- You need: just a description of the shot. Everything else is optional.
⚙ More options camera controls · add a character photo · AI enhance
How it works: the person in the photo becomes the character — AI focuses on motion and environment, not appearance.
- You need: a story idea (2–3 sentences is enough).
- Optional cast: up to 3 people — one clear, front-facing photo each + their name. The story is written about them and they speak the dialogue.
- Faces — pick one path: 🆓 Free tier + a group photo → the film opens on your real faces, AI carries the look onward · 💳 Paid tier + ✨ Stylize → Pixar/anime versions of your cast in every scene (the paid model blocks real photos — anti-deepfake policy, not a bug).
Cast your family & friends: one clear, front-facing photo per person (like a profile picture) + their name — the story is written about them and they speak the dialogue. These are identity photos; for "moment" photos use 📸 Album mode instead.
⚙ More options aspect ratio · draft/final · sound · title card
Draft = 480p preview at ~⅕ the cost — iterate here, then re-create at Final (720p free / paid tier) for the keeper.
Plain text, Markdown or screenplay (.txt, .md, .fountain, .fdx). For PDF / Word, paste the text into the box above. Best with the Series or Film format.
⚙ More options
Draft = 480p at ~⅕ the cost — preview cheaply, then re-create at Final.
- 📖 Living moments (default): each photo becomes a ~6s living scene — people move, talk, laugh — joined with cuts & dissolves. Like a slideshow where every slide is alive.
- 🌊 Seamless journey (toggle below): ONE continuous shot morphing from photo to photo — the person literally ages, stands up, changes places between frames. For growing-up / young-to-old films, upload photos in age order.
- You need: 2–5 moment photos in story order — snapshots of situations, NOT headshots. Whoever is in a photo is who appears in that scene. Plus a short brief connecting them.
Moment photos, not headshots: snapshots of situations — people doing things somewhere. Whoever is in photo N appears in scene N. Well-lit, similar framing morphs best. Captions ("age 7, 2003") help the Director understand each moment.
⚙ More options draft/final · film sound · title card
Draft costs ~⅕ of Final — preview the morphs first, especially for journeys.
- You need: ① one clear, front-facing photo (your avatar) · ② your script, word-for-word (450–900 characters ≈ 30–60s of speech) · ③ a description of the world behind you.
- Voice: pick from your ElevenLabs voices — or your own cloned voice. The avatar's mouth follows the audio exactly.
- Smart extras (toggles below): backgrounds that change as your script moves through topics · captions synced word-by-word.
- Requires API keys in .env: HEYGEN_API_KEY (lip-sync, ~$0.05/sec) + ELEVENLABS_API_KEY (voice, free tier OK).
Record yourself speaking naturally for 1–3 minutes. The AI will clone your voice — all future presenter videos will sound exactly like you.
Best results: front-facing portrait, eyes visible, neutral or slight smile. Avoid dark glasses or heavy shadows on the face.
Tip: natural conversational sentences work best. ~130 words per minute — 3000 chars is roughly 3–4 minutes of speech.
Describe the world behind you — environment, mood, movement. Don't describe yourself; the AI adds you on top.
Microsoft Azure Neural — included in D-ID plan, zero extra cost.
⚙ More options visual style · mood · aspect · captions
- You need: a description of what to draw.
- Tip: start on 🧪 Draft to nail the prompt, then upgrade the one you like to ✨ Premium.
- Also here: ✂️ Remove Background and 🙈 Blur Faces — upload your own photos, get a transparent PNG or a privacy-safe copy.
⚙ More options reference image → variations
How it works: the uploaded image guides the subject & identity — describe the new pose, outfit, or scene you want it in.
How it works: an open-source AI (BiRefNet) finds the subject and erases everything behind it — even wispy hair and fur edges. Great for stickers, thumbnails, product shots, or the Reference Character slot above.
How it works: the AI finds every face — including small ones at the back of group photos — and blurs them all. Tap a face to unblur it (e.g. the youth worker); tap again to re-blur. Tap anywhere else to add a blur patch the AI missed. Runs on your own server: photos are never sent to a third party.
- How: ① describe your song → ② pick a voice → ③ pick a genre → Generate. Lyrics are written for you (and stay editable).
- Tip: pick an Asian voice (e.g. Mandarin ♀) and the lyrics are written and sung in that language.
Sung by an AI voice in the style you pick — not your own voice.
🎤 Sing in my voice EXPERIMENTAL optional · use your own voice
Add a voice clip and your song is re-sung in that voice.
- Record ~15–30s or upload a clip — clean speech or singing, no background music/noise.
- Train voice — up to ~10 min, once. Then play its preview in “My voices” above to hear the clone.
- Use → Generate — only once the preview sounds right. Training and song generation are separate steps, so a bad clone never wastes a song.
The clip above is your input. Training renders a preview of the clone singing so you can check it in “My voices” before spending a song.
⏱ Training is one-time per voice (up to ~10 min). Reusing a trained voice for a song is much faster. Uses Replicate credits. Only use your own voice, or someone's with their permission.
⚙ More options length · DJ tempo lock
Controls how many verses the AI writes — longer lyrics = longer song.
- You need: a few videos and/or photos (a trip, a party, a project…). More footage = better picks.
- Optional: a short brief — what the video is about, the vibe, who it's for.
Shorter = a punchy highlight reel · longer = a fuller recap. Portrait 9:16 suits phones & Reels, 16:9 suits YouTube & TV.
The editor picks a camera move per shot — a person-to-person highlight across a group, a push, a reveal… Auto uses the full repertoire and matches it to your theme & the beat. Or choose exactly which moves it may use.
- Tip: editing a single scene only re-generates that clip (~1 call), not the whole film.