Sampler Combo (Compact)

Every KSampler needs four settings wired in: a sampler, a scheduler, steps, and CFG. Stock ComfyUI makes you scatter those across four separate widgets and remember which one does what. Sampler Combo (Compact) collects all four into a single node and hands them to KSampler as one clean set of outputs. It's the "everything in one place" node that keeps your graph tidy without asking you to think about anything new.

What it actually does

This is the compact sibling of the full Sampler Combo node in the same pack - same idea, shorter parameter names (sampler, sched instead of sampler_name, scheduler) and a few tighter ranges. It doesn't change how sampling works; it's a config pass-through. Wire its four outputs into a KSampler's matching inputs and you get one obvious place to tweak everything that controls the denoise.

The genuinely nice bit is under the hood: the node pulls its sampler and scheduler lists straight from comfy.samplers.KSampler at load time. So the dropdown always shows exactly what your ComfyUI build supports - all 44 samplers and 9 schedulers, including the newer _cfg_pp variants. No hardcoded list that goes stale when you update ComfyUI. If something invalid sneaks in, it falls back to euler/normal/20/7 instead of crashing mid-run.

The inputs that matter

You'll actually touch two of these:

• sampler - the denoising algorithm. Default euler, which is a sane starting point for most models. For SDXL-lineage checkpoints, DPM++ 2M Karras is the community's safe default; for flow-matching models (Flux and friends), stick with euler and don't pair it with a Karras schedule.
• sched - the noise schedule. Default normal. Karras concentrates denoising in the middle steps, which helps SD 1.5/SDXL and hurts straight-line flow-matching models.
• steps (1–50) and cfg (1–15) - the compact version keeps these ranges modest. 20 steps and CFG 7.0 are the classic starting points, but remember CFG is architecture knowledge, not taste: guidance-distilled models want CFG at or near 1.

Outputs are named sampler, scheduler, steps, cfg - four wires, straight into KSampler.

Installing it

This ships in the ComfyUI-KikoTools pack, so one install covers every node here:

1. Open ComfyUI Manager, search for "ComfyUI-KikoTools", click Install.
2. Restart ComfyUI, then find it under the ComfyAssets → Samplers menu.

Manual route is the same clone-and-restart dance:

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

Restart ComfyUI. Runtime deps are minimal (nothing beyond PyTorch), so this pack won't drag in half of PyPI.

Common gotchas

The trap here isn't the node, it's the settings. People grab a workflow that says "euler, karras, 30 steps, CFG 7" and paste it onto a Flux or distilled model, then wonder why the image looks overcooked. The compact combo faithfully passes through whatever you choose - it won't stop you from picking a bad schedule for the model. If output quality suddenly tanks after you "just tweaked the sampler," check the sched field first.

One more thing to know: because the dropdowns are built from your running ComfyUI, an update to ComfyUI can add or remove sampler names. If an old workflow complains about a missing sampler, re-pick it from the dropdown - the old string may simply not exist anymore.