Nodes/comfyui-photoshop/šŸ”¹modelPass
ComfyUI Node

šŸ”¹modelPass

The MODEL pass-through that keeps your graph tidy

By NimaNzriiĀ·Created 3 years agoĀ·Updated 11 days agoĀ· 1,735
šŸ”¹modelPass
  • model
  • model

modelPass is ClipPass's sibling with different types: it takes a MODEL, and it hands you back the exact same MODEL. The whole function body is return (model,). No sampling, no patching, no hidden behavior. If you're staring at it wondering what dark magic it performs, the answer is that it performs none.

What it's for

This is a layout node, not a feature node. The comfyui-photoshop workflows route a checkpoint through a fairly busy graph - canvas in, mask in, prompts in, and a render going back out to Photoshop. In a graph that busy, the MODEL cable from your checkpoint loader to the sampler is one of the longest and easiest to snag. modelPass gives you a fixed anchor point: run the model into the pass node, then route from there, and you can rearrange half the canvas without tracing which sampler it fed.

It also works as a kind of live label. In a wall of šŸ”¹-prefixed nodes, a pass node is a clean spot to confirm the right model is loaded before anything expensive runs - a cheap sanity check before you waste a queue on a graph wired to the wrong checkpoint.

Using it

One input, one output, both type MODEL:

  • model (MODEL) in → model (MODEL) out.

Wire your checkpoint loader's model output in, wire the pass node out to your sampler. That's the entire setup. If the output looks wrong, the node is innocent - check what's feeding it.

Installing it

It rides along with the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/NimaNzrii/comfyui-photoshop

or ComfyUI Manager → search comfyui-photoshop → install → restart.

If you build your own workflows from scratch, you'll probably never use this. It exists for the pack's shipped graphs, where it earns its keep as an organizing widget. Knowing that saves you the twenty minutes I spent looking for the hidden feature.

Categoryutils

Inputs (1)

NameTypeDefaultDescription
modelMODEL—

Outputs (1)

NameTypeDescription
modelMODEL—