Weather Modifier (Single)

Weather Modifier is the node you reach for when the scene is right but the sky is wrong. Twenty weather conditions - perfectly clear, overcast, steady rain, heavy snowfall, dense fog, post-storm clearing - each a hand-written description appended to your base prompt by ID. It's the fastest way to test a street scene in drizzle vs. heat shimmer vs. dramatic storm clouds without touching your subject prompt.

The modifiers are organized into four groups. Clear/fair weather sits at IDs 1–4 (DRY): clear conditions, high pressure, light breeze, calm air. Cloudy and overcast is 5–8 (WET): scattered clouds, diffused light, heavy cover, dramatic formations. Precipitation is 9–13 (WET): light drizzle through heavy snowfall. And 14–20 are the atmospheric effects - morning mist, dense fog, heat shimmer, dust haze, post-storm clearing, wind-swept, and stormy - mixed between DRY and WET. The README flags #10 (steady rain, wet surfaces) as one of the pack's most popular modifiers, and it's a good example of the format: concise, concrete, and full of words checkpoints actually know.

The inputs: text (base prompt), modifier_id (1–20, default 7), and preserve_base (default true). With preserve_base off, the node strips common weather keywords from your base text first - so "rainy city street" becomes just "city street" before it appends snowfall. That swap behavior matters more here than in most dimensions, because weather is the most likely thing you've already typed in your base prompt. Output: single STRING, modified_prompt.

The mechanism is the pack standard - a loaded modifier table and a string append, zero dependencies, no models. Install: ComfyUI Manager → search "Variationator" → install → restart, or cd ComfyUI/custom_nodes && git clone https://github.com/MushroomFleet/ComfyUI-Variationator.git. (The repo's install-portable.bat mentions a requirements.txt that isn't shipped - skip that step.) Look under Prompt Orchestration.

Where this one really earns its place is paired with the lighting dimension - the pack's DRY/WET system was designed so you can deliberately mix "clear-sky lighting" with "wet-road weather" for contrast, and weather + lighting stacked is how you get those moody blue-hour-wet-road shots the README loves. One tip from the pack's own troubleshooting: if all your weather variations come out looking similar, the problem is usually that you're staying inside one category - mix DRY and WET, or bring in a different dimension like mood, rather than trying the same five IDs again. And as always with comma-joined modifier text, the output plays best with tag-based checkpoints; on LLM-encoded models it still reads fine as description, just watch total prompt length.