fps=30[v1]; &lt;/span&gt;&lt;span class="se"&gt;\&lt;/span&gt;&lt;span class="s2"&gt;
   [v0][0:a][v1][1:a]concat=n=2:v=1:a=1[v][a]"&lt;/span&gt; &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;-map&lt;/span&gt; &lt;span class="s2"&gt;"[v]"&lt;/span&gt; &lt;span class="nt"&gt;-map&lt;/span&gt; &lt;span class="s2"&gt;"[a]"&lt;/span&gt; &lt;span class="nt"&gt;-c&lt;/span&gt;:v libx264 &lt;span class="nt"&gt;-crf&lt;/span&gt; 18 &lt;span class="nt"&gt;-c&lt;/span&gt;:a aac launch_9x16.mp4
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Render your coded segment at the target size rather than cropping it. Then, whatever made each half, a human watches the whole thing with sound on before it ships. That's the step every failure log in this trend has in common.&lt;/p&gt;

&lt;p&gt;If you've made a launch video this way, I'd like to see it. What did you keep in code, and what did you record? And if you want to try the recorded half, it's at &lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;mirrorcut.ai&lt;/a&gt;.&lt;/p&gt;

&lt;h3&gt;
  
  
  Sources (checked Oct 30, 2026)
&lt;/h3&gt;

&lt;ul&gt;
&lt;li&gt;HTML as a video primitive: &lt;a href="https://dev.to/wx4766/your-html-page-is-already-a-video-renderer-it-just-needs-one-global-function-k31"&gt;https://dev.to/wx4766/your-html-page-is-already-a-video-renderer-it-just-needs-one-global-function-k31&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Launch videos from a web page: &lt;a href="https://dev.to/susanacharya/i-make-my-launch-videos-with-a-web-page-not-a-video-editor-5f32"&gt;https://dev.to/susanacharya/i-make-my-launch-videos-with-a-web-page-not-a-video-editor-5f32&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Six versions of a Claude-made promo: &lt;a href="https://dev.to/zaradevto/what-breaks-when-claude-makes-your-product-video-and-how-to-fix-it-before-version-6-26no"&gt;https://dev.to/zaradevto/what-breaks-when-claude-makes-your-product-video-and-how-to-fix-it-before-version-6-26no&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Release notes to product video: &lt;a href="https://dev.to/sheen_417f0f3a7f4/how-i-triage-release-notes-before-claude-code-turns-them-into-a-product-video-1h3"&gt;https://dev.to/sheen_417f0f3a7f4/how-i-triage-release-notes-before-claude-code-turns-them-into-a-product-video-1h3&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Claude Motion help page: &lt;a href="https://support.claude.com/en/articles/17454997-get-started-with-claude-motion" rel="noopener noreferrer"&gt;https://support.claude.com/en/articles/17454997-get-started-with-claude-motion&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;HyperFrames (Apache 2.0): &lt;a href="https://github.com/heygen-com/hyperframes" rel="noopener noreferrer"&gt;https://github.com/heygen-com/hyperframes&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Remotion: &lt;a href="https://www.remotion.dev/" rel="noopener noreferrer"&gt;https://www.remotion.dev/&lt;/a&gt; · licence: &lt;a href="https://github.com/remotion-dev/remotion/blob/main/LICENSE.md" rel="noopener noreferrer"&gt;https://github.com/remotion-dev/remotion/blob/main/LICENSE.md&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Descript pricing: &lt;a href="https://www.descript.com/pricing" rel="noopener noreferrer"&gt;https://www.descript.com/pricing&lt;/a&gt; · Underlord revert: &lt;a href="https://help.descript.com/ai-assistant/revert" rel="noopener noreferrer"&gt;https://help.descript.com/ai-assistant/revert&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;video-use: &lt;a href="https://github.com/browser-use/video-use" rel="noopener noreferrer"&gt;https://github.com/browser-use/video-use&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;MirrorCut: &lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;https://www.mirrorcut.ai&lt;/a&gt; · &lt;a href="https://www.mirrorcut.ai/pricing" rel="noopener noreferrer"&gt;https://www.mirrorcut.ai/pricing&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>ai</category>
      <category>webdev</category>
      <category>video</category>
      <category>productivity</category>
    </item>
    <item>
      <title>7 Best Tools to Convert Horizontal Video to Vertical (2026)</title>
      <dc:creator>MirrorCut</dc:creator>
      <pubDate>Fri, 30 Oct 2026 15:22:34 +0000</pubDate>
      <link>https://dev.to/mirrorcut/7-best-tools-to-convert-horizontal-video-to-vertical-2026-d5f</link>
      <guid>https://dev.to/mirrorcut/7-best-tools-to-convert-horizontal-video-to-vertical-2026-d5f</guid>
      <description>&lt;p&gt;&lt;em&gt;Every tool here is described from its own website or docs, checked October 30, 2026. Prices and plans change, so click through before you pay.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;You recorded a demo, a meetup talk or a podcast episode in 16:9, because that's what cameras, screen recorders and Zoom give you. Now you want it on Shorts, Reels or TikTok, which are 9:16.&lt;/p&gt;

