❄️ Procedural Fractal

Procedural Fractal Generator renders classic escape-time fractals - Mandelbrot, Julia, Burning Ship, Newton - as images, computed on the CPU in pure math. No diffusion model, no weights, no VRAM, no downloads. It's one of the pack's LCARS procedural generators, and it does exactly what it says on the label: math in, image out.

Why would you want fractals in ComfyUI? They're a great source of procedural texture and background material for compositing, masking experiments, and dataset augmentation - and they're a surprisingly nice way to test your output pipeline without generating anything with a model. It's also just fun to explore; the Mandelbrot set at increasing zoom never gets old, and it costs you nothing to iterate.

The inputs that matter

• width / height - 64 to 2048 (step 64). Keep in mind this is pure-Python per-pixel math; large sizes at high max_iterations get slow.
• fractal_type - Mandelbrot, Julia, Burning Ship, or Newton. Mandelbrot is the safe start; Julia gives you infinite variety by changing the constant.
• max_iterations - 10 to 500. More iterations = deeper detail but slower. 100 is a good balance; go to 300+ for deep zoom.
• zoom - 0.1 to 10. This zooms the view.
• center_x / center_y - the center of the view in the complex plane (−2 to 2). Mandelbrot's default center of −0.5, 0 shows the classic shape.
• julia_c_real / julia_c_imag - the Julia constant (the c that defines the set), used when fractal_type is Julia. Defaults of −0.7 + 0.27i are a known-interesting point.

Output is the image - an IMAGE tensor ready to wire into anything that takes an image.

The workflow angle

Because it's deterministic, it's a great procedural texture source: generate a Mandelbrot, use it as a background under a Layer Composer composite, or feed it into an image-based node for stylization. It also pairs naturally with the pack's Procedural Gradient and Pattern generators if you want a scifi dashboard look with zero AI involved.

Installing it

Same pack, 24oiduts-ComfyUI:

• ComfyUI Manager: search 24oiduts, install, restart.
• Manual:

  cd ComfyUI/custom_nodes/
  git clone https://github.com/GeekyGhost/24oiduts-ComfyUI
  pip install -r 24oiduts-ComfyUI/requirements.txt

  then restart.

Needs nothing beyond Pillow/numpy. The GitHub README is a stale Studio42 template that claims the project is WIP with no license - an unmaintained banner, not a warning about this node.

Troubleshooting

• Slow renders - you're at a big size with high max_iterations; drop one of them. It's CPU math, so this is expected.
• Image is a solid color - the view might be outside the set (or in a spot where nothing escapes). Zoom out or nudge center_x/center_y.
• Julia looks identical every time - Julia sets depend on the constant, not a seed; change julia_c_real/julia_c_imag for variety.

A no-cost, no-download source of infinite procedural imagery. If you're sick of generating test images with a model, this is the node to grab.