TakeBoard: an open-source AI creation canvas built on ComfyUI

[Fourques]

I'm the maintainer of TakeBoard, an open-source AI creation canvas built on ComfyUI.

It brings image and video creation into one workspace: choose a model through an available workflow, connect reference media, write a prompt and generate. Keep several directions on the canvas, then use a result as the input for your next step—for example, taking an image into an image-to-video workflow.

The canvas is for creative work with media and generation nodes. Your ComfyUI still executes the underlying graph, and you can open its editor when a workflow needs deeper changes.

• Image and video tasks show the inputs and parameters supported by the selected workflow.
• Add recommended templates or import your own workflows; custom graphs may need input mapping and dependency fixes.
• Connect local or remote ComfyUI. A project can stay on your laptop while outputs are collected from your GPU server.

Watch the 42-second film—a real project, from a reference to a generated result and the next creative step. Generation wait is omitted; both editions show the actual Chinese app interface.

Try the desktop beta · Getting started · Source

macOS, Windows and Debian/Ubuntu installers are available. Local use needs no account. Bring your own ComfyUI, models and nodes; this is beta software and workflow compatibility varies. Installation notes cover the unsigned/notarization-pending packages.

I'd welcome feedback on the creation flow: can you go from a reference to a generated result and use it in the next step without losing your place? If something blocks that, tell us which step.

Apache-2.0. Independent project, not affiliated with Comfy Org.

[BlackBox-Labs]

Looks great , i wish you more success.