Prompt Styler All

This is the kitchen-sink node of wolfden's SDXL Prompt Styler pack. Every other node in the pack carries one themed list of presets - cameras, horror scenes, art movements. SDXLPromptStylerAll throws all of them into a single dropdown with ~5,000 style choices. You type a plain subject, pick a style off the list, and the node wraps your text in a canned template so you don't have to remember the magic keywords yourself.

That's the whole pitch, and it's a genuinely handy one if you can't be bothered to memorize what makes a good "cinematic" or "neonpunk" prompt. It's also a lot to scroll through. More on that in a second.

How it actually works

There's nothing clever under the hood, and that's a feature. Each style is a JSON entry with two fields: a prompt string containing a {prompt} placeholder, and a negative_prompt string. When you pick a style, the node drops your text_positive into {prompt} and hands you the finished line. So the "enhance" style turns a fox in a forest into breathtaking a fox in a forest . award-winning, professional, highly detailed. No model runs, no API is called, no key is needed - it's string substitution against a big lookup table.

The negative side is the part people miss. If you type something into text_negative, it gets combined with the style's built-in negative. If you leave it blank, you just inherit the template's negative. So every style quietly ships its own negative prompt whether you asked for one or not.

The inputs that matter

• text_positive - your actual subject, in plain words. Keep it short; the style does the dressing-up.
• style - the preset. This is the giant dropdown. Start typing to filter it, or you'll be scrolling for a while.
• text_negative - optional. Merged with the style's negative, not replacing it.

log_prompt (No/Yes) prints the final combined prompts to your ComfyUI console - set it to Yes once so you can actually see what the style built. This node also has two extras the themed nodes don't: auto_select_style and auto_refresh, both booleans, off by default. Leave them off unless you're deliberately experimenting; the standard flow is pick-a-style-yourself.

The outputs are positive_prompt_text_g and negative_prompt_text_g, both plain strings. Wire the positive one into a CLIP Text Encode feeding your sampler's positive, and the negative one into the encoder feeding the negative. On SDXL you can send the same string into both the G and L fields of a CLIPTextEncodeSDXL. The _g in the name is just a nod to SDXL's dual encoders - don't overthink it.

Installing it

Easiest path is ComfyUI Manager: open it, search SDXL Prompt Styler (the wolfden one), install, restart. Manual works too:

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

Then restart ComfyUI fully. After the restart you'll find a new Style Prompts submenu in the add-node menu - the node lives there, not at the root. No models to download, no pip dependencies, nothing heavy. It's pure text handling.

Where people get tripped up

The honest caveat: these style templates were built for SDXL, which reads prompts through CLIP. On SDXL and its descendants - Illustrious, Pony, NoobAI, WAI - the padding earns its keep. On the 2026 LLM-encoded models (Z-Image, Flux 2 Klein, Anima), the quality-tag fluff does little, and the big boilerplate negative does nothing at all at CFG 1, which is where every Turbo/Lightning/distilled checkpoint runs by default - the negative pass literally isn't computed. The node will happily bolt its templates onto any model because it's just text, but "works" and "was tuned for your model" aren't the same thing.

Two more: if the node doesn't appear, you didn't do a full restart, or you're looking in the wrong menu (check Style Prompts). And if a style comes out overcooked, it's usually the inherited negative fighting your positive - flip log_prompt to Yes, read what it generated, and gut the negative if it's a forty-word generic block. The styles are just JSON in the pack folder, so you can open them and edit the templates to taste.

With 5,000 options this node is powerful but overwhelming. If you already know you want a camera or a specific art style, the themed nodes in the pack are far easier to navigate.