Camera

Of all the blocks this pack lets you file away, camera language is the one where a dropdown earns its keep. Naming a physical camera - 85mm, f/1.8, shallow depth of field - is one of the highest-leverage things you can put in a prompt on caption-trained models, and it's also the thing people fiddle with most while keeping everything else identical. The Camera selector exists so that swap is a click, not a paste-and-edit.

It's part of the Prompt String Selector family, a pack that treats a prompt as a set of text-file blocks - style, character, pose, camera, background - each swapped by a dropdown. Camera is one of the more satisfying categories in it, because the effect is immediate and you can feel the difference when you switch from a wide establishing shot block to a tight closeup block on the same seed.

How it works

Reads ComfyUI/custom_nodes/ComfyUI-Prompt-String-Selector/data/camera/. Each .txt file is a dropdown option, filename minus extension is the label, and the leading number becomes the id output. Ship contents, comment with // or /* */, sort by the number prefix - same rules as every other selector in the pack.

Out of the box there's just 01_none.txt (effectively a no-op). You'll want to add your own: 02_full_body_shot.txt, 03_portrait_85mm.txt, 04_low_angle.txt. Each file can be a couple of camera phrases long - these blocks are meant to be compact, since the model reads them as a tag group.

Inputs and outputs

• camera - dropdown of your saved camera blocks.
• camera_custom - multiline override. Choose custom_input and this box's text wins; id becomes 0.

Outputs: text (STRING, comma-terminated) and id (INT). The comma termination matters - this pack expects you to chain blocks through its Aggregator, and the node adds the separator so camera glues onto character without you thinking about punctuation.

Install

Via ComfyUI Manager (search "Prompt String Selector") or:

cd ComfyUI/custom_nodes
git clone https://github.com/itom0717/ComfyUI-Prompt-String-Selector.git

Restart ComfyUI. There is no requirements.txt and nothing to download - it's the standard library only, which makes this one of the lower-risk custom node installs you can do.

Gotchas

Dropdowns rebuild at startup, so a new .txt file needs a restart before it appears. Plan for it - write your camera files in a batch, then restart once.

Also worth knowing: this pack's blocks are built for CLIP-style tag models (SDXL, Illustrious, NoobAI, Pony). If you're on a 2026 LLM-encoded model like Z-Image or Flux 2 Klein, the node still works - it's just text - but treat your camera phrasing as prose inside a scene paragraph rather than a bare tag list. The mechanism is agnostic; the phrasing is on you.