&lt;p&gt;Post the wide video as it is and it sits in a thin strip across the middle of a phone. Crop it down the middle and you get a different problem: the moment the person on the left starts talking, they're gone.&lt;/p&gt;

&lt;p&gt;Here are seven ways to convert horizontal video to vertical, from AI editors that follow whoever is speaking to a one-line ffmpeg command you can drop into a script.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why a centre crop isn't enough
&lt;/h2&gt;

&lt;p&gt;A 9:16 slice of a 1920×1080 frame is about 608 pixels wide, roughly a third of the picture. That's fine for one person sitting dead centre. It breaks when:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Two people are on screen.&lt;/strong&gt; A host and a guest sit apart, so one fixed window can't hold both.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The speaker moves.&lt;/strong&gt; People lean, gesture and walk to a whiteboard.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;There's text or UI at the edges.&lt;/strong&gt; Slides and screen recordings lose their menus and half their words.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So the real job isn't "crop to 9:16". It's "keep the right thing in frame, shot by shot".&lt;/p&gt;

&lt;h2&gt;
  
  
  The three ways to go vertical
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Fixed crop.&lt;/strong&gt; One window position for the whole clip. Reliable and free, but it doesn't follow anyone.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Tracked crop (auto reframe).&lt;/strong&gt; The window follows a face, the active speaker or a moving object. This is what the AI tools do.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Fit with padding.&lt;/strong&gt; The whole wide frame, shrunk to the phone's width, on a blurred copy of itself or a colour. Nothing is lost, but everything is small.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Most real edits mix all three: tracked crops for talking, a fit for the slide, a fixed crop for the locked-off shot.&lt;/p&gt;

&lt;h2&gt;
  
  
  Quick picks
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Tool&lt;/th&gt;
&lt;th&gt;Best for&lt;/th&gt;
&lt;th&gt;Tracks the speaker?&lt;/th&gt;
&lt;th&gt;Price to reframe&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;MirrorCut&lt;/td&gt;
&lt;td&gt;Talking-head, interview and podcast recordings&lt;/td&gt;
&lt;td&gt;Yes, follows whoever is speaking&lt;/td&gt;
&lt;td&gt;7-day trial (card required), then from $10/mo&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Adobe Premiere Pro&lt;/td&gt;
&lt;td&gt;Editors already in Premiere&lt;/td&gt;
&lt;td&gt;Follows the action (motion presets)&lt;/td&gt;
&lt;td&gt;Premiere Pro subscription&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;DaVinci Resolve Studio&lt;/td&gt;
&lt;td&gt;One-time purchase, full NLE&lt;/td&gt;
&lt;td&gt;Smart Reframe (Studio only)&lt;/td&gt;
&lt;td&gt;$295 one-time&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;CapCut&lt;/td&gt;
&lt;td&gt;Quick edits on desktop or phone&lt;/td&gt;
&lt;td&gt;Subject tracking&lt;/td&gt;
&lt;td&gt;CapCut Pro required&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Kapwing&lt;/td&gt;
&lt;td&gt;Browser editing, multi-speaker calls&lt;/td&gt;
&lt;td&gt;Speaker Focus&lt;/td&gt;
&lt;td&gt;Free to start (720p, watermark)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OpusClip&lt;/td&gt;
&lt;td&gt;Turning long videos into many clips&lt;/td&gt;
&lt;td&gt;ReframeAnything&lt;/td&gt;
&lt;td&gt;Free plan (9:16, watermark); paid from $15/mo&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ffmpeg&lt;/td&gt;
&lt;td&gt;Scripts, batches, CI&lt;/td&gt;
&lt;td&gt;No&lt;/td&gt;
&lt;td&gt;Free, open source&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  The 7 best tools to convert horizontal video to vertical
&lt;/h2&gt;

&lt;h3&gt;
  
