Prompt Settings

The hardest part of a good prompt isn't the subject - it's the how. How it's shot, by whose style, under what light, on what film. Prompt Settings is the pack's answer to that second half: 21 dropdowns covering shot styles, artists, film directors, lighting, effects, cameras, film stocks and art styles, all curated from a data file, assembled into a single comma-joined style string. It's the natural companion to Prompt Subject - that node builds the who and what, this one builds the cinematography.

What's in the dropdowns

The categories read from a settings.json shipped with the pack. The ones you'll actually care about:

• Shotstyle 1 and Shotstyle 2 - 106 options of framing and lens language: 14mm wide angle lens, 85mm long lens, aerial shot, angle from below, between fingers, and so on. Two slots so you can combine a lens with an angle.
• Artists - 154 named artists (Alphonse Mucha, Alfred Sisley, Alberto Seveso...) plus Artists Special (21 curated contemporary/niche names). Yes, the code spells that category "Arists Special" - a typo that made it into the shipped JSON key; you'll see it misspelled in the node.
• Film Directors - 72 of them, from Akira Kurosawa to Alejandro Jodorowsky, for the "in the style of" crowd.
• Photography Styles, Lighting, Effects, Magic Elements, Camera, Films - the sensory layer: lighting rigs, bokeh, film stocks, camera bodies.
• Image Type, Composition Method, Art Style, Art Style+, Scene Atmosphere, Theme, Printing Materials, Illustration Style, Color Theme - the genre and finish layer.

Every category defaults to None and is skipped when unset; Random is available everywhere and picks from the non-None options. A seed input makes the randomness reproducible.

The single prompt output is a comma-joined style string for a CLIP Text Encode. Because it's built from real artist and camera names, it reads the way tag-based checkpoints like Illustrious and Pony actually want - these are the exact strings the models were trained on.

How to use it with the rest of the pack

The intended composition is three nodes: Prompt Subject (character + action), Prompt Environment (the setting/location, also in the pack), and Prompt Settings (style/cinematography). Feed all three outputs into a text-concat or a CLIP Text Encode and you get subject, scene and style in one coherent prompt - which is a lot easier to reason about than one giant string you hand-wrote.

Installing it

Part of ComfyUI-RvTools_v2 - Manager → search "ComfyUI-RvTools_v2", or:

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

The data lives in the pack's json folder - no model downloads.

Gotchas

A few reality checks. This is tuned for CLIP-era, tag-based checkpoints (SDXL family, Pony, Illustrious); on LLM-encoded models the artist names still carry weight but the tag-soup formatting is less of an advantage. The categories are curated lists, not exhaustive - if the artist you want isn't there, you can't type it in, so keep a fallback text node handy. And the typo'd "Arists Special" is cosmetic; it works fine, it just looks wrong. Pack-level caveat as always: RvTools_v2 is unmaintained, superseded by ComfyUI_Eclipse, and [RvTools]-suffixed names can trip Manager on old shared workflows.