Prompt Styler Cyberpunk Surrealism

This is the maximalist one. Fifty-one styles, all variations on wet-neon dystopia bleeding into dream logic: Neon Metropolis, Biomechanical Dreamscape, Cyberpunk Underwater Abyss, Data Hacker's Nightmare, City of Whispering Machines, Neo-Tokyo Redux, Bioluminescent Urban Jungle, Psychedelic Cyber Alley. You bring a subject, pick a flavor of chrome-and-fever-dream, and the node paints the whole atmosphere around it.

If you've ever tried to hand-write a convincing cyberpunk prompt you know it's a grind - the rain, the reflections, the volumetric haze, the color palette, the negative prompt to keep it from going flat. This node from wolfden's SDXL Prompt Styler pack is that grind pre-baked into a dropdown, fifty-one times over.

How it works

Each style is a JSON entry with a prompt template (holding a {prompt} placeholder) and a negative_prompt. Your positive text goes into the placeholder, the cyberpunk-surrealist scaffolding wraps around it, and you get back a finished string plus a matching negative. Leave your negative blank to use the template's; type one and it merges. Nothing runs on the GPU - it's string assembly, and the image quality comes entirely from the checkpoint you feed it into.

Given how heavy these templates are, the styler is doing real work for you here: a good cyberpunk prompt is long and specific, and that's exactly the kind of boilerplate worth factoring out so you can iterate on the subject alone.

The inputs and outputs that matter

• style - your dystopia. The node's reason to exist; there are 51 to try.
• text_positive - the subject the neon world forms around. "a lone courier," "a temple," "a woman in a trench coat."
• text_negative - optional. Blank inherits the template's negative (already tuned to fight muddiness); add your own for specific recurring problems.

log_prompt (No/Yes) echoes the full assembled prompt to the console - useful, because these templates are dense and you'll want to see what's actually being sent.

Outputs: positive_prompt_text_g and negative_prompt_text_g. Two strings, into your positive and negative CLIP Text Encode nodes respectively.

Installing it

ComfyUI Manager: search SDXL Prompt Styler (wolfden), install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/wolfden/ComfyUi_PromptStylers

Restart ComfyUI. No pip dependencies, no model files. The nodes appear under a Style Prompts submenu once ComfyUI restarts.

Common issues

Over-styling. These are the heaviest templates in the pack. If your own positive is already elaborate, stacking a dense style on top pushes you into prompt soup where every concept dilutes the others. Keep your text lean - a subject and maybe one anchoring detail - and let the style do the heavy lifting.

"none" is a passthrough. The first list entry, none byCyberpunkSurrealism, applies nothing. If the node seems to do nothing, that's the usual culprit.

Best on SDXL, not Flux/Z-Image. The templates are SDXL-era, and they lean on tags and a working negative prompt. On guidance-distilled or LLM-encoded models the negative doesn't fire at CFG 1 and the descriptors read as literal instruction, so you lose most of the punch. Run these on Juggernaut, Illustrious, or another SDXL checkpoint.

Sameness across styles. Fifty-one presets sound like variety, but many share DNA - neon, rain, reflections. If two look alike, it's because the subject is doing more than the style. Change the subject, not just the preset.