Runware Bria Rmbg V2 0 Open

If you've already met Runware_bria_rmbg_v2_0, this node will look like a deja vu. Same job - remove the background, return a cutout - and nearly the same shape: one required image input, one IMAGE output, plus the usual output format/quality and TTL options. The difference is in the plumbing: this one is the open variant of the RMBG v2.0 model, and it's served under Runware's own catalog entry (AIR runware:110@1) rather than Bria's (bria:2@1). That's why it's grouped under Runware/Image/runware in the menu instead of the Bria folder - it's the same model family, released through a different, more permissive route.

It also has one fewer knob than its Bria-branded sibling: no preserveAlpha toggle. The Bria node lets you explicitly keep an alpha channel when your source has one; this one just does the job. For most uses that's not a loss - you're here to strip a background, not to babysit channel semantics - but if you're feeding in images that already have transparency and need it preserved, reach for the other node.

What you'd actually touch

• image - required. Your subject.
• outputFormat - the important one, same as always: switch from the default JPG to PNG if you want a real cutout with transparency instead of something flattened onto a background.
• outputQuality, ttl - defaults are fine.

Output is a single IMAGE socket. Into Save Image or a composite workflow.

Install

The whole pack installs once, same as every node here:

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

Restart, add your Runware API key (Settings → Runware API key, or RUNWARE_API_KEY), and you're set. No checkpoints download; the cutout happens on Runware's GPUs and the result comes back as a native IMAGE tensor, with cost printed on the node after the run.

The honest take

Two nearly identical background-removal nodes in one pack is the kind of thing that makes you wonder which one you're supposed to use. The answer: whichever shows up in the workflow you loaded. For a beginner, pick by cost transparency and don't overthink it - both do the same job, this one is the open-weights route with a lighter settings surface, and the other gives you alpha preservation. If you're a heavy user, the per-cutout pricing is the real constraint either way; local BiRefNet in core ComfyUI is free and excellent if the cloud bill starts to matter.