🦙 Winamp Viz

If you were on the internet in the early 2000s, you already know the vibe. Winamp Viz takes an AUDIO input and renders classic media-player visualizations - the oscilloscope, the spectrum analyzer, the MilkDrop-style bars - as a batch of images, so you can composite them under your own UI chrome, use them as a music-video background, or just watch a waveform dance while your video renders. The name is a promise, and it mostly delivers.

It's one of the more charming nodes in DJZ-Nodes, and it's a reminder that the pack isn't all serious film effects - sometimes it's just fun.

How it works

The node analyzes the audio with a sliding-window FFT, splitting the spectrum into bass/mids/highs bands, and uses those values to drive the visualization math each frame. The viz_type dropdown picks the renderer, and there are eight of them:

• oscilloscope - the classic waveform trace.
• spectrum - the frequency analyzer.
• particle_storm - a particle system whose emission reacts to the beat.
• plasma_wave - wavy plasma distortion driven by the signal.
• milkdrop_bars - frequency bars with that unmistakable MilkDrop smoothness.
• circular_wave - the waveform wrapped into a ring.
• butterfly - a butterfly-curve animation (yes, like the MilkDrop preset).
• tunnel_beat - a beat-reactive tunnel you fly through.

Two global dials shape how reactive everything is: sensitivity (0.1–5) amplifies the audio response, and smoothing (0–0.99) eases the visualization's motion - higher smoothing = lazy, liquid transitions; lower = twitchy, immediate. color_scheme (classic, rainbow, fire, matrix) repaints all of them.

The inputs that matter

• audio - the AUDIO input that drives everything.
• width/height (64–4096), fps (1–60) - output resolution and timing.
• max_frames - length cap. Default 0 means "auto": it renders until the audio runs out, which is what you want for a full-length track. Set a number to truncate.
• viz_type - the effect. Start with spectrum and oscilloscope; they're the most legible.
• sensitivity/smoothing - the two dials you'll actually tweak.

Output is images (IMAGE batch) - one frame per video frame of your audio. Feed it to a Video Combine alongside the audio and you've made a music video.

Installing it

Standard DJZ-Nodes install:

cd ComfyUI/custom_nodes
git clone https://github.com/MushroomFleet/DJZ-Nodes
cd DJZ-Nodes
pip install -r requirements.txt

Restart ComfyUI, or search "DJZ-Nodes" in ComfyUI Manager. Needs the pack's audio stack (librosa, soundfile, numpy) - the same heavier requirements as the other audio nodes.

Troubleshooting

• Visualization is dead / barely moving. Check sensitivity - at 0.1 the audio barely registers. Also confirm the audio actually has signal; silence visualizes as silence.
• Frame count doesn't match the song. max_frames at 0 auto-sizes to the audio, but your fps setting determines how many frames that computes to. If the viz ends early, it's usually the fps/frame-count math, not the audio.
• Stutters on long tracks. Big resolutions at 60fps for a 4-minute song is a lot of FFT + rendering. Drop to 512px and it flies.
• "Winamp Viz" name missing from search. Some Manager lists show it under the pack. Just install DJZ-Nodes and all the viz nodes come along.

It's unapologetically a party node. Not something you'll use for serious work, but for music-driven video it's dead simple and reliably fun. In 2026, that still whips the llama's ass.