Prompt Environment Slider

If your prompts are half scenery tags you keep forgetting to type, this node is for you. Prompt Environment Slider is a drop-in prompt builder that handles the setting of an image - sky, indoor/outdoor location, buildings, flowers, background - one dropdown per category, and spits out a ready-to-use string. It's from RvTools v2, r-vage's grab-bag pack of workflow glue, and it works exactly the way it looks: pick a category, get a prompt chunk.

How it works

Under the hood the node reads a bundled environment.json (ships in the pack's json/ folder) and turns each category into a dropdown. The lists are genuinely big - Flowers alone has 179 entries, Outdoor has 87, Building 86 - so this is a faster way to find "alley at dusk with cherry blossoms" than digging through a tag list. Every category has three states: None (skip it), a specific pick, or Random, which rolls one option from the list.

Each category also gets a weight slider (0.1 to 2.0, default 1). Pick a value and it's emitted plain; pick 1.3 and the node wraps it as (alley:1.3) before joining everything with commas. The output is a single prompt STRING, which you wire straight into a CLIP Text Encode positive.

The inputs that matter

• The six category dropdowns - Sky, Indoor, Outdoor, Building, Flowers, Background - with their matching weight sliders. That's the whole job.
• seed - and here's the honest part: in the current source the seed is declared on the node but the Random pick never actually reads it. The roll uses Python's global random state, so if you need a reproducible random environment, this node won't give you one from that widget. Set each dropdown to a specific value if repeatability matters more than surprise.

Where it sits in a workflow

This is an SDXL/booru-tag-era tool. If you're prompting Illustrious, Pony or NoobAI, the comma-joined tags slot right in. On Flux, the tags mostly still land (T5 reads natural language well enough), but the (tag:weight) syntax is silently discarded by Flux's encoder - so the weight sliders are doing nothing for you there. Keep that in mind before you blame the node for a muddy result.

Installing it

ComfyUI Manager: search ComfyUI-RvTools_v2 (or "RvTools") and Install, then restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/r-vage/ComfyUI-RvTools_v2

Restart ComfyUI and it appears under 🫦 RvTools II → Text. No model downloads, nothing heavy - the pack's requirements (torch, numpy, Pillow, opencv, pilgram, pynvml, piexif) are almost all already in your environment.

Gotchas

The README is upfront that RvTools v2 is no longer maintained, superseded by ComfyUI_Eclipse - it still works fine, just no new features. The other trap is v1 vs v2: the author renamed every node between versions to fix ComfyUI Manager's node library, so a workflow built on the original ComfyUI-RvTools won't match these names and Manager will report it missing. If you hit that, you need the v1 repo, not this one. Here, the only real annoyance is the dead seed widget - set specific values when you need the same scene twice.