Preposition Styler (Advanced)

I'll level with you: this is the most niche node in the pack. PrepositionStylerAdvanced injects a preposition - aboard, about, above, across, against, among - from a menu of ~66, into your prompt. A preposition on its own doesn't describe a subject, a style, or a look; it describes a relationship between things. So this node really only earns its place as a connective piece when you're stacking stylers and want to glue two style fragments together with spatial phrasing. On its own, a single preposition at heavy weight is close to a no-op.

It exists mostly for completeness and for people building elaborate multi-node styling chains. If that's not you, you can safely skip it. This is the SDXL dual-CLIP variant of the menu.

How the "Advanced" plumbing works

SDXL uses two text encoders, CLIP-G and CLIP-L. This node feeds them separately - text_positive_g to G, text_positive_l to L - and applies the chosen preposition to both, wrapped at the pack's strong default weight (~2.0). MileHighStyler is a fork of the classic SDXL Prompt Styler. negative_prompt_to (Both / G only / L only) routes any built-in negative; Both by default.

The inputs and outputs that matter

• preposition - the ~66-entry dropdown; none is off.
• text_positive_g / text_positive_l - your subject per encoder; same text in both if unsure.
• text_negative - your negative, routed by negative_prompt_to.

log_prompt prints the assembled prompt so you can actually see how the preposition wove in - genuinely useful here, since its whole value is placement. Six outputs: text_positive_g, text_positive_l, merged text_positive, and text_negative_g / text_negative_l / merged text_negative. Feed _g / _l into a CLIPTextEncodeSDXL, or take the merged text_positive / text_negative into a normal 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

• Expecting it to do something alone. A lone preposition rarely changes an image. Its job is connective - inside a stack of stylers, not as a standalone effect.
• Spatial relationships are hard anyway. Even a well-placed above/below won't reliably enforce object placement on a diffusion model; that's a regional-prompting or ControlNet problem, not a word-menu one.
• Advanced when basic would do. The G/L split only matters with a CLIPTextEncodeSDXL and different per-encoder text. Otherwise use the merged outputs.
• Model era. Dual-CLIP is SDXL-only. Meaningless on the 2026 LLM-encoded models (Flux 2, Z-Image, Qwen) - which, ironically, handle spatial language far better when you just write a normal sentence.