xmarre-ComfyUI-ScaleLockedResidualDiffusion/js/docs/ScaleLockedResidualKSampler/en.md

Scale-Locked Residual KSampler

All-in-one sampler node for large-resolution generation where the model's native-resolution generations are much better than its direct high-resolution generations.

Core idea

A native / low-resolution planner branch is sampled first. Its denoised trajectory is cached. The final high-resolution branch then uses a custom guider that keeps the base model's high-frequency residual detail but continuously pulls the low-frequency denoised structure back toward the planner trajectory.

Inputs

• model, positive, negative, latent_image: standard Comfy sampler inputs
• seed, steps, cfg, sampler_name, scheduler, denoise: standard sampler controls
• target_megapixels: planner resolution in pixel-space MP
• lock_strength: overall lock multiplier
• lock_strength_start: early-step lock amount
• lock_strength_end: late-step lock amount
• lock_schedule: linear / cosine / flat
• coarse_cutoff: strongest coarse-band resolution fraction
• mid_band_cutoff: second, looser mid-band resolution fraction
• mid_band_strength: relative strength of the mid-band lock
• nested_noise_strength: amount of extra high-frequency detail noise
• pin_anchors: pinned-memory staging for planner anchors when possible
• sampler_guard: warn / error / off handling for samplers outside the conservative safe set
• lock_mask (optional): spatial mask to focus the lock on anatomy-critical regions

Outputs

• output: final high-res latent
• lowres_planner: final low-res planner latent
• denoised_output: final denoised x0 high-res latent when available

Notes

• Lower coarse_cutoff = stronger global structure control.
• Lower mid_band_cutoff and higher mid_band_strength = tighter control over medium-scale body/shape structure.
• Higher nested_noise_strength = more detail freedom, but also more chance of drift.
• A lock_mask is recommended for body-heavy and anatomy-sensitive generations.