How to make an AI video from a photo

Turn any photo into a moving AI video in a few clicks. Learn what motion suits portraits, product shots, and old photos, with prompt examples for each.

To make an AI video from a photo, upload the image to Morphic, describe the motion you want in plain language, and generate. The model animates your still into a short moving clip. The key choice is how much motion to ask for: portraits want subtle life, product shots want a slow reveal, and old photos want gentle movement.

Steps at a glance

  1. Upload a clear, well-lit photo
  2. Choose motion that suits the subject
  3. Describe the movement in the prompt
  4. Generate and keep the subtle take
  5. Add sound and export

Animate your photo step by step

1.

Start with the right photo

Pick a photo with one clear subject, good light, and an uncluttered background. Sharp focus matters more than resolution: a crisp subject animates cleanly, while a soft or busy image gives the model more to invent and get wrong. If the picture is old, scratched, or faded, restore and enhance it first, which is a paid feature, so you are animating a clean frame rather than amplifying damage. A strong starting image is the single biggest factor in a believable result, so it is worth spending a moment to choose the best frame you have before you write a word of prompt.

2.

Decide what should move

The biggest factor in a believable result is choosing motion that fits the subject. A portrait wants stillness with small signs of life, like breathing or a blink; a product wants a slow, controlled reveal that shows it off; a landscape wants environmental movement like drifting clouds or rippling water. Decide what should move before you write the prompt, and deliberately leave everything else still. Trying to animate the whole frame at once is what produces the warped, uncanny results people associate with bad AI video.

3.

Describe the movement, gently

Write the motion in plain language and keep it subtle, naming the movement and its pace rather than asking for a lot of it. Different subjects call for different prompts, so match the words to the photo in front of you.

subtle breathing, a soft blink, gentle hair movement in a light breeze, slow push-in

4.

Generate and keep the take that holds

Generate the clip and use the seconds that stay convincing. If the face drifts or an edge warps, you asked for too much motion, so dial it back and regenerate rather than adding more, since more movement only gives the model more to get wrong. Subtle almost always beats dramatic when the goal is a photo that has come to life rather than an obvious special effect. Generate a couple of variations and keep the one where the subject holds together longest; you only need a few good seconds.

5.

Add sound and export

A little audio finishes the piece: ambient sound for a landscape, a soft music bed for a memory, or a voiceover if the person should speak. Assemble on the Compose timeline if you are combining several animated photos into one sequence, ordering them and laying a single audio bed underneath, then export. Free exports carry a watermark; a paid plan exports clean and at higher resolution, which matters if the clip is going somewhere that shows it at full size.

What motion suits which photo

Different photos call for different movement. This is the quickest way to choose: pick your photo type and use the motion that keeps it believable.

Photo typeBest motionAvoid
Portrait / headshotSubtle breathing, blink, slow push-inHead turns, big expression changes
Product shotSlow rotation or reveal, light shiftingFast spins, background motion
Old / family photoGentle life in the face, very slow zoomDramatic movement, style changes
Landscape / sceneDrifting clouds, water, foliageMoving the camera through the scene

If you want the bare mechanics of the upload-and-animate flow, create a video from your image covers it directly, and the image-to-video generator is the tool itself. To animate a photo of yourself so you move and speak, see how to make an AI video of yourself.

Put it together

Animating a photo well comes down to one discipline: choose motion that suits the subject and keep it subtle. Start from a clean, sharp image, name the movement and its pace, and use the take where the face and edges hold. A portrait that quietly breathes is far more convincing than one you asked to do too much; restraint is what separates a photo brought to life from an effect that gives itself away.

Preguntas frecuentes

What kind of photo works best for AI video?
A clear, well-lit image with one obvious subject and an uncluttered background animates most smoothly. High resolution helps, but sharp focus on the subject matters more than raw megapixels. Busy scenes give the model more to get wrong, so simpler compositions produce cleaner motion.
Can I animate an old or damaged photo?
Yes. Old photos animate well, and gentle motion suits them better than dramatic movement. If the photo is scratched or faded, restore and enhance it first (enhancement is a paid feature) so the animation works from a clean image rather than amplifying the damage in every frame.
How much motion should I ask for?
Less than you would expect. Subtle, named motion like 'gentle breathing' or 'a slow push-in' reads as real, while 'fast' or 'sweeping' motion gives the model more to get wrong. Match the movement to what the photo is of.
Will the person still look like themselves?
For subtle motion, yes: the model animates the existing image rather than redrawing the face. Keeping the movement gentle protects the likeness; the more extreme the motion you ask for, the more the face can drift. Start subtle and add only if needed.
Do I need to edit the photo first?
Usually not, but a quick cleanup helps for older or lower-quality images. Enhancing sharpness, fixing exposure, or restoring damage before you animate gives the model a better starting frame, and a better starting frame produces a cleaner, more believable video.