Create a cinematic AI video from one photo

Turn one photo into a short cinematic AI video: add the image, describe the motion, pick quality and generate. Includes prompt tips and what each clip costs.

BeginnerTutorial video 0:233 min read

Create video turns a still image into a short moving shot. The photo fixes the look of the scene — the people, the light, the set — and your prompt decides what happens next. A five-second clip takes about a minute, and the price is printed on the Generate button before anything runs.

The clip below started from a single still of a knight facing a stone golem. We added it as the first frame, wrote one paragraph of action and camera movement, and generated it with MiniMax H3 Max.

Made from one still image with MiniMax H3 Max, 5 seconds, image to video.

Step 1: Open Create video

Go to Create video. The left panel holds everything you need: the photo slot, the prompt, the model and the quality. The right side shows examples until you generate, then your results.

The Create video panel with the empty photo slot, prompt box, model picker and orange Generate button
The whole workflow lives in the left panel; Generate shows the credit price before you click.

Step 2: Add your photo

Click Start from a photo and choose an image, or pick one from your library. The photo becomes the first frame, so everything you see in it — faces, clothes, colors, framing — carries into the video. A second slot, End on this photo, appears if you want the clip to finish on a specific frame.

The photo slot now shows the knight and golem still, with an optional End on this photo slot beside it
The uploaded still is the first frame. Add an end frame only when you need the shot to land somewhere exact.

Use a sharp, well-lit image at the aspect ratio you want to publish. A wide 16:9 still makes a wide clip; a vertical still makes a vertical one.

Step 3: Describe the motion

Write what moves, in the order it happens, then how the camera behaves. You do not need to describe what is already visible in the photo — the model can see it. Spend your words on action and camera.

The prompt box filled with a description of the golem's attack, the knight's dodge and an arcing camera move
Action first, camera second. One continuous shot keeps the scene from cutting away.
The prompt we usedModel: MiniMax H3

The golem slams its fist down and the floor shatters; the knight rolls aside, leaps onto its arm and drives the glowing sword into a molten crack. Camera swings in a dynamic arc around the clash. One continuous shot.

Step 4: Pick quality and generate

Open the quality options and choose a resolution. Standard is the cheapest way to check an idea; Sharp is a good default for sharing; Ultra sharp is for the final take. Each option shows its price, and so does the Generate button.

The resolution list with Standard, Sharp and Ultra sharp, plus duration and number of videos above the Generate button
Resolution, length and the number of videos all change the price on the button.

With MiniMax H3, a five-second clip costs 125 credits at Standard, 185 credits at Sharp and 290 credits at Ultra sharp. Quick try makes a cheaper preview first, so you can check the motion before paying for a sharper take. If a generation fails, the credits go back to your balance.

Tips for better results

  • One idea per clip. Five seconds fits one action and one camera move. Chain clips with Extend video for longer scenes.
  • Keep the subject consistent. Ending with "Same faces, hairstyles and outfits throughout" reduces drift on people.
  • Name the camera. "Slow push-in", "arc around", "FPV dive" and "locked-off" all change the feel. See Camera moves.
  • Avoid text in the frame. Signs and captions tend to come out garbled; add text afterwards.

FAQ

Can I make a video without a photo?

Yes. Switch the mode to text to video and describe the whole scene in the prompt. Starting from a photo gives you much more control over how the shot looks.

Which model should I start with?

The default, MiniMax H3, is a good balance of quality and price. Switch to MiniMax H3 Max, Kling or Seedance in the model picker when you need more detail or a specific look; the button price updates as you switch.

How long can one clip be?

It depends on the model; most produce five to ten seconds per generation. Use Extend video to continue a clip from its last frame.

Do I lose credits if a video fails?

No. If a generation fails, the credits it used are returned to your balance automatically.

Next tutorials

All tutorials

Create video2 min read

Prompt writing for cinematic AI video

A four-part structure for AI video prompts that move like film: scene, action written as cause and effect, one camera move, and a style lock. With examples.

Pricing2 min read

What one AI clip really costs on PopNow

How PopNow credits turn into dollars, what a five-second clip costs on each popular model, how bonus packs lower the price, and how to spend less per clip.

Try it while it’s fresh

The tool opens with the settings from this tutorial. You’ll see the credit price on the button before anything runs.