Focal Point Styler

FocalPointStyler injects a "focus" tag into your prompt - a booru-style cue for which part of the subject the image should center on. The ~122-entry menu runs from the tame (eye focus, hand focus, back focus) to the very much not (ass focus, breast focus, and plenty more). In Danbooru's vocabulary a <part> focus tag is a real, learned convention for compositional emphasis, and this node is a dropdown for it. Like most of this pack, it leans NSFW, and it's described here plainly because that's the job it does.

The thing to know up front: a "focus" tag is a compositional suggestion, not a crop or a mask. It biases the model toward emphasizing that region - larger in frame, sharper, more central - but it won't guarantee it, and on a model that never learned the tag it does nothing at all.

How it works

MileHighStyler is a fork of the classic SDXL Prompt Styler. Each menu entry is a JSON template with a {prompt} slot; FocalPointStyler drops your text_positive into the chosen tag's template and wraps the style text at a strong default weight (~2.0, per the pack docs). It's text only - the output goes to your CLIP Text Encode and the model interprets the emphasis.

The inputs and outputs that matter

• focal point - the ~122-entry <part> focus dropdown. none is the off switch.
• text_positive - your subject and scene; it lands in the template.
• text_negative - optional negative text.

log_prompt prints the finished prompt so you can confirm the focus tag landed. Outputs are text_positive and text_negative strings, into a CLIP Text Encode.

Installing it

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

cd ComfyUI/custom_nodes && git clone https://github.com/TripleHeadedMonkey/ComfyUI_MileHighStyler

then restart. No models, no requirements.txt, no dependencies - Python and JSON. Nodes appear under ali1234/stylers.

Where people get burned

• These are booru tags. X focus is Danbooru vocabulary and needs a booru-trained model (Pony, Illustrious, NoobAI, WAI, SD1.5 anime) to mean anything. On photoreal or base SDXL, mostly nothing.
• Focus isn't a crop. It nudges emphasis; it doesn't reframe. If you need the subject genuinely centered and large, also say so in your positive text and set your composition accordingly - the tag alone won't do it.
• Model era. The 2.0 weighting is CLIP-encoder syntax, discarded on the 2026 LLM-encoded models (Flux 2, Z-Image, Qwen). Use this on the SDXL/SD1.5 family; on newer models write the emphasis in plain words.
• Over-weighting distorts. A heavily weighted focus tag can warp proportions to force the emphasis. If anatomy goes strange, dial back your expectations of what one weighted tag should carry.