Most families have a small pile of photographs that matter more than everything else in the album. Animating one is a way of looking at it again — a few seconds where the light moves and the person breathes.
Prepare the photo first
This step decides the result more than any prompt does.
- Capture flat and even. If you are photographing a print, lay it down, avoid glare, fill the frame, and shoot straight on rather than at an angle.
- Crop to the subject. Give the model a face, not a whole album page.
- Keep the best copy. If there is a scan and a phone snapshot of the same picture, use the scan.
- Don't over-restore. Heavy denoising can smooth away the features that make the face recognisable.
Ask for small, real motion
The instinct is to ask for a lot. The results people keep are almost always the restrained ones.
Good prompts here are quiet: "she blinks and smiles gently, slight head turn, soft window light." Ask for a dance routine and you will get something that looks like a different person. Subtle motion is both more convincing and more moving.
The general workflow is the same one described on the photo to video page: upload, describe, generate, save. If your source is a recent iPhone Live Photo rather than an old still, live photo to video has the specifics for that.
What people make with it
- Birthday and anniversary videos — a short animated clip of the person the day is about.
- Memorial pieces — a few seconds of movement instead of a static slide in a family montage.
- Family history — a grandparent's wedding photo in motion, shared with people who never met them.
- Gifts — one clip, sent to a sibling, that lands harder than anything bought.
For a longer piece from several photos, generate multiple scenes and stitch them as described in long AI video. For the two-person embrace many people are looking for, see the AI hug video guide.
Honest limits
Faded and low-resolution faces drift. Damage, creases and glare confuse the model. Group photos rarely animate well — crop to one or two people instead. And the model is guessing: it generates plausible motion, not a recovered memory. It does not know how that person actually moved, and pretending otherwise is the one claim we won't make.
There is also a judgement call, especially with photos of people who have died. Keeping the motion small tends to keep the clip in comforting territory. Making them speak words they never said is a different thing entirely, and worth thinking twice about — it is easy to do, and it is not always kind.
What it costs
Rigel is free to download on iOS 16.6+ and Android and includes starter credits, so the first photo costs nothing to try. Credit packs and a subscription are available in-app; the credit cost of each generation is shown before you confirm.
Start with one photo and ask for one small movement. That is usually the version you send to the family group chat.