Nox Prompt Camera Master

Naming a physical camera is one of the highest-leverage phrases you can put in a prompt on modern models - the encoder was trained on real captions, so "85mm, medium close-up, low angle" does more work than "beautiful, detailed" ever did. Nox Prompt Camera Master is the pack's way of writing that language for you. Instead of remembering the difference between a dutch angle and a low angle, you pick shot_size, lens_angle, and camera_basic from dropdowns and it composes a camera prompt block, a framing summary, and rigging notes.

The README flags it as having an alias, NoxPromptCameraLanguage, for legacy graphs - if you have an old workflow with that node name, this is the same thing, so don't go hunting for a separate node. Its intended role in the pack is as the camera half of the "Camera Master + Lighting Master feed the Builder" pipeline.

How it works

Same machinery as the rest of the pack: curated dictionaries in constants.py for each dropdown, resolved into prompt fragments and summaries, assembled into the three outputs. shot_size, lens_angle, camera_basic, and camera_advanced cover framing; lens_focal and lens_type handle optics; motion_type, camera_motions, movement_beats, and camera_rig cover camera movement; time_of_day, color_tone, and visual_style tie into the look; and focus_subject / focus_environment keep the attention split under your control. There are safety-focused fields too - safety_considerations and rig_support - which the author clearly carries through every node in this pack.

Every dropdown has a Custom Input option that unlocks a *_custom field, and there are free-text slots (accent_keywords, camera_language, additional_notes) for the stuff vocabulary can't say.

Inputs and outputs that matter

The three required fields are shot_size, lens_angle, and camera_basic - set those three and you already have a credible camera prompt. lens_focal is the one I'd add next; on sentence-based models it's cheap and effective. Outputs: camera_prompt (the text to encode), framing_summary (what the framing actually is, in plain English), rigging_notes (the technical notes you'd give a crew - keep these out of the encode and in your production doc), and preset_status. The camera_prompt is what you feed into Nox Prompt Builder's camera_prompt input or into Pipeline Combiner.

Installing it

Whole pack, one install. ComfyUI Manager: search "Nox Prompter", install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/jamesk9526/NoxPrompter-ComfyUI-CustomNode

Restart, find the NoxPrompter category. Pure Python, no extra dependencies.

Gotchas

The README's clone URL and folder name are stale (ComfyNoxPrompter); the repo is the one above. Presets land in presets/ inside the pack folder, so watch write permissions. And a real caveat about how this reads: camera language is at its most powerful on LLM-encoded models like Wan 2.2 and Flux-style encoders - on SDXL-lineage checkpoints the same phrases still help, but they compete with booru-tag vocabulary, so keep the camera clause compact and don't bury your subject tags under it.