ComfyUI-Sigmax

ComfyUI-Sigmax provides model-aware sigma schedules for ComfyUI, with verified Krea 2, Z-Image, FLUX.1-schnell, Qwen Image, SD3, AuraFlow v0.2, Lumina-Image 2.0, HunyuanImage 2.1, MiniMax H3 Base (qualification slice), Anima, Wan 2.1/2.2, and LTX profiles plus editing tools.

Features

• Explicit model and variant selection with no silent generic fallback.
• Verified Krea 2, Z-Image, FLUX.1-schnell, Qwen Image, SD3, AuraFlow v0.2, Lumina-Image 2.0, HunyuanImage 2.1, MiniMax H3 Base qualification, Anima, Wan 2.1/2.2, and LTX recipes.
• Schedule slicing, concatenation, resampling, inspection, and comparison.
• Experimental Krea 2 CONDITIONING tap rebalancing for explicitly selected RAW or Turbo workflows, with fixed RMS preservation and no scheduler/model patching.
• Checkpoint header inspection without loading model weights.
• Versioned schedule information and fingerprints for reproducible workflows.
• No mandatory third-party runtime dependencies beyond the libraries provided by ComfyUI.

Sigmax builds SIGMAS for ComfyUI custom sampling. It does not replace the sampler and does not download models.

Installation

ComfyUI Manager

Search for ComfyUI-Sigmax in ComfyUI Manager, install it, and restart ComfyUI. If it is not available in Manager, use the Git installation below.

Git

Stop ComfyUI, open a terminal in ComfyUI/custom_nodes, and run:

git clone https://github.com/rookiestar28/ComfyUI-Sigmax comfyui-sigmax

Restart ComfyUI. The expected package entry point is:

ComfyUI/custom_nodes/comfyui-sigmax/__init__.py

Python 3.10 or newer and ComfyUI 0.29.0 or newer are required. Do not install the repository's development or parity dependencies into ComfyUI just to use the nodes.

Use in ComfyUI

Search the node menu for Sigmax. The package registers 24 namespaced nodes.

Build a model schedule

Model	Node	Recommended settings
Krea 2 Turbo	Sigmax.Krea2SigmaScheduler	Turbo, 8 steps, Euler, CFG 1.0
Krea 2 RAW	Sigmax.Krea2SigmaScheduler	RAW, 52 steps with CFG 4.5, or 28 steps with CFG 5.5
Krea 2 RAW + RAW-to-Turbo LoRA	Sigmax.Krea2SigmaScheduler	Experimental RAW/Turbo mu; user-selected steps; community starting point: 12 steps, Euler, CFG 1.0
Z-Image Base	Sigmax.ZImageSigmaScheduler	Base, 28-50 steps, default 50, CFG 4.0
Z-Image Turbo	Sigmax.ZImageSigmaScheduler	Turbo, 8 steps, CFG 1.0
FLUX.1-schnell	Sigmax.Flux1SchnellSigmaScheduler	1-4 steps, default 4, CFG 1.0
Original Qwen Image	Sigmax.QwenImageSigmaScheduler	Comfy Fixed, 50 steps, or Diffusers Dynamic with explicit image_seq_len; host true CFG 4.0
Original Stable Diffusion 3	Sigmax.SD3SigmaScheduler	Select Publisher Reference (1.0) at 50 steps or Comfy/Diffusers Fixed (3.0) at 28 steps; source mode is required
Original AuraFlow v0.2	Sigmax.AuraFlowSigmaScheduler	Official Fixed (1.73), 50 steps, CFG 3.5; source mode is explicit
Lumina-Image 2.0	Sigmax.Lumina2SigmaScheduler	Official Fixed (6.0), 50 steps, CFG 4.0; source mode is explicit
HunyuanImage 2.1	Sigmax.HunyuanImage21SigmaScheduler	Base (5.0), 50 steps, CFG 3.5, or Distilled (4.0), 8 steps, CFG 3.25; variant is explicit
MiniMax H3 Base	Sigmax.MiniMaxH3SigmaScheduler	Select H3 Base FL2VA or H3 Base Ref2VA; endpoint-inclusive video sigmas, 20 grid points by default; audio remapping remains model-owned
Anima Base v1.0 / Aesthetic / Turbo	Sigmax.AnimaSigmaScheduler	Base, 30-50 steps, default 50, CFG 4.5; Aesthetic uses the same recipe; Turbo, 8-12 steps, CFG 1.0; fixed shift 3.0
Wan 2.1 T2V / I2V	Sigmax.WanSigmaScheduler	Select generation, task, source, and resolution explicitly; official T2V 50 steps (5.0), official I2V 480P/720P 40 steps (3.0/5.0)
Wan 2.2 TI2V 5B / A14B T2V/I2V	Sigmax.WanSigmaScheduler	Native or Diffusers-reference source lanes; TI2V 5B uses 5.0; A14B T2V/I2V use 12.0/5.0 with caller-owned boundary metadata
LTXV 0.9.8 / LTX-2 19B / LTX-2.3 22B	Sigmax.LTXSigmaScheduler	Dev adaptive token shift (20/40/30 default steps) or explicit LTX-2/LTX-2.3 distilled Stage 1/2 vectors; generation and stage are explicit

For normal use:

1. Add the scheduler node for the selected model.
2. Select the exact variant and keep strict_official enabled.
3. For Krea 2, enter the actual output width and height.
4. Connect the node's SIGMAS output directly to a custom-sampling path that accepts external sigmas.
5. Do not pass the result through another scheduler or apply another time shift.
6. Read schedule_info when checking the selected recipe, dimensions, step range, or warnings.

