Draggen Board to Image

Draggen Board to Image gives you what you see in Draggen: the entire canvas - images, boxes, notes - composited into a single IMAGE, with element positioning and z-index respected. Where Extract Images takes the board apart, this one puts it back together as one picture.

The mechanism is a small compositor, and understanding it explains every limitation you'll hit. It computes the bounding box of all elements, pads it by 50px, and creates a transparent RGBA canvas. Then it draws in z-index order: each image is resized with Lanczos to its element's size and composited at its position; boxes become rectangles with parsed hex fill and a hardcoded 1px border; text is drawn with a default font. One input (moodboard), one output: a single IMAGE.

What it's good for

That single image is genuinely useful for two jobs:

• A visual overview - drop it on a PreviewImage and eyeball the whole board before generating.
• A composite style reference - the entire board in one conditioning slot, rather than juggling dozens of extracted images. That's the appeal: one canvas, one reference, no per-image management.

What it gets wrong

Now the honest part. Text rendering is crude. The code reaches for arial.ttf at a fixed size 24 and falls back to PIL's default bitmap font if the system font isn't found. Your board's styled notes will not look like they do in Draggen - think functional, not pretty. Two more quirks: the canvas is transparent, so empty areas of the board stay see-through if you composite the result over something; and box borders are locked to 1px with only the fill and border colors carrying over. If your board leans on fine typography or thick outlines, the render will disappoint.

None of that is a bug - it's a pragmatic first cut at flattening an editor document, and the author picked a reasonable 50/50: positions and layering exact, typography approximate. If pixel-faithful output matters, export from Draggen itself; this node is for when "good enough to preview or reference" is exactly what you want.

Installing

Install is the same pack-wide story - ComfyUI Manager → search ComfyUI-Draggen, or:

cd ComfyUI/custom_nodes
git clone https://github.com/kadima-tech/comfyui-draggen

then restart. No extra dependencies; the pack uses only requests and Pillow, both already in ComfyUI. The example workflow ships with this node wired straight into a PreviewImage, which is the right first thing to try.