Sampler Settings (Simple)

Every shared ComfyUI workflow ends up with a KSampler buried under a pile of wires, and every time you tweak steps you have to dig through the spaghetti to find it. Sampler Settings (Simple) exists so you don't have to. It's one node that collects your sampler, scheduler, steps, CFG and guidance into a single pipe, plus - the bit most settings nodes skip - an upscale model you can tag along for the ride.

What it actually does

Under the hood it's just a tidy tuple. The node packs your choices into a plain Python list - [sampler, scheduler, steps, cfg, flux_guidance, upscale_model, scale_by] - and hands it out on a single pipe output. To get those values back out as separate wires you pair it with Pipe Out Sampler Settings (Simple), also in this pack, which unpacks the list into the individual inputs of a KSampler. Think of it as a "settings bundle" convention: the pipe type isn't a ComfyUI-native thing, so it only works with the matching Pipe Out node. If you drop a random pipe consumer on it, it'll complain.

The inputs that matter

• Sampler - 44 options, straight from ComfyUI's KSampler.SAMPLERS, so everything you're used to is there: euler, euler_ancestral, dpm_2, heun, and the CFG++ variants if you've installed that sampler set.
• Scheduler - 16 of them, and here's a nice touch: it's ComfyUI's scheduler list plus the extras (AYS SDXL, AYS SD1, AYS SVD, OSS FLUX, OSS Wan), so you don't need a separate scheduler-extras pack to reach the AYS schedules.
• Steps (default 20) - the number you'll actually tune constantly.
• CFG (default 3.5) and Flux_Guidance (default 3.5) - heads up, that 3.5 default is Flux-era thinking. For SD 1.5 or SDXL you'll want CFG around 6–9, and on Flux the CFG field is basically ignored in favor of Flux_Guidance. Don't blindly run the defaults on an old checkpoint and wonder why it looks flat.
• Upscale_Model (optional) and Scale_By (optional, default 0.5) - pick any model in your ComfyUI/models/upscale_models folder and it gets loaded and passed down the pipe too, so a hires-fix pass can grab it without a separate Upscale Model loader node. It loads through the spandrel library, which is already in your ComfyUI, and raises an error if you hand it something that isn't a single-image upscaler.

Installing it

This is part of ComfyUI-RvTools_v2. The easiest path is ComfyUI Manager → Install Custom Nodes → search "ComfyUI-RvTools_v2". Or drop it in manually:

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

The pack's requirements.txt lists torch, numpy, Pillow, opencv-python, pilgram, pynvml and piexif - most of that ships with ComfyUI already; you only realistically need the extras if you use the video or save-image nodes.

Gotchas

Honest warning before you build a workflow on this: the README says the pack is no longer maintained and has been superseded by ComfyUI_Eclipse, which carries most of these nodes forward. The nodes still work fine and show up in a lot of shared workflows, but if you're starting fresh today, check Eclipse first. Also note the display name ends in [RvTools] - the author renamed everything when he fixed clashes with the original RvTools repo, so don't be surprised if a downloaded workflow references a slightly different name and Manager can't auto-find it. That's usually a case of the workflow using the old repo's node name; swap in the [RvTools] version and it'll wire up.