The Qwen Image node covers the original text-to-image family only: Comfy Fixed mirrors 1.15, while Diffusers Dynamic requires image_seq_len; later variants and image-quality parity are out of scope.

The SD3 node covers only the original Stability AI SD3 Medium text-to-image schedule. Its two source-qualified modes preserve the publisher 1.0 versus pinned ComfyUI/Diffusers 3.0 conflict; SD3.5, Turbo, ControlNet, model execution, and image quality are outside this support claim.

The AuraFlow node covers only original fal AuraFlow v0.2 with fixed ratio 1.73 and 50 steps; other versions, finetunes, dynamic shifts, execution, and image quality are outside scope.

The Lumina-Image 2.0 node covers only original Alpha-VLLM text-to-image with fixed ratio 6.0 and 50 steps; video, mGPT, editing paths, dynamic shifts, execution, and image quality are outside scope.

The HunyuanImage 2.1 node constructs schedule-only Base and Distilled direct-ratio paths (5.0/4.0). Base is the pinned ComfyUI-compatible lane; Distilled remains publisher-schedule-only until a native host path is qualified. Refiner, encoders, conditioning, weights, and image quality are outside scope.

The MiniMax H3 node is a narrow Base qualification slice: FL2VA/Ref2VA are explicit, only the Diffusers endpoint-inclusive video lane (shift 12.0) is exposed, and native simple plus model-owned audio mapping (shift 3.0) remain separate. The model-free workflow helper can preflight the generated graph against a caller-supplied ComfyUI /object_info schema, including Ref2VA autogrow inputs, without loading weights or submitting a prompt. Context-IR, Regenerate-2K, sparse attention, hosted/API behavior, weights, samplers, and quality claims are outside scope.

The Wan node constructs schedule-only unit-flow sigmas for the released 2.1/2.2 matrix; 2.1 I2V requires 480P or 720P, and 2.2 A14B boundaries are caller-owned metadata (never expert routing). Diffusers-reference lanes describe scheduler math only; video execution, weights, and quality parity are outside scope.

The LTX node constructs schedule-only unit-flow sigmas for pinned LTXV 0.9.8, LTX-2 19B, and LTX-2.3 22B lanes. Dev mode derives one token-count shift; distilled modes use immutable publisher vectors. It does not load video weights, run encoders, or claim video-quality parity.

Sigmax.ModelAwareSigmaScheduler can inspect a connected Krea 2 model, but weak evidence may identify only the family; if Auto reports ambiguity, select Turbo or RAW explicitly.

For experimental conditioning, connect Krea 2 CONDITIONING to Sigmax.Krea2ConditioningRebalance, select RAW/Turbo and a profile; it changes only the primary tensor, preserves metadata, and makes no prompt-adherence or image-quality claim. Strength 0 is identity.

For a RAW checkpoint with a compatible RAW-to-Turbo model-difference LoRA, choose LoRA Experimental (RAW mu) or LoRA Experimental (Turbo mu) and set the desired steps. The node automatically forces strict_official off and disables that widget for either Experimental selection. The former derives mu from width/height; the latter fixes mu = 1.15. This path is experimental: the scheduler does not load or scale the LoRA, enforce CFG/sampler settings, or claim that 12 steps is official. Apply the LoRA to RAW only; do not stack equivalent RAW-to-Turbo LoRAs or apply one to the Turbo checkpoint.

Inspect or modify a schedule

• Sigmax.ProfileInspector and Sigmax.ScheduleInspector show profile and schedule details.
• Sigmax.ScheduleComparison compares two schedules.
• Sigmax.ScheduleSlice, Sigmax.ScheduleConcatenate, and Sigmax.ScheduleResample create a modified schedule while retaining traceable schedule information.
• Sigmax.CheckpointEvidenceInspector reads only the bounded header of a local .safetensors file. Its suggested variant is advisory, not confirmation.
• Sigmax.AdvancedFlowMatchScheduler is an experimental schedule constructor, not a universal model profile or sampler.

Update or remove

For a Git installation:

cd ComfyUI/custom_nodes/comfyui-sigmax
git pull --ff-only

Restart ComfyUI after updating. Back up important workflows before changing versions. To remove Sigmax, stop ComfyUI, remove or move only the comfyui-sigmax directory, and restart ComfyUI.

Troubleshooting

Problem	Resolution
Nodes do not appear	Restart ComfyUI, inspect the startup log, and confirm there is only one direct comfyui-sigmax folder under custom_nodes.
Import fails	Confirm Python 3.10+ and ComfyUI 0.29.0+, then remove duplicate or nested installations.
Auto rejects Krea 2	Select Turbo or RAW explicitly. Do not rely on the filename alone.
The schedule appears shifted twice	Remove the second scheduler or model time shift; use the Sigmax SIGMAS output once.
Krea 2 RAW dimensions differ	Check the requested and effective dimensions in schedule_info; RAW dimensions are normalized to the supported grid.
A newer ComfyUI release behaves differently	Review the supported boundary in Compatibility.

ComfyUI custom nodes execute Python code inside the host process. Install reviewed sources only. Model weights are not included, and current validation does not claim real-model GPU image quality, stochastic sampling, resume behavior, or arbitrary-model compatibility.