AI product video
Make a product video from your own photos
Give a video model a few photos of your product and it keeps the product the same while it builds the shot around it. On Root, the seven video models below take product photos as reference images, and a five-second 720p clip made from them starts at $0.49, with Kling 3.0 Omni. Or start from one packshot as the first frame for a hero shot.
Updated
First month free on the Beginner plan, with 650 credits to try any model.
Product work made in Root
Video models that take product photos
How many photos of your product each model takes in one clip, and what one 5-second clip made from them costs on Root, at 720p where the model offers it and with sound off where it can be. 100 credits is $1.
| Model | Product photos in one clip | The clip | Price | Where it runs |
|---|---|---|---|---|
| Seedance 2.5 | Up to 30 | 5 s, 720p, no sound | up to $2.03 | App, API, Claude, ChatGPT |
| Seedance 2.0 | Up to 9 | 5 s, 720p, no sound | up to $1.33 | App, API, Claude, ChatGPT |
| Seedance 2.0 Fast | Up to 9 | 5 s, 720p, no sound | up to $1.06 | App, API, Claude, ChatGPT |
| Seedance 2.0 Mini | Up to 9 | 5 s, 720p, no sound | $0.65 | App, API, Claude, ChatGPT |
| Kling 3.0 Omni | Several | 5 s, 720p, no sound | $0.49 | App, API, Claude, ChatGPT |
| Veo 3.1 | Up to 3 | 8 s, 720p, with sound | $3.68 | App, API, Claude, ChatGPT |
| MiniMax H3 reference to video | Up to 3 | 5 s, 1080p, with sound | about $2.39 | Root app only |
Veo 3.1 always makes clips of eight seconds when it has reference photos. Veo 3.1 and MiniMax H3 reference to video always make sound. MiniMax H3 reference to video is priced at its High quality, the default; at Turbo quality the same clip is about $0.33. “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.
Three product shots and how to get them
- A turntable: wire in the product photos as references and ask for the turn, with the camera still and a plain background.
- A hero shot from one packshot: wire the packshot in as the first frame instead and describe the move, such as a slow push in or a light sweeping across the label. The clip starts on your photo, so the first frame is exactly your product.
- The product in use: keep the references and change the scene. Ask for a hand pouring it, a runner lacing it up, a kitchen in morning light.
From packshot to product clip in Root
- Bring in the product photos. Use the product photos you already have; the product photography guide says how to shoot them. For a hero shot, add one packshot from the angle the clip should open on.
- Choose the model and wire them in. Add a Video Gen node with a model from the table and connect the photos to it as references, or the packshot as the first frame.
- Describe the scene and the camera, not the product. The model can already see the product. Say where it is, what happens around it and how the camera moves: “on wet black stone, light rain, slow push in”.
- Run a few takes and check the label. The price shows before each run. Keep the take where the product and its lettering look right; the others stay in the project.
Related guides
Product video questions
What is the best AI product video generator?
One that takes photos of your own product, so the product stays the same while the shot changes around it. On Root, the seven models in the table above do, and Seedance 2.5 takes the most, up to 30 photos in one clip. A five-second 720p clip made from product photos starts at $0.49, with Kling 3.0 Omni; at 480p, Seedance 2.0 Mini is $0.31.
Should I use reference photos or a first frame?
Reference photos for a turntable or the product in use, where the model frames the shot; one packshot as the first frame for a hero shot that must open on your photo. The image to video guide says which models cannot mix the two in one clip.
How do I make a product turntable video with AI?
Wire the product photos in as reference images and ask for the turn, not the product: “the bottle turns slowly on a white plinth, the camera stays still, soft studio light”. Keep the background plain so the product stays the centre of the shot, and make a few takes.
Will the AI keep my label and logo right?
Not always: small lettering can change from frame to frame. A sharp close-up of the label among the photos helps. Run a few takes and keep the one where the label reads correctly; every take is saved in the project, so nothing is lost.
Can I make product videos by API or from ChatGPT?
Yes. Seedance 2.5, Seedance 2.0, Seedance 2.0 Fast, Seedance 2.0 Mini, Kling 3.0 Omni and Veo 3.1 run in the Root app, by API, and from Claude and ChatGPT. By API, send the photos as reference_urls, or one packshot as image_url for a hero shot. MiniMax H3 reference to video runs in the Root app only.
Can a product video have sound and music?
Yes. Seedance 2.5, Seedance 2.0, Seedance 2.0 Fast, Seedance 2.0 Mini, Kling 3.0 Omni, Veo 3.1 and MiniMax H3 reference to video can make sound with the picture, and Veo 3.1 and MiniMax H3 reference to video always do. Root also makes music tracks to your description, up to 3 minutes long for $0.16 a track, saved in the same project as the clips.
Can I try AI product video for free?
For a month, yes. The first month of Root's Beginner plan is free and comes with 650 credits, enough for 13 5-second Kling 3.0 Omni clips from product photos. After that the plan is $9 a month, and you can cancel whenever you like.
Try Root free for a month
The first month of the Beginner plan is free and comes with 650 credits to spend on any model. After that it's $9 a month, and you can cancel whenever you like.