Video

How to turn a still image into a video clip

Image-to-video, step by step: choosing a good start frame, writing a motion prompt, and picking length and resolution for the clip.

2 min read
A 35mm film strip unspooling in the air with violet glowing frames

Text-to-video asks a model to invent everything at once: the scene, the look and the motion. Image-to-video splits the job. You settle the look first, as a still you are happy with, and the model only has to decide how that picture moves. The result is usually more controlled and more consistent with what you had in mind.

Step 1: start from a frame that can move

The start frame becomes the first frame of the clip, so its composition sets the whole shot. A few things make a still animate well:

  • Room to move. Leave space in the direction of motion - sky above rising smoke, road ahead of a car.
  • Something that naturally moves. Water, fabric, hair of an animal, smoke, clouds, leaves, light. A perfectly static object on a plain background gives the model little to do.
  • A clear subject. Busy frames with many small elements tend to smear once they move.
  • The right frame. A clip keeps the proportions of its start frame, so render the still in the ratio you want the video to be.

Step 2: animate it

In the studio, every image in your feed and library has an Animate action. It switches the composer to video with that image attached as the start frame. You can also attach any picture of your own with + Reference in video mode.

Step 3: describe the motion, not the picture

The model can already see the image. Your prompt should spend its words on what happens: what moves, how fast, and what the camera does.

Prompt
Slow push-in toward the lighthouse, waves breaking against the rocks below, spray drifting in the wind, clouds moving across the sky
Camera move first, then the motion in the scene.

Keep it to one camera move per clip. "Slow push-in", "gentle pan left", "static camera" or "orbit around the subject" each read clearly; stacking three of them in a short clip gives jittery results. Our text-to-video guide has a longer list of camera terms.

Step 4: choose length and resolution

Clips run up to 30 seconds, depending on the model. Shorter clips hold together better: a 5-second shot of one clear action usually looks more convincing than a long one where the model has to keep inventing. Price scales with length and resolution, and the composer shows the exact total before you generate.

If you are testing motion ideas, start at the lowest resolution the model offers and a short length. Once the movement is right, render the final at the resolution you need.

Which model for image-to-video

  • Motion Turbo - the fastest turnaround, up to 30 s · 1080p.
  • Motion - the highest fidelity, up to 30 s · 4K.
  • Motion Lite - the lowest cost per clip, up to 15 s · 1080p.
  • Motion Real - realistic motion and texture, up to 30 s · 1080p.

The model comparison goes through when each one is the right call.

#Video#Guides

Keep reading

All articles →