  
  1. MirrorCut: reframes around whoever is speaking
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;MirrorCut, AI video editor (mirrorcut.ai)&lt;/a&gt; is an AI agent that edits your recordings into finished videos, and going vertical is one of the things you can simply ask it for. Upload a horizontal recording, type something like &lt;em&gt;"Reframe this to 9:16 and keep the speaker centred"&lt;/em&gt;, and it &lt;a href="https://www.mirrorcut.ai/use-cases/reframe-video-to-vertical" rel="noopener noreferrer"&gt;reframes each shot around the person speaking&lt;/a&gt; instead of using a fixed centre crop. In a host-and-guest conversation it frames whoever is talking and moves to the other person when they start.&lt;/p&gt;

&lt;p&gt;What makes it different from a resize tool is that the reframe is part of an edit, not a separate export:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;One edit, every format.&lt;/strong&gt; Switch the same edit between 16:9, 9:16, 1:1 and 4:5. Captions, logos and graphics are laid out again for each format.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Camera switching.&lt;/strong&gt; With a camera file per person, it cuts to whoever is speaking on each change of voice.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Nothing is baked in.&lt;/strong&gt; Every AI change lands on the timeline, outlined and explained in one line. Accept it, revert it, or fix one badly framed shot by hand or by asking the agent to redo it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The rest of the cleanup in the same place.&lt;/strong&gt; It can also tighten dead air and filler words, add word-by-word captions and &lt;a href="https://www.mirrorcut.ai/use-cases/long-video-to-shorts" rel="noopener noreferrer"&gt;cut Shorts from a long recording&lt;/a&gt;, all as reviewable changes (&lt;a href="https://www.mirrorcut.ai/features" rel="noopener noreferrer"&gt;full feature list&lt;/a&gt;).&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Price:&lt;/strong&gt; 7 days of Creator free (a card is required to start), then Creator $10/month, Pro $30, Studio $60. Every plan has every feature; they differ in credits and hours. &lt;a href="https://www.mirrorcut.ai/pricing" rel="noopener noreferrer"&gt;Pricing&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; you're turning talking-head, interview or podcast recordings into vertical videos and want captions and cleanup done in the same pass.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. Adobe Premiere Pro: Auto Reframe
&lt;/h3&gt;

&lt;p&gt;If you already edit in Premiere, &lt;a href="https://helpx.adobe.com/premiere-pro/using/auto-reframe.html" rel="noopener noreferrer"&gt;Auto Reframe&lt;/a&gt; is built in. Choose &lt;strong&gt;Sequence &amp;gt; Auto Reframe Sequence&lt;/strong&gt;, pick a target aspect ratio, and Premiere creates a duplicate sequence at the new size with the Auto Reframe effect on every clip.&lt;/p&gt;

&lt;p&gt;You choose a motion preset: &lt;strong&gt;Slower Motion&lt;/strong&gt; for talking-head interviews (nearly static, few keyframes), &lt;strong&gt;Default&lt;/strong&gt; for most content, and &lt;strong&gt;Faster Motion&lt;/strong&gt; for sports and other fast action. Adobe notes that complex sequences with several points of interest may need the keyframes fine-tuned afterwards.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Price:&lt;/strong&gt; part of a Premiere Pro subscription.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; your edit already lives in Premiere and you want the vertical version to stay in the same project.&lt;/p&gt;

&lt;h3&gt;
  
  
  3. DaVinci Resolve Studio: Smart Reframe
&lt;/h3&gt;

&lt;p&gt;Resolve's &lt;a href="https://www.blackmagicdesign.com/products/davinciresolve/studio/" rel="noopener noreferrer"&gt;Smart Reframe&lt;/a&gt; uses the DaVinci Neural Engine to keep the subject in frame when you change the timeline's aspect ratio. One catch: it's a &lt;strong&gt;Studio-only&lt;/strong&gt; feature. Blackmagic's own &lt;a href="https://documents.blackmagicdesign.com/SupportNotes/DaVinci_Resolve_Studio_21_Features.pdf" rel="noopener noreferrer"&gt;feature comparison&lt;/a&gt; lists Smart Reframe as Studio on desktop, so the free version of Resolve won't do it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Price:&lt;/strong&gt; DaVinci Resolve Studio is a one-time $295. Blackmagic is running a sale at $236 until November 20, 2026 (&lt;a href="https://www.blackmagicdesign.com/event/davinci-resolve-special" rel="noopener noreferrer"&gt;source&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; you'd rather pay once than subscribe, and you want a full colour-and-edit suite around the reframe.&lt;/p&gt;

&lt;h3&gt;
  
