<?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: JME Freestyle AI</title>
    <description>The latest articles on DEV Community by JME Freestyle AI (@jmeukrap).</description>
    <link>https://dev.to/jmeukrap</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%2F4244136%2F13d759ff-9e92-4485-8631-6bdb8923b99a.png</url>
      <title>DEV Community: JME Freestyle AI</title>
      <link>https://dev.to/jmeukrap</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/jmeukrap"/>
    <language>en</language>
    <item>
      <title>FaceScored: an AI face score you can audit, computed entirely in your browser</title>
      <dc:creator>JME Freestyle AI</dc:creator>
      <pubDate>Tue, 17 Nov 2026 08:18:20 +0000</pubDate>
      <link>https://dev.to/jmeukrap/facescored-an-ai-face-score-you-can-audit-computed-entirely-in-your-browser-55j</link>
      <guid>https://dev.to/jmeukrap/facescored-an-ai-face-score-you-can-audit-computed-entirely-in-your-browser-55j</guid>
      <description>&lt;p&gt;FaceScored is a free AI face rating tool that performs the whole analysis inside your browser tab.&lt;/p&gt;

&lt;h2&gt;
  
  
  What happens to the photo
&lt;/h2&gt;

&lt;p&gt;Nothing is uploaded. The app reads the photo locally, runs MediaPipe face landmark detection, measures facial symmetry and proportion ratios, then renders a score next to the formula that produced it. No account, no server-side copy of the image.&lt;/p&gt;

&lt;h2&gt;
  
  
  How the symmetry number is built
&lt;/h2&gt;

&lt;p&gt;The symmetry test averages the horizontal imbalance across six landmark pairs, normalises it by face width, and applies the same formula the homepage rating uses. Two of the pairs are the inner and outer eye corners, measured against the vertical reference line through the face. The result also surfaces the two largest measured deviations, so the score can be read against the measurements behind it.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Face rater: &lt;a href="https://facescored.com" rel="noopener noreferrer"&gt;https://facescored.com&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Symmetry breakdown: &lt;a href="https://facescored.com/face-symmetry/" rel="noopener noreferrer"&gt;https://facescored.com/face-symmetry/&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why two photos of the same face score differently
&lt;/h2&gt;

&lt;p&gt;The calculation works on landmark positions in a 2D image, so camera perspective, head rotation, expression and landmark-detection noise all move those positions. The site states this itself and does not present the output as a medical assessment.&lt;/p&gt;

&lt;h2&gt;
  
  
  Free vs paid
&lt;/h2&gt;

&lt;p&gt;Free: overall symmetry score, left and right mirrored comparisons, deviation list, browser-local processing.&lt;/p&gt;

&lt;p&gt;Optional full report (currently listed at $16.99, one-time): six dimensions - symmetry, proportions, jawline, eye-angle balance, face shape, skin clarity - written feature notes, heuristic percentile estimates and a printable PDF. Percentiles are model heuristics, not measured population rankings.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>frontend</category>
      <category>privacy</category>
      <category>showdev</category>
    </item>
    <item>
      <title>JME Freestyle AI: a template-first pipeline for AI rap videos</title>
      <dc:creator>JME Freestyle AI</dc:creator>
      <pubDate>Tue, 17 Nov 2026 01:55:04 +0000</pubDate>
      <link>https://dev.to/jmeukrap/jme-freestyle-ai-a-template-first-pipeline-for-ai-rap-videos-2khe</link>
      <guid>https://dev.to/jmeukrap/jme-freestyle-ai-a-template-first-pipeline-for-ai-rap-videos-2khe</guid>
      <description>&lt;p&gt;JME Freestyle AI is a browser-based tool for making rap-style AI videos from ready-made performance templates. You pick a template, describe the scene you want, and the app renders a 720p clip — no editing software needed.&lt;/p&gt;

&lt;p&gt;What is interesting about this approach is that the template already encodes the camera moves and pacing, so the user mainly writes the prompt and chooses a pass.&lt;/p&gt;

&lt;p&gt;Useful links:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Website: &lt;a href="https://jmeukrap.com/" rel="noopener noreferrer"&gt;https://jmeukrap.com/&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Pricing: &lt;a href="https://jmeukrap.com/pricing" rel="noopener noreferrer"&gt;https://jmeukrap.com/pricing&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;One-time passes: &lt;a href="https://jmeukrap.com/one-time" rel="noopener noreferrer"&gt;https://jmeukrap.com/one-time&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;It is aimed at short-form creators who want a rap video clip in a few minutes rather than an afternoon in a timeline.&lt;/p&gt;

