Mood Styler

MoodStyler is the emotional-tone knob for your prompt. Pick from the dropdown - Aggressive, Calm, Chaotic, Energetic, Happy, Mysterious, Relaxed, Romantic, Sad, Serene - and it wraps your prompt in language meant to steer the overall vibe of the image. It's a gentler lever than something like lighting or camera; mood shifts the atmosphere rather than restructuring the shot. Reach for it when a render is technically fine but emotionally flat.

Mechanically it's the same trick every MileHighStyler node runs. Each mood is a text template with a {prompt} placeholder. Your text slots in, and the mood words the style adds get automatically weighted up (around ("...":2.0)) so the atmosphere pushes while your subject stays put. Nothing is generated or filtered - it's a curated phrase inserter, full stop.

The inputs worth knowing

• text_positive - your prompt, subject and scene.
• text_negative - your negatives; the mood preset can add its own here.
• mood - the dropdown. No option passes your prompt through unchanged.
• log_prompt - on by default; prints the final assembled prompt to the ComfyUI console.

It outputs two strings, text_positive and text_negative, that go straight into a CLIP Text Encode pair and onward to your sampler. If your workflow uses SDXL's split text_g / text_l inputs, use MoodStylerAdvanced instead - this basic node hands you a single clean pair, which is what most graphs actually want.

Installing it

Through ComfyUI Manager, search ComfyUI_MileHighStyler. It's a niche older pack that Manager sometimes doesn't index, so the dependable route is a manual clone:

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

Restart ComfyUI. There's no model download and no dependency install - the pack is entirely JSON style templates, so it's featherweight.

Common snags

Mood is subtle by nature, so the usual complaint is "I picked Melancholy-ish and nothing changed." Two things: first, turn on log_prompt and confirm the mood text actually got inserted; second, remember that a strong subject and busy scene can just swamp a mood cue. Mood works best on portraits and simple compositions where there's room for atmosphere to register.

The pack-wide caveat applies here as much as anywhere: this is SDXL / SD 1.5 technology. On the SDXL lineage (Juggernaut, RealVis, Illustrious, Pony) the weighted template does its job. On a 2026 LLM-encoded model - Flux 2, Z-Image, Anima - the (word:2.0) weighting is silently thrown away, so a mood preset barely registers. On those, you'll get more mileage writing the feeling into a plain sentence than routing it through this node.