AI video canvas
A branching canvas for AI images and video
Every prompt becomes a node you can branch from. Explore alternatives, compare them side by side and keep characters consistent across scenes, without wiring up a graph by hand.
Start creatingHow it works
-
Start a thread
Write a prompt for an image or video. The result appears as a node on the canvas of your thread.
-
Branch from any result
Pick any node and generate from it again. Each branch keeps its own history, so trying a new idea never overwrites the old one.
-
Reuse characters with @references
Upload a photo, give it a name, then type @Name in any prompt to bring that character or object into new scenes.
-
Compare and keep the best
Look at the alternates together and send the winners to the timeline to build a longer video.
Frequently asked questions
What is an AI video canvas?
It is a workspace where each generation is a node and each new prompt branches from one. You see the whole exploration at once instead of a single linear chat.
Do I need to connect nodes myself?
No. Branches are created for you when you generate from a result, so there are no connections to wire up by hand.
How do @references keep characters consistent?
A reference is a named photo of a character or object. Mentioning @Name in a prompt gives the model that image as input. Results can still vary, so branch and compare.
Can I generate images as well as video?
Yes. Threads hold both image and video generations, and you can start a video from an image result.
What happens to the branches I don't use?
They stay on the canvas. Nothing is overwritten, so you can return to an earlier branch whenever you like.