Solarize

There is exactly one input on this node, and that's the whole appeal. Solarize flips the pixels above a brightness threshold to their inverse - like the photographic Sabattier effect where part of the image develops inverted. Push the highlights toward their negative while the shadows stay put, and you get that striking, slightly otherworldly high-contrast look that's been used in album covers since the 60s. It's the pack author's favorite, and it's about as close to zero-effort as a creative effect gets.

How it works

The math is one line: every pixel channel brighter than the threshold becomes 1 - pixel, everything darker stays as-is, then the result is clamped. Because it runs per-channel, it inverts colors rather than just brightness - reds flip to cyan, greens to magenta. That's what gives the effect its punch.

• threshold (0–1, default 0.5) - the brightness cutoff. At 0.5 the top half of your tonal range inverts. Push it up toward 0.8 and only the brightest highlights (speculars, sky) flip; drop it to 0.2 and most of the image goes negative. This single slider is your whole creative range.

Output is a single IMAGE.

Install

Solarize is one of ~22 nodes in ComfyUI-post-processing-nodes.

• ComfyUI Manager: search "post processing nodes", install, restart.
• Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/EllangoK/ComfyUI-post-processing-nodes

Restart after cloning. No requirements.txt, no model downloads, no extra pip - just torch ops, so it works on any stock ComfyUI. It's under postprocessing → Color Adjustments in the node menu.

Notes from playing with it

At the default 0.5 you often get a wild, near-total inversion that reads as "poster art" more than "photo." The subtler look lives near the extremes: 0.8–0.9 inverts just the sky and brightest highlights, which photographs surprisingly well. And since the node is cheap, you can afford to chain it with the pack's FilmGrain or ColorTint afterwards to give the inverted image some texture instead of a flat, glossy negative. One input, but it rewards experimentation.