APNext Glitch Art FX

Digital decay as a look - RGB channels torn apart, scanlines, pixel-sorted streaks, JPEG-mangled blocks. APNext Glitch Art FX applies those corrupted-signal effects to a finished image on purpose, so you can get the vaporwave / databending / broken-VHS aesthetic without actually corrupting files by hand (the old-school way, which is fiddly and destroys images unpredictably). Pick a glitch type, set how hard to push it, and reroll until you like the damage.

Image in, image out, after VAE decode. No model, no API key.

How it works

The glitch_type dropdown picks which corruption to simulate, and each one is a different pixel operation: RGB_Shift and Color_Channel_Shift slide the color channels apart for that 3D-without-glasses fringing, Pixel_Sort reorders pixels along rows into smeared bands, Scanlines overlays CRT lines, Digital_Noise adds static, Data_Moshing and Compression_Artifacts fake the blocky mess of a broken video codec. intensity controls how far it goes; randomness adds variation to where and how the glitch lands; and seed makes a given glitch reproducible so you can lock a result you like.

The seed matters more here than in most FX nodes, because glitch is inherently random - without a fixed seed every run gives you different damage, which is great for exploring and annoying when you found the perfect one and want it back.

Inputs and outputs that matter

• images - your decoded image.
• glitch_type - the effect. RGB_Shift is the classic, most-recognizable glitch; Pixel_Sort is the more artful, streaky one.
• intensity (0–2, default 0.5) - how broken it looks. Small values give a subtle digital edge; high values obliterate the image.
• randomness (0–1, default 0.3) - how chaotic and scattered the glitch placement is.
• seed - lock a glitch you like; roll it to keep generating new ones.

The single output is image.

Installing it

ComfyUI Manager: search comfyui_dagthomas, install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/dagthomas/comfyui_dagthomas
cd comfyui_dagthomas
pip install -r requirements.txt

Runs on the pack's image-processing dependencies. No API key, no model files.

Common issues

Glitch is a taste effect and it tips into ugly fast - full intensity on Data_Moshing usually just looks like a rendering bug, not art. The convincing versions live at low-to-mid intensity where the image is still readable and the glitch reads as an intentional layer on top. Start around 0.3–0.5.

Because the output is random, treat this like a slot machine: set a seed to randomize while you explore, then freeze the exact seed once you've got a keeper - otherwise the good one is gone on the next run. And it stacks: a subtle RGB_Shift plus a few Scanlines plus some grain reads as a much more deliberate broken-screen aesthetic than any single effect maxed out.