Nodes/ComfyUI-RMBG/Image Preview (RMBG) 🖼️
ComfyUI Node Runs on cloud

Image Preview (RMBG) 🖼️

See an image mid-graph, and pass it through

By 1038lab·Created 2 years ago·Updated 8 days ago· 2,051
Image Preview (RMBG) 🖼️
  • image
  • IMAGE

A preview window you can drop anywhere in a graph to see what an image looks like at that point. That's it, and that's genuinely useful - when a workflow isn't producing what you want, the fastest way to find the broken link is to eyeball the image at each stage. This is the RMBG pack's version of that, tuned to fit its own image and mask tools.

Every serious ComfyUI graph is dotted with preview nodes during debugging. They're your print statements. Is the cutout clean before compositing? Preview it. Did the resize do what you thought? Preview it. You pull them out (or ignore them) once things work, but while you're building, they're how you see inside the pipeline instead of only judging the final save.

How it works

It's an output node - it renders the connected image right there in the graph, in the node's body, when the workflow runs. Being an output node matters: ComfyUI will actually execute the branch feeding it, so a preview at the end of a side-branch forces that branch to run even if nothing else consumes it. It also passes the image straight through on its IMAGE output, so you can keep the chain going instead of dead-ending at the preview.

The input and output that matter

  • image (required) - the image to display. Connect the output of whatever stage you want to inspect.

Output: IMAGE, the same image, passed through so you can chain further. That pass-through is the small nicety over a pure sink - you can insert this inline without breaking the flow.

How to install it

  • ComfyUI Manager: search Comfyui-RMBG, install, restart.
  • Manual: cd ComfyUI/custom_nodes && git clone https://github.com/1038lab/ComfyUI-RMBG, then pip install -r requirements.txt, and restart.

No model - it's a viewer.

Common issues

The classic one, and it's called out in the pack's own README: a preview that errors with "Required input is missing: images." That means the node upstream of the preview didn't actually produce an image - either it's not connected, or it failed to run. The preview isn't broken; it's telling you the thing feeding it is. Check that the upstream node ran successfully and its output is wired in. Beyond that, previews are cheap but not free - a graph stuffed with them re-renders each on every run, so once a section works, disconnect the previews you don't need. And remember this shows an IMAGE; for a MASK, use the Mask Preview node instead, since a raw mask won't display sensibly through an image viewer.

Category🧪AILab/🖼️IMAGE

Inputs (1)

NameTypeDefaultDescription
imageIMAGE

Outputs (1)

NameTypeDescription
IMAGEIMAGE