ImageCombineOneColumn

ImageCombineOneColumn is ImageCombineOneRow rotated ninety degrees: instead of gluing up to twelve images side by side, it stacks them top to bottom into one tall image. Same inputs, same idea, same one-word difference in the output direction. In the ComfyUI-Champ reference workflow it works alongside its row twin to build the comparison grid - your CHAMP output frames plus the depth, pose, and semantic-map previews that drove them - so you can save one vertical strip showing everything at once.

How it works

Under the hood it's a torch.cat along the height axis of the IMAGE tensor. Output height is the sum of the inputs; width stays whatever your images were. Same two rules as the row version:

• image1 and image2 are required; image3 through image12 are optional and it stops at the first disconnected one. You can leave nine slots hanging and it just concatenates what's there.
• Everything must match in width, channels, and frame count - this is a concat, not a resize. A 512-wide frame beside a 640-wide frame errors, and a 16-frame batch next to an 8-frame batch errors too. If you want a montage that resizes, this isn't the node; grab a proper image-grid utility from another pack.

Which one do you actually use?

Honestly, whichever reads better for what you're checking. One column suits a long sequence where you want to trace a single artifact down the stack frame by frame; one row suits comparing conditions across the same frame. The shipped workflow wires the same content into both and lets VHS_VideoCombine save whichever you care about. For a 16-frame CHAMP run, a stacked column gets tall fast - expect to zoom.

Output and install

One IMAGE, same frame count and width as the inputs, just taller. Nothing CHAMP-specific about it: it's a general helper that lives in this pack because the author needed it for the preview grid. Install it with the pack - ComfyUI Manager (search "Champ") or:

cd ComfyUI/custom_nodes
git clone https://github.com/chaojie/ComfyUI-Champ

then restart ComfyUI. The pack itself is the heavy part (see the ChampLoader and ChampRun pages); this node is as lightweight as it gets.