Flow Control Parameters

Flow Control Parameters is a small unpacker node. It takes the Searge workflow's processed parameters bundle and pulls out a handful of integer selectors - the numbers that decide which branch of the graph actually runs. It's the node that translates your dropdown choices ("inpainting," "style focus") into the plain integers that drive the switches and muxers scattered through the workflow.

It's part of SeargeSDXL, the SDXL base-plus-refiner workflow pack Searge released at SDXL 1.0's launch in 2023. The Searge/_deprecated_/UI/Generated category tells you it's a v3-era internal node - the "Generated" part meaning it was one of the auto-generated UI-output nodes in that architecture. The v4.x rewrite handles this differently through its data stream.

How it works

The Searge graph doesn't have three separate pipelines for txt2img, img2img, and inpainting - it has one graph with switches that route the right path. Those switches are driven by integers, and this node is where the integers come from. It reads the PARAMETERS bundle (which encodes your mode and style choices) and emits selector values like an operation selector and prompt-style selectors. Downstream, a node like the 3-way latent muxer reads that operation selector to decide which latent path flows through.

So it's the bridge between "human picked a dropdown" and "graph takes the correct branch." You never set anything on it - it just surfaces values that were decided elsewhere.

The inputs and outputs

• parameters (PARAMETERS) - the processed bundle, in. Required.
• parameters (PARAMETERS) - passed straight through, so you can keep chaining output nodes off the same bundle.
• operation_selector (INT) - which operation mode (drives the mode switches/muxers).
• prompt_style_selector and prompt_style_group (INT) - which prompt-combination recipe and group.

Those integers aren't things you read or tune; they're wiring signals for other nodes.

How to install it

ComfyUI Manager: search SeargeSDXL, install, restart. Manual: cd ComfyUI/custom_nodes && git clone https://github.com/SeargeDP/SeargeSDXL.git, then restart. Manual installs need python -m pip install opencv-python in ComfyUI's Python environment first, per the README, or the nodes won't load. This node needs no models.

Common issues

If the workflow keeps running the wrong mode - you set inpainting but it behaves like txt2img - the trail runs through here: this node emits the operation selector that the muxers obey, so a stale or mismatched parameters bundle shows up as "wrong branch executed." The fix is almost never this node itself; it's making sure your mode is actually set (in Misc Parameters / Operating Mode) and that your workflow JSON matches your node version.

That last point is the recurring Searge lesson: load the latest workflow file with the latest custom nodes. And the honest context - this is a deprecated internal node you don't place by hand, inside a workflow built around SDXL's refiner. It works, but it's the machinery of a setup the wider community has largely moved past in favor of single fine-tuned checkpoints.