  
  4. CapCut: Auto Reframe
&lt;/h3&gt;

&lt;p&gt;In CapCut desktop, select the clip, open &lt;strong&gt;Video &amp;gt; Auto Reframe&lt;/strong&gt;, choose 9:16, set image stabilisation and camera moving speed, and apply (&lt;a href="https://www.capcut.com/tools/auto-reframe" rel="noopener noreferrer"&gt;CapCut's guide&lt;/a&gt;). It's on mobile and CapCut Online too.&lt;/p&gt;

&lt;p&gt;Know that it isn't free any more: CapCut's help centre says Auto Reframe &lt;a href="https://www.capcut.com/help/why-do-i-have-to-pay-for-auto-reframe" rel="noopener noreferrer"&gt;moved to CapCut Pro&lt;/a&gt; after an introductory period. The basic crop and zoom tools remain free.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Price:&lt;/strong&gt; requires a CapCut Pro subscription.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; you're already editing short clips in CapCut, especially on your phone.&lt;/p&gt;

&lt;h3&gt;
  
  
  5. Kapwing: Speaker Focus
&lt;/h3&gt;

&lt;p&gt;Kapwing runs in the browser. Click &lt;strong&gt;Resize Canvas&lt;/strong&gt;, pick a vertical preset (9:16, 4:5 or 3:4), and choose &lt;strong&gt;Speaker Focus&lt;/strong&gt; to &lt;a href="https://www.kapwing.com/ai/auto-speaker-focus" rel="noopener noreferrer"&gt;keep the active speaker centred&lt;/a&gt;. It's aimed at podcasts, Zoom meetings, webinars and interviews. If you don't want AI framing, &lt;strong&gt;Fill and Crop&lt;/strong&gt; and &lt;strong&gt;Fit to Center&lt;/strong&gt; (blurred or coloured padding) are right there.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Price:&lt;/strong&gt; Free to start (720p with a watermark); paid plans remove the watermark.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; you want a quick browser tool for multi-speaker recordings and don't want to install anything.&lt;/p&gt;

&lt;h3&gt;
  
  
  6. OpusClip: ReframeAnything
&lt;/h3&gt;

&lt;p&gt;OpusClip is built to find clips in long videos, and its &lt;a href="https://www.opus.pro/ai-reframe" rel="noopener noreferrer"&gt;ReframeAnything&lt;/a&gt; model reframes them, tracking moving speakers and objects and switching layouts to fit the video and platform. You can take over with manual reframing.&lt;/p&gt;

&lt;p&gt;According to its &lt;a href="https://www.opus.pro/pricing" rel="noopener noreferrer"&gt;pricing page&lt;/a&gt;, the free plan reframes to 9:16 only, adds a watermark and keeps exports for 3 days. Starter ($15/month) removes the watermark; Pro ($29/month) adds 1:1 and 16:9, moving-object tracking and export to Premiere Pro and DaVinci Resolve.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; your goal is many short clips from one long video, and reframing is a step in that pipeline.&lt;/p&gt;

&lt;h3&gt;
  
  
  7. ffmpeg: one command, no tracking
&lt;/h3&gt;

&lt;p&gt;For scripts and batch jobs, &lt;a href="https://ffmpeg.org/ffmpeg-filters.html#crop" rel="noopener noreferrer"&gt;ffmpeg's crop filter&lt;/a&gt; does a fixed 9:16 crop in one line. It doesn't track anyone, so it only suits a centred or locked-off subject, but it's free, fast and runs in CI.&lt;/p&gt;

