Turn an image into a video with AI

Images turned into shots in Root

What each model can start from

The longest clip is for a clip made from an image. Luma Ray 3.2 makes up to 10 seconds from a prompt alone.

What a clip from an image costs

A dash means the model doesn't take a last frame. “In the app” means the model makes that clip and the app shows its price before you run it. Where a length is shown after the price, the model always makes a clip that long with a last frame. On Kling 3.0, a last frame makes the clip at 1080p, so it costs more. “Up to” is the most it can cost; you are charged what the model reports it used. Prices marked “about” are for work that runs on Root's own machines: the price shown before the run is an estimate, and the bill is set by how long the run really takes.

What makes a good starting image

• Treat it as the first frame of the shot: the framing, the light and the subject should already be where the shot begins.
• Leave room for what moves. If a car will drive into frame, leave the road clear.
• On Seedance 2.5, Seedance 2.0, Seedance 2.0 Fast, Seedance 2.0 Mini and MiniMax H3 (Hailuo 3.0), the clip takes the shape of your image, so crop it to the shape you want first.
• Seedance needs an image at least 300 pixels on each side and no more than 6000.
• No image yet? Make the still in Root first with an image model, then animate it in the same project.

First frame, last frame or references

• First frame: the clip starts on your image. Use it when you know exactly how the shot opens.
• First and last frame: the clip moves from one image to the other. All four Seedance models, Kling 3.0, Kling 3.0 Omni, Kling 2.6, Veo 3.1, Luma Ray 3.2 and MiniMax H3 (Hailuo 3.0) can do it.
• Reference images: the model keeps a character, a costume or a product from your pictures but frames the shot itself. Seedance 2.5 takes up to 30; Seedance 2.0, Seedance 2.0 Fast and Seedance 2.0 Mini take up to 9; Veo 3.1 takes up to 3; Kling 3.0 Omni takes several.
• On Seedance and Veo, use either first and last frames or reference images in one clip, not both.
• FLUX 3 Video takes the first frame and up to 4 more frames, and moves through them in the order you give them.

How to turn an image into a video in Root

1. Start a project. Create an account, then open a project. Upload your image, or make one there with an image model.
2. Pick a video model. Add a Video Gen node and choose the model. The tables above show what each one takes and what it costs.
3. Wire in the image. Connect your image as the first frame. Add a last frame or reference images if the model takes them.
4. Say what moves. Describe the action and the camera, not what is already in the picture. Then set the length, the size and the sound.
5. Check the price and run. The price shows before you run it. Every take is saved in the project, so you can compare them and keep the best.

Image to video by API

curl https://root.film/api/v1/generations \
  -H "Authorization: Bearer rfk_your_key" \
  -H "Content-Type: application/json" \
  -d '{"model":"kling-v3","project_id":"YOUR_PROJECT_ID",
       "input":{"prompt":"the wind picks up and the flags start to flap",
                "image_url":"https://example.com/still.png",
                "duration_seconds":5,
                "resolution":"720p"}}'

• Add last_frame_url to end on a second image, on a model that takes one.
• Make an API key in Root under Settings, then API, and choose a daily spending limit for it.
• The finished file is saved to the project you name, so your team sees it in the app too.
• Send an Idempotency-Key header and a retry never charges twice.

Related guides

• Best AI video generator
• Text to video
• AI video prompts
• Consistent characters
• Kling
• Seedance
• Every guide

Image to video questions

Try Root free for a month