Turn an image into a video with AI
Attach an image in the Tella Agent chat and say how it should move. Tella generates a short AI clip that starts on your image and places it in your video.
Try it in Tella
Paste a prompt into the Tella Agent in the editor, or into Claude or ChatGPT connected to Tella, then ask for changes.
I've attached a photo of our Tallyfern ceramic mug on a wooden table. Turn it into a five-second video clip with steam rising from the mug and a slow push-in, in warm morning light. Place it as b-roll over the part where I talk about the new range.
Use the Quillhaven backpack image in this video's media library as the first frame of a 10-second clip. Orbit the camera slowly around the backpack on a plain studio background and keep the image's aspect ratio. Put it over my opening line.
Animate the attached photo of a mountain lake for my intro: clouds drifting, small ripples on the water and a gentle pan to the right. Use Seedance 2.0 Fast for a first draft and show it as b-roll for the first five seconds.
How do I turn an image into a video with AI?
In Tella's editor, attach the image in the Tella Agent chat and ask for a video clip that starts from it, saying what should move and how the camera goes. Before it generates, the agent shows a card with the model and clip length; click Generate clip to start. It then places the clip in your video as b-roll or an overlay. To do it yourself, add the image as a reference in Generate at the top of the Media Library and choose Video.
How to do it
- Open your video in the Tella editor and open the Tella Agent chat (Shift + S if the sidebar is hidden).
- Click Add (+) → Attach… and pick the image, or name an image already in the video's media library.
- Describe the motion, the camera move, the style, the length and where the clip goes.
- Check the AI video clip card and click Generate clip.
- Preview the clip and ask the agent to move, trim or regenerate it.
Good to know
- Needs a paid plan or the 7-day trial. Messages use your workspace's credits.
- Your image becomes the clip's first frame. Seedance 2.5 keeps the image's aspect ratio; the Seedance 2.0 models crop it to 16:9.
- Video generation refuses an image that shows a person, real or AI-generated, so start from a product, place or object.
- Clips are 5 seconds unless you ask for longer: up to 30 seconds on Seedance 2.5 or 15 seconds on the Seedance 2.0 models.
Frequently asked questions
Tella uses Seedance 2.5 at 1080p by default. Seedance 2.0 Fast and Seedance 2.0 Mini are 720p options that use fewer credits. To pick one, name it in your message to the Tella Agent.
ChatGPT can. Connected to Tella's MCP server, it passes an image from your Tella library to the generate_video tool as the clip's first frame, then places the clip in your video. Claude, Cursor and other MCP clients don't list the generation tools, so use the Tella Agent in the editor instead.
Yes. Ask the Tella Agent to generate the image first, then to turn it into a video clip. Leave people out of the image, or the video step is refused.