Pass Images

Pass Images is the ComfyUI equivalent of a power strip: one input, one output, nothing in between. The whole node is a single IMAGE in and a single IMAGE out, and its only job is to be a stable connection point for an image tensor. It doesn't resize, convert, batch, or flip anything. It literally returns whatever you feed it.

That sounds useless until you've lost an afternoon to a workflow that breaks every time you look at it. ComfyUI lets you drag a wire from an image output to as many inputs as you want, which works fine in a small graph - but the moment you're editing a big one, that source node moves, gets bypassed, or gets replaced, and suddenly six downstream wires are pointing at a hole. A passer breaks that coupling: the wire from your real source terminates at the passer, and everything downstream connects to the passer. You can then bypass the source, swap it, or move it without ripping up the rest of the graph.

This is exactly why r-vage built it. The RvTools README says passers were "created to be placed in front of nodes that had problems getting information from bypassed nodes," and to work inside managed groups - you need at least two nodes to form a group, and a passer is the natural second node that lets you hide inputs and widgets behind one tidy connection. It's the same problem rgthree's Display Anything exists to solve, and it's the quiet infrastructure that keeps big shared workflows load-bearing.

The one input and output that matter

• image (IMAGE) - the tensor you want to route. Required.
• Output IMAGE - the same tensor, unchanged.

There's nothing else. No Purge_VRAM toggle here - that's a v2 thing, and the plain Pass Images node deliberately stays dumb. Feed it a batch and you get the batch back. Feed it a single frame and you get a single frame.

Installing it

The node ships in ComfyUI-RvTools_v2 by r-vage. Easiest path is ComfyUI Manager: search for "ComfyUI-RvTools_v2" and hit install, then restart. Or by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/r-vage/ComfyUI-RvTools_v2

The pack pulls in torch, numpy, Pillow, opencv-python, pilgram, pynvml, and piexif - all of which you almost certainly already have if you run ComfyUI, so there are no model files to download and nothing heavy to babysit.

Gotchas worth knowing

The pack's README opens with a big warning that it's no longer maintained and has been superseded by ComfyUI_Eclipse, which "includes most of the RvTools nodes." That doesn't make this node stop working - it just means fixes land in Eclipse from here on. Also, if you ever loaded a workflow that references RvTools v1 (the original RvTools repo), that repo got pulled from GitHub, which is why those workflows fail to resolve; the v2 pack renamed everything, and this node is part of that rename. Install v2 and the missing-node wall clears.

One honest caveat: if you only have two nodes on screen, a passer is pure ceremony. Reach for it when the graph starts getting crowded, not before - it earns its place as organizational glue, not as a feature.