Focus Styler

FocusStyler is a 67-entry menu built mostly around focus-area percentages - "10% Focus Area," "50% Focus Area," "100% Focus Area," plus a few named ones like "Autofocus." The idea is to describe how much of the frame is sharp. I'll be honest with you: this is one of the more gimmicky menus in the pack. "30% Focus Area" isn't standard photographic language, so the model doesn't have a precise, learned meaning for it - it's a soft nudge at best, and the results between adjacent percentages tend to blur together.

If you want to actually control the in-focus region, the DepthStyler depth-of-field menu speaks the language diffusion models actually learned (shallow/deep DoF, bokeh, macro). FocusStyler is worth a spin for curiosity or variety, but don't expect the percentages to behave like a real focus dial.

How it works

MileHighStyler is a fork of SDXL Prompt Styler. Each focus option is a JSON template with a {prompt} slot; the node drops your text_positive in and wraps the focus phrase at a strong weight (~2.0). The catch, as above, is that the model has to understand the phrase for the weight to matter, and "40% Focus Area" is fuzzier to it than "shallow depth of field." Output is a rewritten prompt string into your CLIP Text Encode; nothing touches the model.

The inputs and outputs that matter

• focus - the 67-entry focus-area dropdown. No option is the off switch.
• text_positive - your subject; the focus cue attaches to it.
• text_negative - optional negative text.

log_prompt prints the finished prompt to the console. 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

• Percentages aren't a real control. This is the main thing. Don't expect "70% Focus Area" to sharpen exactly 70% of the frame - the model has no such concept. Treat it as a vague vibe, not a slider.
• Use DoF language instead. For anything you actually care about, depth-of-field terms steer far more reliably. Reach for DepthStyler and leave FocusStyler for random exploration.
• Model era. The 2.0 weighting is SDXL/SD 1.5 syntax, silently discarded on 2026 LLM-encoded models (Flux 2, Z-Image, Qwen). And since the phrasing is already weak, there's even less reason to bother with it on those - just describe the focus you want in words.