Produce Remotion clips

This is the rendering adapter in the content-production collection, alongside screenstudio-edit. Keep tool-specific acquisition, timing, layout and exports here. Use content-repurpose for selecting complete source ideas, video-edit for editorial judgment, social-post for platform copy, and content-publish for delivery checks or authorized publication. Read sibling skills only when their phase is needed; if a dependency is not installed, follow the handoff below and report the missing capability instead of assuming it exists.

Default for every clip request

Apply this collection's approved clip preset automatically whenever the creator asks for clips, reels or shorts. A request such as "create 5 clips from this YouTube video" is sufficient: download, transcribe, select five complete ideas, source relevant real images and generate conceptual images where useful, render and verify the finished clips in the saved style. Do not ask which style to use or require the creator to mention a previous project, "my usual style", Remotion or this skill.

The visual style is a permanent creator preference, independent of source origin, speaker, subject and language. It applies to YouTube interviews, other hosted videos, local recordings, podcasts and screen recordings. Adapt source acquisition, transcription language, speaker crops and illustrative content to the new recording while retaining the preset. Source-specific interview names, timestamps, images and personal claims are not defaults.

Override the visual treatment only when the current request or an explicit project preference calls for another style. If no local CONTENT_STYLE.md exists, use the bundled preset directly; its absence is not a reason to ask again or choose a generic style. When the recording has no speaker camera, use the relevant source screen or footage in the camera region, keeping the same typography, timing, image treatment and audio defaults. Do not manufacture a speaker portrait.

Style and resources

The default is already approved for Anthony. Use the approved creator profile when installed and the bundled style preset. A project CONTENT_STYLE.md and current instructions override those defaults. For another creator, adapt the preset to their preferences. Social-copy lowercase does not change source quotations or subtitle spelling.

The usual treatment is 1080×1920 at 30 fps, full-bleed speaker camera, hard cuts with modest punch-ins, measured pause removal and pitch-preserving 1.15× speech. Adjust speed and silence thresholds for the actual voice. Keep completed ideas, necessary caveats and ending words.

Space Grotesk captions use 64 px, weight 620, white on translucent black, centered 320 px above the bottom with 65 px side insets. Highlight the active word in the clip accent. Hold related images through their explanation, using both full-screen and camera-top/image-bottom layouts. Fill every region edge to edge with media. Do not add backgrounds, cards, borders, padding, headings, name labels or source credits over cutaways. Keep source attribution in the production ledger. Subtitles and the one-frame preview title remain part of the preset. The accepted examples hold images for several seconds; do not repeat sub-second flashes. At frame zero, show a camera/image split with the title for exactly one frame, export a matching PNG, then resume without shifting audio. Mix a classic short opening swipe below the voice.