</description>
    </item>
    <item>
      <title>JME Freestyle AI: template-first AI rap video workflow</title>
      <dc:creator>JME Freestyle AI</dc:creator>
      <pubDate>Mon, 16 Nov 2026 02:11:58 +0000</pubDate>
      <link>https://dev.to/jmeukrap/jme-freestyle-ai-template-first-ai-rap-video-workflow-18b</link>
      <guid>https://dev.to/jmeukrap/jme-freestyle-ai-template-first-ai-rap-video-workflow-18b</guid>
      <description>&lt;p&gt;JME Freestyle AI is a browser-based tool for making rap-style AI videos from ready-made performance templates. You pick a template, describe the scene you want, and the app renders a 720p clip — no editing software needed.&lt;/p&gt;

&lt;p&gt;What I find interesting about this approach is that it removes the usual pipeline of shooting, cutting, and syncing. The template already encodes the camera moves and pacing, so you mainly write the prompt and choose a pass.&lt;/p&gt;

&lt;p&gt;Useful links:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Website: &lt;a href="https://jmeukrap.com/" rel="noopener noreferrer"&gt;https://jmeukrap.com/&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Pricing: &lt;a href="https://jmeukrap.com/pricing" rel="noopener noreferrer"&gt;https://jmeukrap.com/pricing&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;One-time passes: &lt;a href="https://jmeukrap.com/one-time" rel="noopener noreferrer"&gt;https://jmeukrap.com/one-time&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;It is aimed at short-form creators who want a rap video clip in a few minutes rather than an afternoon in a timeline.&lt;/p&gt;

</description>
      <category>aivideotools</category>
    </item>
    <item>
      <title>One-time passes vs subscriptions for short AI video clips</title>
      <dc:creator>JME Freestyle AI</dc:creator>
      <pubDate>Sun, 15 Nov 2026 09:31:50 +0000</pubDate>
      <link>https://dev.to/jmeukrap/one-time-passes-vs-subscriptions-for-short-ai-video-clips-4lin</link>
      <guid>https://dev.to/jmeukrap/one-time-passes-vs-subscriptions-for-short-ai-video-clips-4lin</guid>
      <description>&lt;p&gt;Short vertical clips rarely need a monthly plan. JME Freestyle AI takes the one-time route: a pass covers a single 15-second rap video at 720p, rendered from one uploaded photo, and nothing renews.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;a href="https://jmeukrap.com/one-time" rel="noopener noreferrer"&gt;One-time passes&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://jmeukrap.com/pricing" rel="noopener noreferrer"&gt;Pricing&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://jmeukrap.com/" rel="noopener noreferrer"&gt;Product site&lt;/a&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Templates fix camera moves and rhyme placement, so you are not writing prompts. For people who render often, credit packs of 300 / 1000 / 1800 are available instead. Generation runs on Wan 3.0 through an API partner.&lt;/p&gt;

&lt;p&gt;Feedback: &lt;a href="mailto:support@jmeukrap.com"&gt;support@jmeukrap.com&lt;/a&gt;&lt;/p&gt;

</description>
    </item>
    <item>
      <title>JME Freestyle AI: template-first AI rap video notes</title>
      <dc:creator>JME Freestyle AI</dc:creator>
      <pubDate>Sun, 15 Nov 2026 09:14:08 +0000</pubDate>
      <link>https://dev.to/jmeukrap/jme-freestyle-ai-template-first-ai-rap-video-notes-4o4c</link>
      <guid>https://dev.to/jmeukrap/jme-freestyle-ai-template-first-ai-rap-video-notes-4o4c</guid>
      <description>&lt;p&gt;JME Freestyle AI turns one uploaded photo into a 15-second vertical rap video at 720p. Scene templates fix camera moves and rhyme placement instead of asking the user to write a prompt, so the first render is usually usable.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;a href="https://jmeukrap.com/" rel="noopener noreferrer"&gt;Product site&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://jmeukrap.com/pricing" rel="noopener noreferrer"&gt;Pricing&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://jmeukrap.com/one-time" rel="noopener noreferrer"&gt;One-time passes&lt;/a&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Generation runs on Wan 3.0 through an API partner. Credits are sold in 300 / 1000 / 1800 packs, and one-time passes cover a single clip without a subscription.&lt;/p&gt;

