arkennemasis Subject Line (gender + notes)

Here's the failure mode this node exists to kill: you write 24 prompts for a character dataset, all phrased "a woman doing X, a woman wearing Y," and then the client hands you reference photos of a man. Now every prompt is wrong and you're doing a find-and-replace. ArkSubjectLine is the fix - a single node that builds one Subject: ... line from a gender choice plus free-text notes, and you wire that line into every prompt. The prompts themselves stay gender-neutral ("the person ..."), and the gender lives in exactly one place.

It's a trivial node and that's the point. One subject STRING output, three inputs:

• gender - an enum: let the reference decide, a man, a woman, a non-binary person, a boy, a girl. The default omits gender entirely, which is the right move when the reference photo is unambiguous and stating it adds nothing.
• notes - free text for age/build/ethnicity, e.g. "in his late twenties, athletic build".
• prefix - the label in front of the line (default Subject:); blank for none.

The output is a single clean line - Subject: a man in his late twenties, athletic build. - and the node strips trailing punctuation on notes so it reads as one phrase rather than two sentences stapled together.

The captioning rule the tooltip is warning about

The notes tooltip carries a genuinely important warning for LoRA work: describe age, build and ethnicity here, but not hair or eye colour. For a training set, anything you caption is excluded from what the LoRA learns - caption the variable parts, never the identity features. That rule runs through the whole pack (the caption sidecar node says the same thing), and it's the reason a VLM description of a finished image is usually the wrong caption source: it dutifully writes "brown hair, green eyes" and teaches the model those are separable instead of learned.

Where it fits

Wire subject into every prompt in the graph - image prompts, video prompts, anything that mentions the character. Because it's one authoritative source fanned out to many consumers, changing the gender or the notes once updates the whole run. If the pack's character-dataset workflow is your use case, this sits right before the prompt builder, next to the reference photo and the System Instructions node.

Install

ArkSubjectLine is one of the 61 nodes in the comfyui-arkennemasis pack, under arkennemasis/Utility:

cd ComfyUI/custom_nodes
git clone https://github.com/Hishamahmer/comfyui-arkennemasis
pip install -r comfyui-arkennemasis/requirements.txt   # then restart ComfyUI

Or ComfyUI Manager → Install via Git URL with the repo URL. No models, no key - pure string plumbing. If you're only wiring a handful of prompts, you can absolutely type the subject into each one; this node earns its keep the moment a set of prompts gets long enough that a leftover "A woman" could land on a set of photos of a man.