Noise Layer [Replace] ππ π
Overriding noise in a region instead of blending it
- prev_noise_layers
- mask_optional
- NOISE_LAYERS
Noise Layers are the pack's custom noise-scheduling system - the README's phrase for it is "custom noise scheduling via Noise Types, Noise Layers, and seed_override/seed_offset/batch_offset." Chain several together and you're building a deliberate noise recipe instead of accepting one flat draw from the sampler. Of the three noise-layer nodes in this pack, this is the simplest: no weighting, no blending math, just "this layer's noise fully replaces whatever was there before, within its region."
How it compares to its siblings
Add Weighted and Normalized Sum both blend a new noise layer in on top of previous ones, at either an open-ended weight or an automatically-balanced proportion. This node skips blending entirely - wherever it applies (globally, or within a mask_optional region), its noise simply overrides whatever the previous layer in the chain had there. That makes it the right choice when you want a clean override rather than a mix: forcing a specific region of the frame to use a different noise type or seed behavior than the rest of the clip, without the previous layer's contribution leaking through underneath it.
The inputs and outputs that matter
Required: batch_offset (default 0) - shifts which frame this layer's pattern starts from, noise_type - the noise-generation strategy for this layer, seed_gen_override - how this layer derives its seed independent of the base sampling seed, and seed_offset (default 0).
Optional: prev_noise_layers (NOISE_LAYERS) to chain this on top of an earlier layer (which it then overrides, rather than blends with), mask_optional (MASK) to restrict the override to a region instead of the whole frame, and seed_override (default 0) for a fully explicit seed rather than an offset.
Output is a single NOISE_LAYERS, which only fits Sample Settings' noise_layers input.
How to install it
ComfyUI Manager: search AnimateDiff Evolved, confirm the author is Kosinkadink, install, restart. Manually:
cd ComfyUI/custom_nodes
git clone https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved
then restart. No model download tied to this node - pure noise configuration on top of whatever motion setup you already have working.
Common issues & troubleshooting
Added this layer and the previous one's noise still seems to show through. Check that prev_noise_layers is actually chained in correctly and that mask_optional, if set, covers the region you expect - a mismatched or missing mask means the override isn't happening where you think it is.
Expected the whole frame to change but only part did. That's mask_optional at work - leave it unconnected for a full-frame override, or double-check the mask's coverage if you only wanted part of the frame affected but got the whole thing (or vice versa).
You actually wanted a blend, not a full override. Wrong node for that - this one is override-only by design. For a controlled blend with previous layers, use Add Weighted if you want manual control over the mix ratio, or Normalized Sum if you'd rather it stay automatically balanced without extra tuning.
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| batch_offset | INT | 00β9007199254740991 | β |
| noise_type | COMBO | 5 options: default, constant, empty, repeated_context, FreeNoise | |
| seed_gen_override | COMBO | 5 options: use existing, comfy, comfy [gpu], auto1111, auto1111 [gpu] | |
| seed_offset | INT | 0-9007199254740991β9007199254740991 | β |
| prev_noise_layersopt | NOISE_LAYERS | β | |
| mask_optionalopt | MASK | β | |
| seed_overrideopt | INT | 00β18446744073709550000 | β |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| NOISE_LAYERS | NOISE_LAYERS | β |