👚 Text Generator (Outfit Male)

This is the outfit feeder in the pack's Text Generator family, and it's by far the most sprawling one - where the character node covers body and face, this one covers everything from head to feet, plus a couple of categories that go way past "everyday clothing." There's a dedicated armors dropdown with well over a hundred entries spanning bone armor to bronze plate, a uniforms list of similar size running from Roman legionary gear to sci-fi outfits, and - genuinely - a cosplay category with specific named characters like Geralt of Rivia, Goku, Link, and Spider-Man. It's clearly built by someone generating a lot of character art and wanting shortcuts for very specific looks, not just "casual" versus "formal."

How it works

The everyday layer is what most people will actually reach for first: top, bottom, footwear, head_item, eye_item, mouth_item, and accessories, each paired with its own COLOR_* field so you're not stuck with whatever default shade the item comes in. Every one of those follows the same NONE / RANDOM / pick-a-value pattern as the rest of the Text Generator family. Beyond that baseline sits the specialty layer - armors, uniforms, material, pattern, style_details, style, and cosplay - for when you want something more specific than "shirt and pants." cosplay_description (a boolean) expands a chosen cosplay pick into fuller descriptive text rather than just the character name. CUSTOM_PROMPT is the usual escape hatch for anything the dropdowns don't cover. There's no chaining input on this one the way the character node has - it connects out via GEN_OUTFIT into the character generator's optional input, so outfit flows into character, not the other way around.

The inputs and outputs that matter

• top / COLOR_top, bottom / COLOR_bottom, footwear / COLOR_footwear - the basics most beginners will actually set; everything else defaults to NONE fine.
• head_item, eye_item, mouth_item, accessories (each with a matching COLOR_*) - smaller detail layers, worth touching once the basics look right.
• armors, uniforms, cosplay - the specialty categories, huge dropdowns, for when you want a specific historical, fantasy, or cosplay look rather than modern clothing.
• CUSTOM_PROMPT (STRING, multiline) - freeform text on top of everything else.
• Output: GEN_OUTFIT - wire this into the character node's optional GEN_OUTFIT input to merge outfit and character into one description.

How to install it

Via ComfyUI Manager: search Bjornulf_custom_nodes, install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/justUmen/Bjornulf_custom_nodes
cd Bjornulf_custom_nodes && pip install -r requirements.txt

then restart ComfyUI. No special dependency for this node itself - it's a large dropdown/text builder - but it's part of the shared 170-node repo, so the pack's requirements.txt still gets installed alongside it.

Common issues & troubleshooting

Output on its own doesn't do anything to your image. This node only produces the GEN_OUTFIT fragment - it has to be wired into the character node's optional outfit input, and that character node's output in turn needs to reach the main Text Generator node, before anything in it shows up in a generated image.

Too many categories at RANDOM produces a jumbled, over-described outfit. With this many independent categories, leaving everything on RANDOM at once can stack conflicting details (armor plus a modern accessory plus a specific pattern) that fight each other visually. Set the categories you actually care about explicitly and leave the rest at NONE rather than RANDOM if you want a coherent look.

cosplay picks the character name but the outfit doesn't look right. Whether your checkpoint actually knows what "Geralt of Rivia" or "Iron Man" looks like depends entirely on what it was trained on - this node just emits the name (or, with cosplay_description on, a fuller text description) as a prompt fragment. It's not pulling in any reference image or LoRA on its own, so results vary a lot by checkpoint.