&lt;p&gt;Centre crop to 1080×1920:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ffmpeg &lt;span class="nt"&gt;-i&lt;/span&gt; input.mp4 &lt;span class="nt"&gt;-vf&lt;/span&gt; &lt;span class="s2"&gt;"crop=trunc(ih*9/32)*2:ih,scale=1080:1920"&lt;/span&gt; &lt;span class="nt"&gt;-c&lt;/span&gt;:a copy vertical.mp4
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The &lt;code&gt;trunc(...)*2&lt;/code&gt; keeps the width even, which H.264 needs. Your subject sits left of centre? Move the window with &lt;code&gt;x&lt;/code&gt; (here, starting a quarter of the way in):&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ffmpeg &lt;span class="nt"&gt;-i&lt;/span&gt; input.mp4 &lt;span class="nt"&gt;-vf&lt;/span&gt; &lt;span class="s2"&gt;"crop=trunc(ih*9/32)*2:ih:x=iw*0.25,scale=1080:1920"&lt;/span&gt; &lt;span class="nt"&gt;-c&lt;/span&gt;:a copy vertical-left.mp4
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;And for slides or screen recordings, fit the whole frame on a blurred copy of itself instead of cropping:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ffmpeg &lt;span class="nt"&gt;-i&lt;/span&gt; input.mp4 &lt;span class="nt"&gt;-filter_complex&lt;/span&gt; &lt;span class="se"&gt;\&lt;/span&gt;
&lt;span class="s2"&gt;"[0:v]split[a][b];[a]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,boxblur=20[bg];[b]scale=1080:-2[fg];[bg][fg]overlay=(W-w)/2:(H-h)/2"&lt;/span&gt; &lt;span class="se"&gt;\&lt;/span&gt;
&lt;span class="nt"&gt;-c&lt;/span&gt;:a copy vertical-fit.mp4
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;All three produce a 1080×1920 file (tested on ffmpeg with a 1920×1080 source).&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Price:&lt;/strong&gt; free and open source.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pick it if:&lt;/strong&gt; the framing never changes and you want it automated.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which tool should you use?
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;One person, centred, never moves:&lt;/strong&gt; ffmpeg's centre crop is all you need.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Slides or a screen recording:&lt;/strong&gt; fit with a blurred background (ffmpeg's third command, Kapwing's Fit to Center), or crop to the part of the screen in use.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Two or more people talking:&lt;/strong&gt; you need speaker-following, not a crop. MirrorCut, Kapwing's Speaker Focus or OpusClip.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You already own an editor:&lt;/strong&gt; Premiere's Auto Reframe or Resolve Studio's Smart Reframe keeps everything in one project.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You want the vertical version finished, not just resized:&lt;/strong&gt; captions, cleanup and every format from one edit is where MirrorCut fits.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Whatever you use, watch the result on a phone before posting. Auto reframing is a strong first pass, and the shot it gets wrong is usually the one where both people laugh at once.&lt;/p&gt;

&lt;h2&gt;
  
  
  Bottom line
&lt;/h2&gt;

&lt;p&gt;Converting horizontal video to vertical is easy when nothing moves and hard when people talk. Keep ffmpeg for the easy cases, and for recordings with real conversations, start with &lt;a href="https://www.mirrorcut.ai/use-cases/reframe-video-to-vertical" rel="noopener noreferrer"&gt;MirrorCut&lt;/a&gt;: ask for 9:16, check each shot on the timeline, and keep or revert what it changed.&lt;/p&gt;

</description>
      <category>video</category>
      <category>ai</category>
      <category>ffmpeg</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>7 Best AI Video Editors for Developer Tutorials and Demos (2026)</title>
      <dc:creator>MirrorCut</dc:creator>
      <pubDate>Sat, 24 Oct 2026 18:51:26 +0000</pubDate>
      <link>https://dev.to/mirrorcut/7-best-ai-video-editors-for-developer-tutorials-and-demos-2026-mjc</link>
      <guid>https://dev.to/mirrorcut/7-best-ai-video-editors-for-developer-tutorials-and-demos-2026-mjc</guid>
      <description>&lt;p&gt;&lt;em&gt;Every tool below is described from its own pricing page and help docs, linked at the end. Prices are list prices checked on Oct 25, 2026. They change often and some vary by region, so click through before you pay.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Developers record more video than we admit. A PR walkthrough for a teammate. A bug repro. A five-minute "how to set up the SDK" for the docs. A 30-second launch clip for the README. A conference talk you want to cut into Shorts.&lt;/p&gt;

&lt;p&gt;Recording is the easy part: OBS, QuickTime, Loom or your OS screen recorder all work. The editing is where the evening goes. You sit through dead air while &lt;code&gt;npm install&lt;/code&gt; runs, three "ummm, let me just…" restarts, and the take where you typed the wrong password on camera.&lt;/p&gt;

&lt;p&gt;AI video editors promise to handle that boring first pass. Some do it well, some do something else entirely, and they differ a lot in one thing developers care about: &lt;strong&gt;can you see exactly what the AI changed, and undo just that part?&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Here's how seven of them compare for developer video.&lt;/p&gt;

&lt;h2&gt;
  
  
  What matters for developer video
&lt;/h2&gt;

