Style Preset

Style Preset is the most boring-looking node in the Prompt Builder pack and arguably the most influential: one dropdown, 59 options, and it decides the entire visual direction of the image. The README calls it out explicitly - "this has a major influence on the final output" - and the pack's code agrees, because the style string is the first thing Combine Prompt stitches into the final prompt.

It's also the easiest node in the pack to understand. There's no layering, no duplicate slots, no weight fiddling. You pick "anime portrait" or "cinematic portrait" or "photorealistic portrait" and that's the whole interaction. On the CLIP-encoded anime-SDXL models this pack is built around, earlier tokens get stronger attention, and since the style lands first in the join order, it sets the frame everything else gets interpreted through.

The input and the output

There's really only one thing to know beyond that: the input is style_preset, but the output is named style, not style_preset. It's a small thing, but it trips people up when they're hunting for the right wire - expect to see "style" on the output socket. It's a single string that plugs into the style_preset input on the Combine Prompt node.

The dropdown itself leans anime, which tells you exactly who this pack was written for: anime portrait, high quality anime illustration, semi-realistic anime, anime key visual, anime screencap style, semi-realism - and then a set of photography options (studio portrait photography, fashion photography, editorial photoshoot, cinematic portrait). If your model of choice is Pony V6, Illustrious, or NoobAI, this menu maps cleanly onto what those models know.

Install

No dependencies, no model downloads - the pack is plain Python reading a JSON file. Clone it or use ComfyUI Manager:

cd ComfyUI/custom_nodes
git clone https://github.com/zinigo-creations/comfyui-prompt-builder

Then restart ComfyUI.

Common issues

The main misconception is that a style tag transforms a model. It doesn't. "Photorealistic portrait" on an anime checkpoint won't make your render photographic - it gently steers a model toward the photorealism it already saw in training, and on a heavily anime model that's still mostly anime with slightly different lighting. The style preset is a steering wheel, not an engine swap.

Second, the whole approach is a tool for the SDXL-anime lineage. If you're on a 2026 LLM-encoded model like Flux or Z-Image, the tag vocabulary this pack speaks loses a lot of its power - those models read sentences better than tag lists, and the score_x/style conventions were trained in on the anime side. Style Preset will still add words to your prompt; whether those words do anything is up to your checkpoint. And if you edit presets.json to add your own style options, restart ComfyUI to pick up the change.