Prompt Styler [Advanced] (Arctenox's Essentials)

This is the sibling of the plain Prompt Styler in the same pack - literally its subclass, sharing the same 52 style templates. What you get on top is three things: a second style slot to mix with the first, a ratio to control the mix, and a way to emphasize specific words.

The use case is the person who can't pick between Cinematic and Film Noir and wants both, or who wants a style plus "but push the colors harder." If you never want two styles at once, the plain PromptStyler is simpler and you don't need this.

How it works

Under the hood it's still string assembly, not embedding math. It builds the positive from your base_prompt, the quality tier's tags, and then the style tags - and here's the honest bit about the mix: style_mix_ratio doesn't interpolate two styles in a weighted-average sense. At 0.7 it includes the primary style's tags (70% weight, per the style_info output) and includes the secondary's tags too (30%); at 1.0 it drops the secondary entirely; at 0.0 it drops the primary. So the ratio controls whether each block is included and how it's labeled, not a smooth blend of embeddings. Both styles' negative suggestions get merged into the negative when use_negative_suggestions is on.

The genuinely different feature is emphasis_words. Give it comma-separated words and any that appear in the final positive get wrapped in (word:1.2) weighting syntax - the standard ComfyUI/A1111 emphasis format. Then wildcards resolve across the finished strings, seeded by wildcard_seed (−1 = fresh each run).

The inputs that matter

• primary_style and secondary_style - the two dropdowns, both from the same 52 presets.
• style_mix_ratio - default 0.7, meaning mostly primary. 1.0 = primary only, 0.0 = secondary only.
• quality_tier - None / Draft / Standard / High / Ultimate.
• emphasis_words - comma-separated, only affects words already present in the positive.
• use_negative_suggestions, wildcard_seed, plus base_prompt, base_negative, and the optional custom_positive / custom_negative - same as the plain styler.

The outputs

• positive_prompt, negative_prompt - plain STRINGs, wire them into a CLIP Text Encode node.
• style_info - a short string telling you exactly how the ratio split the two styles (e.g. "Primary: Cinematic (70%) + Secondary: Film Noir (30%)"). Nice for logging and for confirming the mix actually applied.

Install

ComfyUI Manager, search Arctenox's Essentials, install, restart. Or:

cd ComfyUI/custom_nodes/
git clone https://github.com/Arctenox/Arctenoxs-Essentials_ComfyUI

Restart. No model downloads - dependencies are just torch, numpy and optional psutil.

Gotchas

• The mix ratio is inclusion control, not true interpolation. If you want actual embedding-space blending, that's a different kind of node; here you get "primary's tag block plus secondary's tag block, with the ratio deciding the balance."
• emphasis_words is silent when it fails: a typo or a word that isn't in the final prompt is simply ignored, no error, no warning. Read the positive_prompt output before you assume the emphasis applied.
• Same CFG-1 caveat as the plain styler: on guidance-distilled models that run at CFG 1, the negative side does nothing, so don't sweat the style negatives there.
• The pack is marked DEPRECATED (remaster incoming), so it works today but isn't the kind of thing you build a permanent workflow on.