Nodes/WWAA-CustomNodes/🪠️ WWAA Sliced Art
ComfyUI Node

🪠️ WWAA Sliced Art

A mirrored kaleidoscope on the GPU, plus the weirdest strip-shuffle you'll see

By hgabha·Created 2 years ago·Updated 6 months ago· 38
🪠️ WWAA Sliced Art
  • image
  • flippedOutput
  • Pass1Output
  • FinalOutput
num_slices20
DoubleSizefalse

WWAA Sliced Art takes one image, mirrors it into a 2×2 kaleidoscope, and then - this is the fun part - chops the result into vertical strips and rearranges them in an alternating pattern: strip 1, strip N, strip 2, strip N−1, 3, N−2, and so on. Then it does the same thing again on the horizontal axis. What comes out looks like the album art your friend made in 1997, and for certain abstract, poster, or background-generation workflows that's exactly the vibe you want.

It's pure visual flavor - there's no "correct" use, it's a creative tool. If you've ever generated a nice texture or pattern and wanted to tile it into something kaleidoscopic without dragging it into Photoshop, this does it on-GPU with zero image editing outside ComfyUI.

The three outputs

  • flippedOutput - the mirrored 2×2 grid: original in the top-left, horizontal flip top-right, vertical flip bottom-left, both flipped bottom-right. This one alone is worth the install for making seamless-ish mirror tiles.
  • Pass1Output - the grid sliced into num_slices vertical strips, reordered 1, N, 2, N−1, 3, N−2…
  • FinalOutput - Pass1 output sliced into horizontal strips, same alternating shuffle applied.

The num_slices control (20–80, default 20, even steps only) sets how many strips each pass uses. Fewer slices = chunky, abstract blocks; more slices = finer, almost glitchy textures. And DoubleSize changes the canvas behavior: False (default) squishes your image to half size on the original canvas; True keeps the image full size and doubles the canvas. For a crisp mirror at native resolution, you want True.

What to watch for

  • Batch size must be 1. Feed it a batch and it throws a clear error - this is a single-image node, so run it after splitting your batch.
  • The shuffle is deterministic. Same image, same slice count, same output. If you want variety, vary the slice count.
  • The strips are produced with bilinear resize when the dimension doesn't divide evenly by the slice count, so a tiny bit of softness can sneak in at the seams on odd resolutions. For a glitch-art tool it's a non-issue.

Install

It's part of WWAA-CustomNodes. In ComfyUI Manager search "WWAA Custom Nodes" and install; or:

cd ComfyUI/custom_nodes
git clone https://github.com/hgabha/WWAA-CustomNodes
# restart ComfyUI

No extra dependencies - the pack needs only ComfyUI's stock torch/numpy/Pillow, and this node is pure PyTorch tensor ops on the GPU. Under the 🪠️ WWAA/image menu.

The verdict

Honestly, this is the pack's personality node - the kind of thing the WeirdWonderfulAI.Art team built because they thought it'd be cool, not because anyone asked. Is it essential? No. Is it fun, fast, and genuinely good at making kaleidoscope and glitch textures that would otherwise take you five steps in an image editor? Yes. If your output feels too plain, run a render through here and pick whichever of the three outputs surprises you most.

Category🪠️ WWAA/image

Inputs (3)

NameTypeDefaultDescription
imageIMAGE
num_slicesINT2020–80
DoubleSizeBOOLEANfalse

Outputs (3)

NameTypeDescription
flippedOutputIMAGE
Pass1OutputIMAGE
FinalOutputIMAGE