&lt;p&gt;Before the list, here are the criteria I used. They're different from what a vlogger or a TikTok creator needs.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Cuts by what you said.&lt;/strong&gt; Tutorials are narration-led, so transcript-based editing (delete words, the video follows) saves the most time.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Handles dead air and retakes.&lt;/strong&gt; Long builds, slow page loads and "let me try that again" are most of a raw dev recording.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Caption files you can ship.&lt;/strong&gt; SRT or VTT export means the same captions can go on YouTube, in your docs player, or into a translation pass.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Multiple aspect ratios.&lt;/strong&gt; 16:9 for docs and YouTube, 9:16 for Shorts and Reels.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Reviewability.&lt;/strong&gt; An AI pass is a diff. You want to review it like one, not accept or reject the whole thing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Where it runs and where it hands off.&lt;/strong&gt; Browser or desktop, and whether you can take the timeline into Premiere, Final Cut or Resolve.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  1. Descript: the transcript-first all-rounder
&lt;/h2&gt;

&lt;p&gt;Descript made "edit video like a doc" mainstream, and it's still the default many people reach for.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;What the AI does:&lt;/strong&gt; Underlord, its chat co-editor (labelled beta), handles Remove Filler Words, Remove Retakes, Edit for Clarity, Studio Sound, captions, clips, translation and automatic multicam.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; Underlord creates a checkpoint before its changes, and you can revert a specific response, alongside normal undo and version history. Per Descript's help docs, those checkpoints only last for the current session, and Descript notes Underlord "might overpromise, make incorrect assumptions".&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hand-off:&lt;/strong&gt; timeline export to Premiere, Final Cut Pro and others.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Free (60 media minutes a month, 720p export); Hobbyist $24/month or $16 billed annually; Creator $35 or $24 annually.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; developers who want one mature tool for screencasts, podcasts and clips and don't mind a bigger app.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  2. Gling: fast rough cuts for talking-head YouTubers
&lt;/h2&gt;

&lt;p&gt;Gling is built for YouTubers who talk to camera. It's a desktop app.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;What the AI does:&lt;/strong&gt; removes bad takes, silences and filler words. It also does captions, auto framing, multicam, AI B-roll, and titles and chapters.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; "Show cuts" reveals what was removed and lets you Uncut it, with pace and padding controls. It's per-cut review for removals, which is genuinely useful.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hand-off:&lt;/strong&gt; XML export to Premiere, Final Cut Pro and Resolve.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Free (1 hour of AI-edited media a month, watermarked); Plus $20/month or $10 billed annually; Pro $40 or $20 annually.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; dev YouTubers who finish in an NLE and want a clean rough cut first.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  3. Opus Clip: long talk in, short clips out
&lt;/h2&gt;

&lt;p&gt;Opus Clip isn't a general editor. It finds clips in a long video.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;What the AI does:&lt;/strong&gt; picks clips with a "virality score", adds hooks and captions, and reframes to vertical.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; mostly one-shot. You get clips and edit them afterwards; text and timeline editing are on paid plans only.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hand-off:&lt;/strong&gt; Premiere and DaVinci Resolve export on Pro.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Free (watermarked exports, no editing); Starter $15/month; Pro $29/month.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; turning a recorded conference talk or livestream into Shorts quickly.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  4. CapCut: free, powerful, social-first
&lt;/h2&gt;

&lt;p&gt;CapCut has a huge free tier and a full timeline editor.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;What the AI does:&lt;/strong&gt; Auto Cut / AI Edit restructures footage from a prompt. It also has auto reframe, captions, background removal and generation billed in credits.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; CapCut's help pages describe AI Edit as producing a draft you refine on the timeline. I couldn't find docs for reviewing it change by change.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Free plan, plus a paid Pro plan. CapCut's pricing page says displayed prices are for reference only and vary by account and region, so sign in to see yours.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; short, punchy dev Shorts when budget is zero.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  5. Riverside: record and edit in one place
&lt;/h2&gt;

&lt;p&gt;Riverside is a recording studio first, and its editor is built around that.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;What the AI does:&lt;/strong&gt; Magic Clips, a chat agent called Co-Creator, and silence and filler-word removal.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; clips open in a transcript-plus-timeline editor for manual changes.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Free (2 hours of multi-track recording, 720p, watermarked); Pro $29/month or $24 billed annually.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; dev podcasts, interviews and live-coding sessions with guests.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  6. VEED: browser editor with lots of AI tools
&lt;/h2&gt;

