Nodes/Comfyui-EasyIllustrious/⚙ KSampler Presets
ComfyUI Node

⚙ KSampler Presets

Six curated configs for when you don't want to think about CFG

By regiellis·Created 12 months ago·Updated 4 months ago· 82
⚙ KSampler Presets
  • model
  • positive
  • negative
  • latent_image
  • LATENT
  • model
  • positive
  • negative
seed0
PresetBalanced
Model VersionAuto Detect
Color Safe Modetrue
Resolution Adaptivetrue
Auto CFGtrue
Seed Offset0
Advanced Logsfalse

IllustriousKSamplerPro gives you every dial. This node gives you six named buttons instead - Balanced, High Quality, Fast, Portrait, Landscape, Anime - and hides steps, CFG, sampler, and scheduler behind them entirely. If you're new to Illustrious and don't yet know that its comfortable CFG range sits around 4–7 rather than the vanilla SDXL default of 7–8, this is the node that gets you there without needing to know it yet.

What's actually behind a preset

The pack doesn't publish the literal numbers each preset resolves to - they're baked into the node rather than exposed as read-only fields - so treat them as a sensible starting point rather than something you can reverse-engineer from the UI alone. If you want to see (and then override) the exact steps/CFG/sampler/scheduler a preset is roughly equivalent to, that's what IllustriousKSamplerPro is for: dial in values by hand there once you know what you like, rather than trying to pry the preset's internals out of this node.

Color Safe Mode is worth understanding precisely because of how it interacts with the other toggles: on (the default), the node uses the preset's values close to exactly, with minimal further tweaking. Off, and Resolution Adaptive and Auto CFG are freer to nudge the preset's numbers based on your actual resolution and detected model version. If you picked a preset because you liked a specific look and then it drifted between runs at different resolutions, that's Color Safe Mode being off doing its job - turn it on to pin the preset down.

The inputs and outputs that matter

  • model, seed, positive, negative, latent_image - the standard set, all required.
  • Preset - the whole point of the node. Balanced is the default and the safest starting pick.
  • Model Version - Auto Detect by default; set explicitly if you know you're on a specific EPS/VPred build, since a wrong guess here changes what "correct" CFG even means for a preset.
  • Seed Offset - added to your base seed, useful for generating variations without touching the seed you actually want to keep as your anchor.

Outputs mirror IllustriousKSamplerPro: LATENT plus passthrough model/positive/negative, so downstream nodes can pull wiring from here instead of routing back to the original sources.

Installing it

ComfyUI Manager: search "EasyIllustrious," install, restart. Manually:

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

No downloads - same checkpoint/VAE/CLIP as everything else in the pack.

Common issues

A preset isn't giving the result its name implies. Since the exact numbers aren't documented, don't assume "Portrait" changes your resolution or "Anime" changes your model - none of the schema fields suggest that's what these presets do; they're almost certainly steps/CFG/sampler/scheduler bundles tuned for different priorities (speed vs. quality) rather than composition or subject matter. If a preset's output doesn't match what its name suggests to you, that's a mismatch between expectation and what the node actually controls, not a bug.

Need to debug a specific setting and the presets won't let you. That's the tradeoff by design - switch to IllustriousKSamplerPro for anything you need to see and tune precisely; this node is for speed, not precision.

Results vary more than expected between runs. Turn Color Safe Mode on and check whether Resolution Adaptive/Auto CFG are quietly adjusting things based on your latent size or detected version.

CategoryEasy Illustrious / KSampler

Inputs (12)

NameTypeDefaultDescription
modelMODELUNet diffusion model to sample with (SDXL / Illustrious).
seedINT00–18446744073709550000Base random seed; use different seeds for different outputs.
positiveCONDITIONINGPositive prompt conditioning (what you want to see).
negativeCONDITIONINGNegative prompt conditioning (what to avoid).
latent_imageLATENTStarting latent tensor (shape [B,4,H/8,W/8]).
PresetCOMBOBalancedCurated configuration presets for Illustrious XL.
Model VersionoptCOMBOAuto DetectPick your Illustrious version or let the node auto-detect by resolution.
Color Safe ModeoptBOOLEANtrueWhen ON, use preset values exactly with minimal auto-tweaks.
Resolution AdaptiveoptBOOLEANtrueGently adjusts steps for larger/smaller images.
Auto CFGoptBOOLEANtrueKeeps CFG within a sweet spot for the selected version.
Seed OffsetoptINT00–100000Added to seed to vary results while keeping the base seed.
Advanced LogsoptBOOLEANfalsePrint extra diagnostics to the console.

Outputs (4)

NameTypeDescription
LATENTLATENT
modelMODEL
positiveCONDITIONING
negativeCONDITIONING