Your idea, brought to life.
Write a scene. Animate your little character. Add a voice and create a short you can keep.
What happens in your video?
1. Character clothes
The selected outfit is applied to the starting preview before generation. Choose clothes, then a background, then write your scene action.
2. Background
Choose a place
Built-in places are empty backgrounds. Your selected outfit is added to the preview. A custom starting frame uses its own character and clothing; the outfit picker does not replace clothing inside an uploaded photo.
For a continuous video, use linked mode and one small action per scene. The next scene starts from the previous final pose; keep direction, props and camera unchanged. All clips are aligned to the same vertical frame size and frame rate. The AI can still drift; review the joins before uploading.
Put actions in the scene boxes. The selected artwork supplies the environment automatically; these legacy notes are not sent as repeated story instructions.
Make one complete video
Write one story idea above, choose an idea chip or upload your character/background image, then tap Generate. The app prepares the scene actions and short English dialogue for the selected length. The current model renders up to 5 seconds from an image. Each next part uses the previous ending frame; it does not receive the whole previous video. The app joins the parts on one exact timeline. Model motion and identity can still drift. Keep the page open until export finishes.
Controlled motion applies negative-prompt guidance. It uses more GPU time and needs visual testing; face consistency is still not guaranteed. Standard accelerated modes use less quota but negative-prompt guidance is inactive.
Review or edit the scene actions and dialogue
Prompt pack for ChatGPT / other tools
Ask ChatGPT for JSON with a scenes array. Each scene needs duration: 5, action and voice (English). Paste the JSON into Your story idea and tap Build my scene plan. For arbitrary ideas, the automatic draft needs review. Uploaded images are starting references; this planner does not analyze their contents.
This voice stays the same across scenes. Write what the character says in each dialogue box, about 6–10 words per scene. Speech is added to the video; automatic lip-sync and singing are not supported.
An uploaded voice or song replaces generated dialogue. Upload it again after reopening.
Built-in styles are simple locally synthesized instrumentals, not YouTube Audio Library recordings. Uploaded music keeps its own style.
Download an MP3 from YouTube Studio → Audio Library, preferably with the “Attribution not required” filter, then choose that file above. The app mixes your selected track behind the voice and lowers it during speech. It does not automatically download music or verify its source. For Creative Commons tracks, copy the required attribution into your video description. Audio Library instructions.
Community GPU availability and daily limits apply. Keep this page open and the screen on while the queue runs. Your reference and prompt are sent to the selected Hugging Face demos when you generate. No paid API is used.
Free account connection / quota help
If the model asks you to sign in, use your own free Hugging Face read token. Tap Save token to remember it on this phone/browser, including after reopening. It is stored in browser storage and sent only to the video demo. Use a read token on your own device. Clearing browser data removes it.
Connected Hugging Face accounts
Add only accounts you own. Tokens stay in this browser and are sent only to Hugging Face. Multiple tokens for one account share one quota. Multi-account pooling is not a verified Hugging Face-approved feature; using extra accounts to multiply free allowance may trigger abuse restrictions. No daily output guarantee.
0 / 20 accounts added · 20 remaining
Encrypted backup / transfer
Prompts, settings, account tokens, queued scenes and saved videos ko doosre browser ya domain par le jane ke liye encrypted backup banao. File sirf tumhare device par banti hai; passphrase kahin save nahi hota.
Backup mein HF tokens shamil hain. File aur password alag safe jagah rakho; restore karne se is browser ke current prompts, accounts aur queue replace honge.
Maximum 20 verified accounts. Switching occurs only on confirmed quota rejection; model and network errors stop generation. Reset times belong to each account.
Checking a valid token activates account mode. Save it to keep it after reopening. Free accounts receive 5 minutes of GPU processing per day; public requests use a shared 2-minute allowance. Quota resets 24 hours after first GPU use. Changing mode does not erase completed scenes.
YouTube upload guidance
Reviewed 8 October 2026 against YouTube’s official guidance. This app prepares video files; it cannot certify copyright clearance, monetization eligibility or your audience setting.
- Use your own character, stories and dialogue. A different background alone does not make a repeated storyline original. Each episode should have a distinct story, lesson or creative idea; avoid misleading titles/thumbnails.
- Review the final video: coherent beginning, action and ending; clear speech; no distorted bodies, unsafe activities or unrelated stitched scenes. Avoid low-quality, confusing or sensational content aimed at children.
- If your intended audience is children, select Made for kids accurately in YouTube Studio. Cartoon style alone does not decide the audience.
- Clearly fictional animation normally does not require GenAI disclosure. Realistic synthetic content and AI-generated music may require the AI use disclosure; follow the official guidance for your final video.
- For music, use your own licensed recording or YouTube Audio Library. Some Audio Library tracks require artist attribution in the video description. Do not upload copyrighted nursery-song recordings or clone a real person’s voice without permission.
Originality & monetization · Kids quality principles · Made for kids · GenAI disclosure · YouTube Audio Library