&lt;p&gt;VEED runs in the browser and bundles subtitles, text-based editing, clips, avatars and generation, with AI generation metered in credits.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; a browser timeline. I didn't find docs on how granular AI review is.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; paid plans are charged per seat and shown in your local currency, so check VEED's pricing page for your region.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; teams that want a shared browser editor for quick social edits.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  7. MirrorCut: an agent that shows every edit as a diff (my tool)
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;MirrorCut&lt;/a&gt; is an AI agent that edits your recordings into videos, in the browser. You upload a recording and the agent makes the first cut: it &lt;a href="https://www.mirrorcut.ai/use-cases/remove-dead-air" rel="noopener noreferrer"&gt;removes awkward pauses&lt;/a&gt;, retakes and filler words.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Review and undo:&lt;/strong&gt; this is the part I built it for. Every change the agent makes shows up on the timeline with its own Accept and Revert, so you keep the cuts you agree with and undo the ones you don't.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hand-off:&lt;/strong&gt; there's no NLE project export today, so if you finish in Premiere or Resolve, Descript, Gling or Opus Clip Pro fit that workflow better right now.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Price:&lt;/strong&gt; Creator $10/month, Pro $30, Studio $60 (&lt;a href="https://www.mirrorcut.ai/pricing" rel="noopener noreferrer"&gt;pricing&lt;/a&gt;). Every plan has every feature; plans differ by how much you can make each month. There's a 7-day free trial of Creator, and it needs a card.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Best for:&lt;/strong&gt; narrated recordings where most of the cleanup is pauses, retakes and filler, and you want to review the AI's first pass like a pull request. If your tutorials are screen-only with a voice-over, try it on one recording before you commit.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Side-by-side
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Tool&lt;/th&gt;
&lt;th&gt;Runs on&lt;/th&gt;
&lt;th&gt;AI first pass&lt;/th&gt;
&lt;th&gt;How you undo AI work&lt;/th&gt;
&lt;th&gt;NLE hand-off&lt;/th&gt;
&lt;th&gt;Entry paid price*&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Descript&lt;/td&gt;
&lt;td&gt;App&lt;/td&gt;
&lt;td&gt;Filler, retakes, clarity, clips&lt;/td&gt;
&lt;td&gt;Revert per chat response (session only)&lt;/td&gt;
&lt;td&gt;Yes&lt;/td&gt;
&lt;td&gt;$24/mo ($16 annual)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Gling&lt;/td&gt;
&lt;td&gt;Desktop&lt;/td&gt;
&lt;td&gt;Bad takes, silence, filler&lt;/td&gt;
&lt;td&gt;Per-cut "Uncut" for removals&lt;/td&gt;
&lt;td&gt;Yes (XML)&lt;/td&gt;
&lt;td&gt;$20/mo ($10 annual)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Opus Clip&lt;/td&gt;
&lt;td&gt;Browser&lt;/td&gt;
&lt;td&gt;Clips + virality score&lt;/td&gt;
&lt;td&gt;Edit the clips after&lt;/td&gt;
&lt;td&gt;Pro plan&lt;/td&gt;
&lt;td&gt;$15/mo&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;CapCut&lt;/td&gt;
&lt;td&gt;App + web&lt;/td&gt;
&lt;td&gt;AI Edit draft&lt;/td&gt;
&lt;td&gt;Refine the draft on the timeline&lt;/td&gt;
&lt;td&gt;Not checked&lt;/td&gt;
&lt;td&gt;Varies by region&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Riverside&lt;/td&gt;
&lt;td&gt;Browser&lt;/td&gt;
&lt;td&gt;Silence/filler, Magic Clips&lt;/td&gt;
&lt;td&gt;Edit in transcript + timeline&lt;/td&gt;
&lt;td&gt;Not checked&lt;/td&gt;
&lt;td&gt;$29/mo ($24 annual)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;VEED&lt;/td&gt;
&lt;td&gt;Browser&lt;/td&gt;
&lt;td&gt;Subtitles, clips, generation&lt;/td&gt;
&lt;td&gt;Timeline (granularity not documented)&lt;/td&gt;
&lt;td&gt;Not checked&lt;/td&gt;
&lt;td&gt;Per seat, varies by region&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;MirrorCut&lt;/td&gt;
&lt;td&gt;Browser&lt;/td&gt;
&lt;td&gt;Awkward pauses, retakes, filler&lt;/td&gt;
&lt;td&gt;Accept/Revert per change&lt;/td&gt;
&lt;td&gt;Not yet&lt;/td&gt;
&lt;td&gt;$10/mo&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;*List prices checked Oct 25, 2026, from each vendor's pricing page. They may differ in your region. "Not checked" means I didn't verify it, not that the feature is missing.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which one should you pick?
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;You finish in Premiere or Resolve:&lt;/strong&gt; Gling or Descript, because both hand off to your NLE today.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You have a long talk and want Shorts:&lt;/strong&gt; Opus Clip, or any tool here with clip-finding.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Zero budget, mostly vertical:&lt;/strong&gt; CapCut.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You record podcasts or interviews with guests:&lt;/strong&gt; Riverside.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You want to review every AI change before it lands, and you finish in the browser:&lt;/strong&gt; MirrorCut. That's the bet I'm making, and I'd rather you try it on a real recording than take my word for it.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  A workflow that works with any of them
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Record big.&lt;/strong&gt; Bump your editor and terminal font to 18–20 pt and hide notifications. Readable code beats a pretty cut.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Say retakes out loud.&lt;/strong&gt; "Again from the import line" gives a transcript-based editor an easy marker to find.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Let the AI cut dead air and filler first.&lt;/strong&gt; That's the safe, boring bulk of the work.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review the cuts that touch code.&lt;/strong&gt; An AI might cut a pause that was really you letting a command's output land. Keep those.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Export captions as SRT or VTT&lt;/strong&gt; and reuse them for docs, YouTube and translation.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Cut Shorts last,&lt;/strong&gt; from the finished long version, so the clips inherit the clean edit.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you want the MirrorCut side of this in more depth, there's a &lt;a href="https://www.mirrorcut.ai/use-cases/talking-head-video-editor" rel="noopener noreferrer"&gt;talking-head editing guide&lt;/a&gt; and a broader &lt;a href="https://www.mirrorcut.ai/best/ai-video-editors" rel="noopener noreferrer"&gt;AI video editor comparison&lt;/a&gt; on our site.&lt;/p&gt;

