Image to video, step by step

You already have a picture you like. Maybe it is a product photo, a portrait, or an image you made in Create. Image to video gives that picture motion. You upload the still, describe what should move, and the model makes a short clip that starts from your image.

When image to video is the better choice

Text to video lets the model invent everything. Image to video keeps one thing fixed: the first frame. That helps when:

• You want a clip that opens on a specific look, such as your product or your own artwork.
• You tried a text prompt and the subject kept changing shape.
• You made a good still in the Image tool and want to see it move.

The first image you add is used as the starting picture. The model then follows your written description for what happens next.

Before you start: pick a good picture

The picture matters more than almost anything else. Aim for:

• A clear subject with plenty of room around it.
• Good light and no heavy filters.
• A shape that suits the clip you want. A wide picture suits a wide clip.
• A file in PNG, JPG or WebP format, up to 10 MB.

If you need a better still first, make one in Create Image, or clean one up with Create Edit, where you describe a change in words.

The steps

1. Open Create Video and switch to the Image to Video tab.
2. Upload your picture into the Start frame box. A start image is required on this tab.
3. Describe the motion in the prompt box. Say what moves and how, for example: "The camera pushes in slowly while steam rises from the cup."
4. Open the Model row and pick a model. If the model you pick does not use frames, Create tells you, and you can pick another one.
5. Set Res., Ratio and Duration. A chip that the model sets by itself shows Default and is locked.
6. If you see a Sound switch, decide whether to turn sound on. Only models that can make sound show it.
7. Choose a Batch size of 1 to 4 clips.
8. Check the credit cost on the Generate button, then press Generate.

Write the prompt for motion, not for the picture

The model can already see your image, so you do not need to describe it again. Spend your words on what changes:

• Camera: slow push in, gentle pan, steady and locked off.
• Subject action: a hand lifts a cup, hair moves in the breeze, a door opens.
• Pace: slowly, quickly, a single smooth move.

Keep it to one motion. Several actions in one short clip often come out muddled. If the result is close but not right, change one detail and run it again.

Compare before you commit

Batch size lets you ask for up to four clips in one run. Each clip is its own request and uses credits, so the cost shown is the single price times the number of clips. Try two or three on a cheaper setting first, then repeat the best idea at a higher resolution if the model offers it.

You can also run the same picture and prompt on a different model. There is no side by side view, so run one, switch the model, and run it again. Models differ in style and motion.

What image to video does not do

It is worth being clear about the limits:

• Each run makes one new clip. To build a longer piece, make several clips and join them in a video editor.
• Faces and small details can shift as the clip plays, so check each result.
• Sound depends on the model. Some models can make audio, and only those show the Sound switch.

What it costs

Create has paid plans only. Each plan includes a monthly amount of credits, and a flagship video clip of 5 seconds is about 186 credits, with the exact price for your settings shown before you generate. A failed generation is not charged. See the pricing page for plans.

Try it