Generate first. Edit the motion after.
Framlit is designed around the practical moment after generation: selecting layers, moving them, fixing placement, recording motion, and keeping the output inspectable.
Generate a Remotion video
Start from a prompt, product image, reference video, changelog, or storyboard. Framlit creates inspectable Remotion code first.
Edit layers directly
Every new video ships with a layer manifest for visible text, images, and shapes. Older projects can still be converted, with a snapshot saved first.
Move layers and record motion
Click visible text, images, and shapes in the preview. Reposition them or drag through time to save motion tracks as Remotion transforms.
Also in the editor
Pick the model
Generate with Claude Sonnet, Claude Opus for harder briefs, or Gemini. The editor shows the credit cost of each before you send.
Sonnet 1 credit, Opus 2 credits, Gemini 1 credit per generation
Brand DNA
Save your colors, fonts, tone, and things to avoid once, or pull them from your website. New generations start from your brand instead of a generic palette.
Applied to generations on Editor Pro and Studio
AI product images
No product photo yet? Turn on AI image and Framlit generates one for the ad instead of drawing a placeholder in CSS.
3 credits per image (FLUX); GPT Image 2 through the CLI
Narration
Add a voiceover to a finished video. Framlit writes the script, records it, fits it to the video, and adds captions that highlight each word as it is spoken.
Editor Pro and Studio, 1 credit per voiceover
Agents, batches, and campaigns live in the CLI
Coding agents can design a video as a Framlit Scene (JSON) and turn it into an editable project with the CLI (npx framlit) or its MCP server, then read back the edits people make. Batch rendering from product rows and multi-segment campaign runs work the same way. Open any resulting project in Motion Editor to adjust layers and motion.
Outputs
- Editable layer manifests for visible text, images, and shapes
- Preview overlays for selecting, moving, and resizing layers
- Motion tracks recorded from drag gestures
- Readable Remotion code that can be polished by humans or agents
Boundaries
Motion Editor does not promise black-box text-to-perfect-video, automatic performance prediction, or a full NLE replacement. The current release focuses on the bridge between generated Remotion code and practical motion polish.