Environment Styler (Advanced)

Same 283-entry location menu as the plain EnvironmentStyler - Beach, Desert, Forest, Mountain, Urban, Studio and a long tail beyond - in the "Advanced" node that opens up SDXL's two-encoder setup. This is actually a menu where the split can earn its keep: environment is broad, conceptual stuff, which is exactly what the big OpenCLIP-G encoder handles well. Still, it's SDXL-only, and if you're not on SDXL you want the basic node.

What "Advanced" means, and where it helps here

SDXL runs two text encoders at once - the big OpenCLIP-G and the smaller CLIP-L. This node splits your positive prompt into text_positive_g and text_positive_l. A reasonable play on an environment menu: broad scene to G (the setting, the mood), tight subject detail to L. That's the closest thing to a real use-case the G/L split has in this pack. But don't overthink it - same text in both still works fine. negative_prompt_to (Both / G only / L only) routes the negative. The whole split is meaningless off SDXL: SD 1.5 has one encoder, and the 2026 LLM-encoded models (Flux 2, Z-Image, Qwen) don't have this architecture.

How it works

It's a fork of SDXL Prompt Styler. Each environment is a JSON template with a {prompt} slot; your subject drops in, the location phrase gets wrapped at a strong weight (~2.0), and rewritten prompt strings come out. Text only - nothing loads a model.

Inputs and outputs that matter

• environment - the 283-entry dropdown. No option / none = off.
• text_positive_g / text_positive_l - your subject, once per SDXL encoder (or split broad-vs-detail).
• negative_prompt_to - Both (default), G only, or L only.

log_prompt prints the built prompt. The node returns six strings - text_positive_g, text_positive_l, combined text_positive, and the three negatives. Route _g/_l into an SDXL dual CLIP Text Encode, or use the combined text_positive / text_negative.

Installing it

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

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

No models, no dependencies. Nodes appear under ali1234/stylers.

Where people get burned

• The environment still eats the subject at 2.0. The G/L split doesn't tame that. If the setting dominates, ease the emphasis or reassert the subject.
• Chasing the "perfect" G/L split. Even on the one menu where it half-makes-sense, the gain is marginal. Start with the same text in both and only split if you see the subject and setting bleeding together.
• Off SDXL it's dead weight. No dual encoders means the _g/_l sockets do nothing - use the plain node.