Misc Parameters

Misc Parameters is the grab-bag node, but don't let the name fool you - two of its three settings are among the most important in the whole Searge workflow: which operation mode you're in (txt2img / img2img / inpainting) and which prompt style controls how your five prompt boxes get combined. The third, LoRA strength, is exactly what it sounds like.

It's from SeargeSDXL, the SDXL base-and-refiner workflow pack Searge released at SDXL 1.0's launch in 2023. The node is in Searge/_deprecated_/UI/Inputs - the v4.x rewrite folded mode selection into the newer Operating Mode node on the data-stream bus - but this is the v3-era equivalent, and understanding the prompt styles here explains a lot about how Searge behaves.

How it works

The operation_mode sets whether you're generating from scratch, from a source image, or inpainting a masked region - the same three-way choice the whole workflow pivots on. The interesting one is prompt_style: it picks the recipe for combining your main, secondary, and style prompts. "simple" just uses them straightforwardly; "subject focus" and "style focus" bias one way or the other; "weighted" and "overlay" mix them mathematically; the "X - Y" styles apply one to the base and another to the refiner. This is the setting that decides whether your style prompt even gets used, which is why "my style prompt does nothing" almost always traces back here.

Like the rest of the input chain, it drops its choices into the PARAMETER_INPUTS bundle rather than acting directly.

The inputs that matter

• operation_mode (enum: text to image, image to image, inpainting) - the workflow mode.
• prompt_style (enum, 11 options: simple, 3 prompts G+L-N, subject focus, style focus, weighted, overlay, subject - style, style - subject, style only, weighted - overlay, overlay - weighted) - how the prompts combine. The big creative lever.
• lora_strength (FLOAT, default 0.2, range -10 to 10) - strength of the LoRA selected in the Model Names node.
• inputs in/out (PARAMETER_INPUTS) - the chain wire.

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, or the README warns the pack won't load. Generation needs the pack's models (SDXL base + refiner, fp16-fix VAE); the LoRA strength here only matters if you've selected a LoRA and installed it.

Common issues

If your style prompt isn't landing, try a prompt style that emphasizes it - style focus, subject - style, or style only - before you touch anything else. simple treats everything plainly, which is often why a carefully written style prompt vanishes.

The default lora_strength of 0.2 is low; if your LoRA barely shows, that's usually why - SDXL LoRAs often want something closer to 0.6–1.0. And the standing rule: latest workflow JSON with the latest nodes. Finally, the honest framing - this node's whole world (SDXL, refiner, elaborate prompt recipes) predates the community's shift to single fine-tuned checkpoints that need far less of this ceremony. It still works; just know what era you're in.