What goes in, and what does each input control?
Two inputs are required: photos and a prompt. Settings decide the length, sharpness and frame.
| Input | Limit | Controls |
|---|---|---|
| Photos (required) | 1–4; JPG, PNG or WebP up to 10 MB | Who appears: face, hair, build and usually clothing |
| Prompt (required) | Up to 1,500 characters | Place, action, camera movement, mood and sound |
| Duration | 5, 10 or 15 seconds | Clip length and credit charge |
| Resolution | 480p or 768p | Sharpness and credit charge |
| Frame | 9:16, 16:9 or 1:1 | Vertical, wide or square output |
fal's endpoint also accepts video and audio references; this site never sends them.
How do you make a video from your photos?
- Pick your photos. Use sharp, evenly lit photos with your face clearly visible and one person per photo. Add a second or third angle if you have one.
- Write the scene. Start with who (Image 1), then where, what they do, how the camera moves and what it sounds like.
- Choose settings. 9:16 for phone video, 16:9 for wide screens. Start at 480p and 5 seconds to check the idea.
- Review the quote, generate, and check the face, hands and clothing before you share the result.
You can preview photos, draft a prompt and estimate cost in the image-to-video tool. New generation on this site uses account credits and passes safety checks. Purchases remain closed.
Which prompt should you use?
Name the photo, then describe one clear scene. Short, concrete prompts work better than long lists.
Image 1 walks along a quiet beach at sunset, smiling at the camera. The camera tracks beside her at walking pace. Soft waves and wind.
Image 1 stands in a bright kitchen and holds up a ceramic mug, talking to the camera. Slow push in. Warm morning light, light background music.
Image 1 and Image 2 are the same man from different angles. He jogs through a city park in autumn. Handheld camera, leaves falling, city sounds.
These are templates, not tested results. Do not name celebrities, ask for a specific real person's voice, or describe intimate or deceptive scenes; such prompts are held or blocked.
Write a short scene description using the templates above. Video to Prompt is not open yet.
How much does a photo-to-video clip cost?
fal charges $0.05 per output second at 480p and $0.06 at 768p for the first five reference images.
| Clip length | 480p at fal's rate | 768p at fal's rate |
|---|---|---|
| 5 seconds | $0.25 | $0.30 |
| 10 seconds | $0.50 | $0.60 |
| 15 seconds | $0.75 | $0.90 |
These are fal's upstream rates, checked October 16, 2026. See pricing for credit packs and the credits each clip uses on this site.
What are the limits?
- Likeness can drift. Faces, hair and hands may change between frames. More angles and clear lighting help.
- Clothing follows the photo unless the prompt asks for something else.
- Short clips. 15 seconds is the maximum; make longer pieces from several clips.
- No editing of real footage. The model cannot keep or modify an existing video, and Image to Video does not accept one.
- Sound is generated. It will not be your voice, and lip movement may not match speech.
Consent and disclosure
Use photos of yourself, or of adults who explicitly authorized this use, and hold the rights to them. Image to Video takes photos and text only, does not swap faces and does not replace people in existing footage. Impersonating anyone, public figures, fabricated statements or evidence, undressing and intimate scenes are prohibited even with permission. See our content policy. Label published results as AI-generated: YouTube requires creators to disclose realistic content made or meaningfully altered with AI, and other platforms have similar labelling rules.
FAQ
Which model generates the video?
MiniMax H3 reference-to-video, a third-party video generation model served by fal, that uses up to nine reference images and a prompt to create a new clip. This site uses it with 1–4 photos and no video or audio references.
How do I tell the model which photo is which?
Refer to photos by order: Image 1, Image 2 and so on. If your prompt does not mention them, we add a short line naming the photos as the main subject.
How long can the video be?
5, 10 or 15 seconds. Credits are charged per second of output at the resolution you choose.
Does the clip have audio?
Yes. The model generates a soundtrack. It does not clone a voice, and you cannot upload audio.
Can I use it to change the person in an existing video?
No. Image to Video takes photos and text only, and the site never swaps faces or replaces people in footage.
Can I make a video of a celebrity or a friend?
Not of public figures. A friend is only allowed with their explicit permission, in an ordinary, non-deceptive scene. Intimate, sexual or misleading content is prohibited even with consent.
Sources
Inputs and prices are from fal's model page and API documentation (checked October 16, 2026).