ComfyUI-CFG-Ctrl (SMC-CFG)

Implements CFG-Ctrl / SMC-CFG (Sliding Mode Control CFG) as a ComfyUI GUIDER.

References:

• Paper: "CFG-Ctrl: Control-Based Classifier-Free Diffusion Guidance"
• Official repo: hanyang-21/CFG-Ctrl

Install

Place this folder into:

ComfyUI/custom_nodes/ComfyUI-CFG-Ctrl/

Restart ComfyUI.

How to use

This node outputs a GUIDER. Use it with ComfyUI custom sampling nodes that accept GUIDER input (e.g. SamplerCustom / SamplerCustomAdvanced workflows).

Steps:

1. Create your model + conditioning (positive/negative) as usual.
2. Add CFG-Ctrl / SMC-CFG Guider and connect model/positive/negative.
3. Feed the GUIDER into your custom sampler node.

Parameters

• cfg: Standard CFG scale. SMC-CFG stabilizes behavior at higher cfg.

• smc_lambda: Sliding-surface decay shaping. Paper commonly uses values around 4–7 (often 6).

• smc_k: Switching gain. Controls how hard the controller pulls toward the sliding manifold. Too low => weak effect. Too high => chattering/instability.

• no_cfg_warmup_steps: First N steps: use pure conditional prediction (no CFG). Helps avoid early noise blowups.

• switch_mode:

  • sign (paper/repo): exact switching
  • sat / tanh: boundary layer to reduce chattering
  • vector_norm: normalizes by ||s|| (per-batch), smoother but slightly more compute

• boundary_epsilon: Only used for sat/tanh/vector_norm. Start with 1e-3 to 1e-2 if you see chattering.

Practical defaults (starting points)

• FLUX-like flow models: cfg 3–6, lambda 5–7, k 0.2–0.8, warmup 1–3

• If you increase cfg above your usual: keep lambda ~6, increase k slightly, add warmup 1–3, or set switch_mode=tanh with epsilon ~1e-3..1e-2.