Remotion Production Workflow

This skill teaches how to produce complete videos with Remotion by orchestrating multiple MCP tools together. It covers the full pipeline from concept to rendered MP4.

Available MCP Tools

You have access to these MCP servers for media production:

remotion-media (via KIE)

• generate_tts — Text-to-speech voiceovers (ElevenLabs TTS)
• generate_music — Background music (Suno V3.5–V5)
• generate_sfx — Sound effects (ElevenLabs SFX V2)
• generate_image — AI images (Nano Banana Pro)
• generate_video — AI video clips (Veo 3.1)
• generate_subtitles — Transcribe audio/video to SRT (Whisper)
• list_assets — List all generated media in the project

TwelveLabs (video understanding)

• Index and analyze video files
• Semantic search within videos ("find the part where...")
• Scene detection, object detection, speaker identification
• Video summarization

Pexels (stock footage)

• searchPhotos — Search free stock photos
• searchVideos — Search free stock videos
• getVideo / getPhoto — Get details by ID
• downloadVideo — Download video to project

ElevenLabs (optional — advanced voice)

• Voice cloning from audio samples
• Advanced TTS with custom voices
• Audio isolation and processing
• Transcription

Replicate (optional — 100+ AI models)

• replicate_run — Run a model synchronously (images)
• replicate_create_prediction — Start async prediction (video)
• replicate_get_prediction — Poll prediction status
• Image models: FLUX 1.1 Pro, Imagen 4, Ideogram v3, FLUX Kontext
• Video models: Wan 2.5 (T2V, I2V), Kling 2.6 Pro

Production Pipeline

Read individual rule files for detailed workflows:

• rules/production-pipeline.md — End-to-end workflow from concept to final render
• rules/audio-integration.md — How to integrate generated audio into Remotion compositions
• rules/voiceover-sync.md — Syncing TTS voiceovers with animations and captions
• rules/music-scoring.md — Generating and timing background music
• rules/stock-footage-workflow.md — Searching, downloading, and using stock footage in Remotion
• rules/video-analysis.md — Using TwelveLabs to analyze and select clips from existing footage
• rules/captions-workflow.md — TikTok-style animated captions using @remotion/captions and Whisper
• rules/animation-presets.md — Reusable animation patterns (fade, slide, scale, typewriter, stagger)
• rules/3d-content.md — Three.js and React Three Fiber via @remotion/three
• rules/data-visualization.md — Animated charts, dashboards, and number counters
• rules/visual-effects.md — Light leaks, Lottie, film grain, vignettes, Ken Burns
• rules/ci-rendering.md — GitHub Actions workflows for automated video rendering
• rules/replicate-models.md — Replicate MCP model catalog, usage, and decision guide
• rules/image-generation.md — AI image prompt engineering, provider selection, Remotion integration
• rules/video-generation.md — AI video clip generation, I2V pipeline, sequencing in Remotion
• rules/sound-effects.md — SFX generation, prompt engineering, timing to visual events
• rules/elevenlabs-advanced.md — Voice cloning, custom TTS parameters, multi-voice scripts
• rules/asset-management.md — File organization, naming conventions, staticFile() reference

Key Principles

1. Audio drives timing — Generate voiceover first, get its duration, then set composition length to match.
2. Assets go in public/ — All generated media files (audio, video, images) must be saved to the project's public/ directory so Remotion can access them via staticFile().
3. Use Remotion's audio components — Always use <Audio> component with staticFile() for audio. Never use HTML <audio> tags.
4. Frame-based timing — Remotion uses frames, not seconds. Convert with fps * seconds. At 30fps, 1 second = 30 frames.
5. Progressive composition — Build the video in layers: visuals first, then voiceover, then music, then SFX.
6. Preview frequently — Use npm run dev to preview after each major change. The Remotion player updates live.