πΈ Multi Distortion
Twelve Ways to Make It Ugly β the Multi-Stage Distortion Rack
- audio
- distorted_audio
Distortion is the simplest effect in audio - clip a signal and it gets ugly - but there are a dozen flavors of ugly. MultiDistortion gives you twelve of them, and lets you chain up to four stages of the same type back to back. It's the pack's dirt pedal, and for a harsh-noise workflow it's often the last processor before AudioSave.
What it is
Pick a distortion_type and the node applies that algorithm to your audio, with drive (0β10) as the amount of dirt and output_gain to set the level after. The twelve types cover the usual suspects and then some:
- Tube, transistor, diode - analog-style waveshaping flavors. Tube is warm, transistor is edgy, diode is crunchy.
- Digital, bitcrush - the aliased, retro-arcade sound. Bitcrush is your chiptune-to-8-bit-doom option.
- Waveshaper, foldback - foldback is the special one: instead of clipping, the waveform folds back on itself, producing bright, aliasing-rich textures that never hard-clip.
- Ring mod - amplitude modulation, metallic bell tones.
- Fuzz, overdrive, chaos, destruction - the escalation ladder, ending in the two modes that exist purely to wreck things.
How it works
stages (1β4) is the clever bit: the signal runs through the distortion multiple times in series. One stage of tube is gentle warmth; four stages is a molten core. asymmetry (β1 to +1) unbalances the positive and negative halves of the wave - that's what generates even harmonics, the sound that makes distortion feel "musical" rather than just harsh. wet_dry_mix blends the clean and processed signals (default 1.0, i.e., fully wet), and the optional chaos_amount adds chaotic modulation on top of the chaos and destruction modes.
The drive control goes up to 10, which is a lot - combined with output_gain up to 2.0, this is the easiest node in the pack to blow way past clipping. That's sometimes exactly the aesthetic, but it's worth knowing the amplitude knob is your safety net.
The inputs that matter
distortion_type- the character. Start withfoldback(musical and safe) ortube, and work up todestructionwhen you're ready.drive- the amount. 0β2 is subtle to medium; 5+ is a wall of grit.stages- the multiplier. Going from 1 to 4 stages is a bigger change than most people expect.
Output: distorted_audio.
Install
The standard pack install:
cd ComfyUI/custom_nodes
git clone https://github.com/eg0pr0xy/comfyui_noisegen.git
cd comfyui_noisegen
pip install -r requirements.txt
(Windows embedded Python: ComfyUI\python_embeded\python.exe -m pip install -r requirements.txt.) Or ComfyUI Manager β search "NoiseGen" / eg0pr0xy. No model files, no keys.
Gotchas
Same pack rule as everywhere: the pip install is mandatory - scipy is imported at load, and without it the whole pack fails to register. Watch your levels with multiple stages and high drive; four stages of destruction at drive 8 is not a sound, it's a fire alarm, and it will clip into a brick wall of digital nastiness. If you want to keep any dynamic range, run high stages with a lower drive instead of one stage at max drive - you get the density without the square-wave flattening.
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | Input audio to process through multi-stage distortion | |
| distortion_type | COMBO | tube | 12 options: tube, transistor, diode, digital, bitcrush, waveshaper, +6 |
| drive | FLOAT | 0.500β10 | β |
| output_gain | FLOAT | 0.700β2 | β |
| wet_dry_mix | FLOAT | 1.000β1 | β |
| stages | INT | 11β4 | β |
| asymmetry | FLOAT | 0.00-1β1 | β |
| amplitude | FLOAT | 0.800β2 | β |
| chaos_amountopt | FLOAT | 0.500β1 | β |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| distorted_audio | AUDIO | β |