Preview Image Bridge (Yogurt Nodes)

ComfyUI has always been stingy with previews in the middle of a graph. The default PreviewImage node works, but it's a terminal - it wants to be the end of the line. YogurtPreviewImageBridge does the thing quality-of-life packs like rgthree made popular: it shows you the image and passes it through, so you can eyeball intermediate results without killing the branch.

One input, four outputs:

• images in - any IMAGE tensor.
• images out - the same tensor, untouched.
• width, height - the image dimensions as integers.
• batch - how many images are in the batch.

So it's a preview node and a free "what's in here?" inspector at the same time. Wire it anywhere a branch needs eyes on it - after a VAE decode, before a crop, at the halfway point of a multi-pass pipeline - and the image keeps flowing to whatever's next while you watch the result.

Why the extra outputs matter

The width, height, and batch outputs are the reason you don't reach for the built-in preview. If you've ever hardcoded a resolution assumption into a workflow and watched it break the moment a different-sized image came through, you know the fix: pull the real dimensions from the node that's actually holding the image. Feed width/height into your crop, resize, or tile math and the workflow stops lying to itself. Same for batch if you're looping over a stack.

Install

Ships in ComfyUI-YogurtNodes. ComfyUI Manager → search ComfyUI-YogurtNodes → install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/yogurt7771/ComfyUI-YogurtNodes.git
cd ComfyUI-YogurtNodes
pip install -r requirements.txt

Restart ComfyUI and look under "Yogurt Nodes" - the display name carries the " (Yogurt Nodes)" suffix. No models, no keys, nothing to download.

One honest note

If you want the same pass-through preview but prefer the name to reflect where it sits, the pack also ships YogurtPreviewImageBridgeOutput with an identical schema - the "(Output)" variant is the pack's convention for the end-of-branch twin. Same inputs, same outputs, either one works mid-graph. Don't overthink the naming; grab whichever reads better in your workflow.