&lt;p&gt;Feedback and template requests: &lt;a href="mailto:support@jmeukrap.com"&gt;support@jmeukrap.com&lt;/a&gt;&lt;/p&gt;

</description>
    </item>
    <item>
      <title>Turning one photo into a 15-second vertical rap video: notes on a template-first AI video pipeline</title>
      <dc:creator>JME Freestyle AI</dc:creator>
      <pubDate>Sat, 14 Nov 2026 08:20:57 +0000</pubDate>
      <link>https://dev.to/jmeukrap/turning-one-photo-into-a-15-second-vertical-rap-video-notes-on-a-template-first-ai-video-pipeline-1bo6</link>
      <guid>https://dev.to/jmeukrap/turning-one-photo-into-a-15-second-vertical-rap-video-notes-on-a-template-first-ai-video-pipeline-1bo6</guid>
      <description>&lt;p&gt;We run &lt;a href="https://jmeukrap.com/" rel="noopener noreferrer"&gt;JME Freestyle AI&lt;/a&gt;, a small tool that takes a single uploaded photo and returns a 15-second vertical rap video. The interesting part is not the model - it is that we ask the user to choose a &lt;strong&gt;template&lt;/strong&gt; instead of writing a prompt.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why templates instead of prompts
&lt;/h2&gt;

&lt;p&gt;Prompt-based video generation pushes the failure mode onto the least prepared person in the loop. If someone has never written a shot list, "cinematic dolly-in, 35mm, moody lighting" is a coin flip. A template encodes the shot list once, and the user only supplies what they actually have: one photo.&lt;/p&gt;

&lt;p&gt;Each template fixes three things:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;the music style and tempo, so the cut points are known in advance&lt;/li&gt;
&lt;li&gt;the camera choreography (push in, whip pan, static), so the motion is plausible for the source image&lt;/li&gt;
&lt;li&gt;the on-screen text structure, so the lyrics land on beats rather than drifting&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That converts an open-ended generation task into a bounded one. Bounded tasks are cheaper to render and far easier to evaluate.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rendering policy that actually matters: shortest duration
&lt;/h2&gt;

&lt;p&gt;We render at 720p vertical and apply a shortest-duration policy across every pass, including one-time passes. In practice: if the template says 15 seconds and the useful content finishes sooner, we trim. Nobody pays for dead air, and the clip is immediately usable on TikTok, Reels and Shorts without a re-export.&lt;/p&gt;

&lt;p&gt;Trimming also hides a class of model artifacts. Long generations accumulate drift - faces soften, backgrounds swim. Cutting at the last coherent frame is a cheaper fix than a quality filter.&lt;/p&gt;

&lt;h2&gt;
  
  
  Model layer
&lt;/h2&gt;

&lt;p&gt;Generation runs on Wan 3.0 through an API partner. We deliberately do not train a custom face model: the product promise is "your photo, one minute, no setup", and a hosted image-to-video model plus strong templates gets us there. The trade-off is that we inherit the provider's failure modes, so the template library is where our own quality control lives.&lt;/p&gt;

&lt;h2&gt;
  
  
  Safety and rights
&lt;/h2&gt;

&lt;p&gt;Two rules are non-negotiable and both are product decisions, not legal boilerplate:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;You upload images you are allowed to use. We do not scrape faces.&lt;/li&gt;
&lt;li&gt;Safety checks run before generation, not after. A blocked render costs the user nothing.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The templates are original arrangements. We are not affiliated with any artist or label, and nothing in the product claims otherwise.&lt;/p&gt;

&lt;h2&gt;
  
  
  What we would tell another team building this
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Constrain the input. One photo is a better contract than "describe your video".&lt;/li&gt;
&lt;li&gt;Put the taste in the template, not in the prompt.&lt;/li&gt;
&lt;li&gt;Bill by the finished clip, not by the GPU minute.&lt;/li&gt;
&lt;li&gt;Ship the trim logic before the style transfer.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you want to see the template list, the tool is live at &lt;a href="https://jmeukrap.com/" rel="noopener noreferrer"&gt;jmeukrap.com&lt;/a&gt;, and credit packs and one-time passes are on the &lt;a href="https://jmeukrap.com/pricing" rel="noopener noreferrer"&gt;pricing page&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>startup</category>
      <category>tooling</category>
    </item>
  </channel>
</rss>