&lt;p&gt;What's in your dev video stack? I'm especially curious how you handle code that's too small to read on a phone.&lt;/p&gt;




&lt;h3&gt;
  
  
  Sources (pricing pages rechecked Oct 25, 2026)
&lt;/h3&gt;

&lt;ul&gt;
&lt;li&gt;Descript pricing: &lt;a href="https://www.descript.com/pricing" rel="noopener noreferrer"&gt;https://www.descript.com/pricing&lt;/a&gt; · Underlord revert: &lt;a href="https://help.descript.com/ai-assistant/revert" rel="noopener noreferrer"&gt;https://help.descript.com/ai-assistant/revert&lt;/a&gt; · Underlord overview: &lt;a href="https://help.descript.com/getting-started/underlord-beta-your-ai-co-editor-in-descript" rel="noopener noreferrer"&gt;https://help.descript.com/getting-started/underlord-beta-your-ai-co-editor-in-descript&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Gling pricing: &lt;a href="https://www.gling.ai/pricing" rel="noopener noreferrer"&gt;https://www.gling.ai/pricing&lt;/a&gt; · Show cuts/Uncut: &lt;a href="https://www.gling.ai/help/manual-cuts-and-trimming" rel="noopener noreferrer"&gt;https://www.gling.ai/help/manual-cuts-and-trimming&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Opus Clip pricing: &lt;a href="https://www.opus.pro/pricing" rel="noopener noreferrer"&gt;https://www.opus.pro/pricing&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;CapCut pricing: &lt;a href="https://www.capcut.com/pricing" rel="noopener noreferrer"&gt;https://www.capcut.com/pricing&lt;/a&gt; · AI Edit/Auto Cut: &lt;a href="https://www.capcut.com/help/how-to-use-auto-cut" rel="noopener noreferrer"&gt;https://www.capcut.com/help/how-to-use-auto-cut&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Riverside pricing: &lt;a href="https://riverside.com/pricing" rel="noopener noreferrer"&gt;https://riverside.com/pricing&lt;/a&gt; · AI features: &lt;a href="https://riverside.com/ai" rel="noopener noreferrer"&gt;https://riverside.com/ai&lt;/a&gt; · Co-Creator: &lt;a href="https://riverside.com/co-creator" rel="noopener noreferrer"&gt;https://riverside.com/co-creator&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;VEED pricing: &lt;a href="https://www.veed.io/pricing" rel="noopener noreferrer"&gt;https://www.veed.io/pricing&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;MirrorCut: &lt;a href="https://www.mirrorcut.ai" rel="noopener noreferrer"&gt;https://www.mirrorcut.ai&lt;/a&gt; · &lt;a href="https://www.mirrorcut.ai/pricing" rel="noopener noreferrer"&gt;https://www.mirrorcut.ai/pricing&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>ai</category>
      <category>productivity</category>
      <category>tooling</category>
      <category>video</category>
    </item>
  </channel>
</rss>
