Image Power Noise

Not to be confused with the noise your sampler adds during generation - this is a straight-up noise image generator, built on power-law noise the way audio people talk about white noise, pink noise, and the rest of that family. Feed it nothing but settings and it hands you an image of pure procedural texture: static, grain, colored noise fields. Useful as an init image to blend into a diffusion pass, a texture layer, or raw material for further filtering.

How it works

The README documents this one in more detail than most, so these are real, not guessed: you control frequency (how fine or coarse the noise texture is - Fourier terms, basically: high frequency reads as fine detail, low frequency as broad blotches), attenuation (how strong the noise is - how far values swing from the mean, weak and smooth versus strong and busy), and noise_type, a choice between white, grey, pink, green, and blue noise. Each of those colors describes a different frequency distribution - white is flat across all frequencies, pink and blue skew toward low or high frequencies respectively - the same vocabulary used for noise coloring in audio, applied here to a 2D image field instead of a waveform.

What comes out and where it goes

The output is a plain IMAGE - a generated noise texture at whatever size you set, ready to feed anywhere an image socket takes input. Common destinations: blending it into a latent or image as an init source (WAS also has an "Image to Noise" node for the reverse direction, converting a real image into noise for exactly this kind of blending), layering it as a texture in a compositing chain, or just using it as raw grain you dial in over a finished image.

How to install it

ComfyUI Manager: search "WAS Node Suite," install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui

then, from inside that folder, install requirements - path/to/ComfyUI/python_embeded/python.exe -s -m pip install -r requirements.txt on portable, pip install -r requirements.txt on system Python. Restart ComfyUI; look under WAS Suite.

Common issues & troubleshooting

Procedural noise generation in WAS Node Suite leans on numba for speed (it JIT-compiles the noise math rather than looping over pixels in plain Python), and numba is a genuinely common source of import failures for this pack - not a rare edge case. People hit ModuleNotFoundError: No module named 'numba' when they installed dependencies with a system Python instead of the embedded interpreter (fixed the same way every time: rerun path/to/ComfyUI/python_embeded/python.exe -s -m pip install -r requirements.txt from inside the WAS folder, using the embedded Python specifically), and separately, numba pins a maximum NumPy version - a newer pack that's pulled NumPy 2.x onto your system can trigger ImportError: Numba needs NumPy 2.0 or less. Got NumPy 2.x and take WAS's import down with it. If Image Power Noise (or the whole suite) won't load, numba/NumPy is one of the first places to look, not a rare theory.

Beyond that, the pack-wide caveat still applies: WAS Node Suite has had no active development since being marked retired in December 2023, so a version clash like this doesn't get an upstream fix - you're managing the dependency pin yourself if it comes up.