Rebuild text, shapes, and animation with Codex + ChatGPT.
Compare the result. Export a video. Keep the project and make it yours.
Try the online studio ↗ · Run on Windows · Full walkthrough
English · 简体中文
Actual saved reconstruction. Visual differences remain.
Watch the examples with playback controls · View a still
Paste a public video link or upload a clip. MotionClone reconstructs it as an editable HyperFrames project: text, shapes, artwork, and animation you can adjust for a product demo, title sequence, or launch video.
| Bring a reference | Check the rebuild | Make it yours |
|---|---|---|
| Import from X, YouTube, Vimeo, a direct video URL, or a local file. | Play both videos together, scrub, and slow down to inspect the match. | Download an MP4 or edit the project’s copy, assets, styling, and timing. |
Bring your own AI account. Connect your own ChatGPT/Codex account to generate videos. Downloading this repository does not include AI access, credits, or unlimited generation. Your account's model access, costs, and usage limits apply. Reconstruction is approximate: small text, unfamiliar fonts, photographs, and complex 3D can need more work.
Windows prerequisites: Git, Python 3.11+, Node.js 22+ with npm, FFmpeg + ffprobe, Google Chrome, and the official Codex CLI with ChatGPT account access. Make git, python, npm, ffmpeg, ffprobe, and codex available on PATH.
git clone https://github.com/blixvip/MotionClone.git
cd MotionClone
codex login
.\start.ps1The launcher installs locked dependencies and opens http://127.0.0.1:4319. First launch needs an internet connection. Next time, double-click Start MotionClone.vbs.
No API key required. MotionClone uses your Codex login. To choose another model available to your account, set FRAMEFORGE_MODEL before starting the app.
Prefer the browser? The online studio is in early access and requires sign-in, a separate ChatGPT connection, and an available processing worker.
Open Settings → Connect ChatGPT, complete the official OpenAI sign-in, and return to MotionClone. Alternatively, run codex login before launching. Signing into the hosted website does not connect the AI account automatically. Current generation supports ChatGPT through Codex; other agent providers are not integrated. Provider access and usage details.
Open Editor from the navigation. Upload references or use saved ones, set each segment's in/out points, and add up to 12 segments (120 seconds total). Reorder or preview the sequence, add segment notes and a project brief, and choose landscape, portrait, or square. Save sequence stores your draft in your workspace. Generate my video uses your connected AI account to rebuild the assembled reference as one editable video. Follow its progress in the studio and download the MP4 or editable project when complete.
Reference imports and sequence preparation do not use AI; generation does. Original audio is retained when selected. See Editor.
01 / Add a reference → 02 / Rebuild → 03 / Compare & export
Start with a short clip. Choose whether to retain its soundtrack in Audio settings, then select Rebuild video. Completed scenes save as the job progresses. Local inputs support up to 120 seconds and 250 MB.
Choose Studio, Editorial, Signal, Cobalt, Peach, or Monochrome. Set the export format and arrangement independently.
| Landscape | Portrait | Square | Ultrawide | Feed |
|---|---|---|---|---|
| 1920 × 1080 | 1080 × 1920 | 1080 × 1080 | 3840 × 1080 | 1080 × 1350 |
Present the original and rebuild side by side, stacked, with the rebuild in the spotlight, or with a wipe reveal. Use Saved videos to reopen projects, mark favorites, or organize collections.
See the full visual walkthrough →
| Download | What you get |
|---|---|
| Compare & export → Download MP4 | Both videos in the selected style, format, and arrangement, with retained original audio. |
| Export details → Rebuilt video only | The reconstructed animation on its own. |
| Editable project | HyperFrames code, assets, fonts, and audio for further editing. |
Extract the editable ZIP, read its README, and run:
npm install
npm run previewEdit the project, then run npm run render. A coding agent can help change the copy, colors, or timing; the agent workflow explains how. The editable ZIP excludes the original reference video and reference screenshots. See recording and exports for details.
| Start creating | Go deeper |
|---|---|
| Your first project | Edit with a coding agent |
| Product demo workflow | Recording and exports |
| Launch video workflow | Development and verification |
| 简体中文指南 | Report a bug |
Troubleshooting
- A command is missing: install its prerequisite, add it to PATH, and reopen PowerShell.
- ChatGPT is disconnected: run
codex login, then refresh the connection in the app. - A video link fails: try uploading the file. Private or unavailable links may not download.
- The server will not start: check
server-error.login the project folder. - The result looks wrong: inspect the comparison. Tiny text, missing fonts, photography, and complex motion can differ from the reference.
Local data, account usage & licensing
Projects and media live in data/ on your computer. Selected reference frames are sent to ChatGPT for analysis, and video downloads contact the source service. The app is not offline-only. Hosted storage differs; check the online studio’s account and privacy information.
Use footage you own or have permission to adapt. Example projects and reference media are not bundled with a fresh clone. Export completion confirms that a file was produced; use the comparison to judge visual similarity.
No project-wide open-source license has been assigned. Third-party components retain their own licenses; see THIRD_PARTY.md and licenses/.
For development, run .venv\Scripts\python.exe -m pytest -q after setup. See the development guide for browser checks.
Built by blix · Follow on X · motionclone.lol
