Photo to video AI

The two kinds of photo-to-video AI

The tools sharing this name split into two families. The first takes your photo plus a written prompt and produces a cinematic clip — camera moves, weather, dramatic motion, whatever you can describe. The second narrows the job: it makes the person in the photograph move the way a person actually moves, and does nothing else. That second kind is what this is.

The distinction matters because the two kinds fail differently. Prompt-driven generation invents freely and shows it — faces drift, hands arrive wrong, the room behind the subject reshapes itself. Portrait motion stays anchored to your image, and its failure mode is simply less movement than you hoped for. One surprises you; the other mostly cannot.

• Prompt-driven: you write the scene, the model invents most of it
• Portrait motion: the photograph is the entire instruction
• This site is the second kind, on purpose

How to judge a result

Watch the eyes and the mouth first — those are where generated motion breaks first, and where the eye goes anyway. A good clip keeps the face recognisably the same person throughout, with movement that a stranger would read as ordinary breathing rather than as an effect.

Second, check the frame. If the background is warping, the edges of the picture are swimming, or the composition drifts, the generator is redrawing your image instead of moving it. Nothing redrawn survives a close look, and a photograph of a family member deserves better than a nearly-right face.

• Eyes and mouth stay the person you uploaded
• The background holds still; the frame does not drift
• Movement reads as breathing, not as an effect

Why this one has no settings

Most generators hand you a prompt box, a model picker and a duration slider, and leave you to find the combination that works. This site removes all of it: the photograph is the only input, and the motion is kept to the range a still image can support without inventing detail.

That narrowness is the trade. You cannot direct a scene or request a gesture, and if you want those things a prompt-driven tool will serve you better. What you give up in control you get back in not having to learn anything — the result does not depend on your prompt-writing skill, because there is no prompt.

What the first run looks like

Sign in with Google — 300 credits are added at that moment, enough for 1 video — then upload one picture and press the single button. Generation takes one to three minutes and returns a few seconds of 720p video. There is no card step anywhere in trying it.

The photo formats are JPG, PNG and WebP, up to 10 MB, which covers a scan or a phone picture without any preparation beyond making sure the face is large and sharp in the frame.

Keeping the result

The finished clip downloads like any file. The free link is temporary because the storage sits with the model provider, so save it when it arrives, or use Save to cloud in your account to keep a permanent copy for a small credit charge.

After that the picture is yours to reuse: nothing about your original changes, and generating again from the same photograph costs the same as the first attempt.