How to Make a Picture Move With AI
A moving picture is a still plus one decision: what should move. Get that decision right and the model does the rest.
18+ only · fictional characters only · no public gallery
The short version
Three decisions that matter more than anything else.
Start from one clear subject
One person, one object or one view, sharp and fully inside the frame. Busy pictures split the motion across everything and nothing moves well.
Decide what moves
Either the subject moves (hair, fabric, a head turn) or the camera moves (push-in, orbit, drift). Pick one. Asking for both is the main cause of warping.
Keep the first clip short
Five seconds is enough to see whether the motion works. Lengthen only a clip that already looks right.
Which pictures move well
The pictures that animate cleanly have three things in common: the subject is in focus, it is the only thing competing for attention, and nothing important is cut off at the edge. A model asked to move a cropped arm has to invent the rest of it, and that invention is where smearing starts.
Portraits, product shots and landscapes all work, but they want different motion. If you are starting from a single photo, an image to video generator keeps your framing and only adds movement, which is usually what you want for a picture you already like.
Subject motion or camera motion
Subject motion changes what is inside the frame: breathing, blinking, hair in the wind, a slight turn of the head. It suits portraits and characters, because a face that stays in place while small things move reads as alive rather than distorted.
Camera motion changes the frame itself: a slow push-in, a parallax drift, an orbit. It suits landscapes, rooms, products and old archive photos, because nothing in the scene has to change shape. For a character you plan to reuse across several clips, the AI character video generator is built around keeping the same face between runs.
Why faces and backgrounds warp
Warping is almost always too much change for the information in the source. A low-resolution face asked to turn ninety degrees has nothing to turn into, so the model guesses. A cluttered background asked to parallax has too many edges to keep consistent.
The fixes are boring and reliable: upscale the still before generating, crop closer to the subject, pick a gentler motion template, and shorten the clip. Changing the prompt rarely fixes warping; changing the input almost always does.
When to start from text instead
If you do not have the picture yet, there is no reason to generate a still first and then animate it. Describing the shot directly in a text to video generator gives the model freedom to compose the frame for motion from the start, which often looks more natural than animating an image that was composed to be still.
The trade-off is control: with text you describe the result, with an image you dictate it. Use text when the idea matters more than the exact look, and an image when the look is already decided.
Frequently asked questions
Can any picture be made to move?
Most can, but sharp, single-subject pictures give the cleanest results. Very small, blurry or crowded images move poorly.
How long should the clip be?
Start with about five seconds. Short clips keep the subject consistent; extend only once the motion looks right.
Why does the face change during the clip?
Too much motion or too long a clip for the detail in the source. Use gentler motion, a shorter clip, or upscale the photo first.
Do I need to write a prompt?
Not with template-based generation. You pick the motion template and supply the picture.
Can I animate an old family photo?
Yes. Slow camera motion such as a gentle push-in works best on archive photos. Upscale small scans first.
Try it on your own image
New accounts get starter credits. No card needed to run a first generation.
Try it yourself