Load Fashion Model

This is the node that tells you everything about the pack it ships in. Load Fashion Model is not a general-purpose utility - it's internal tooling for the author's fashion product, "Faishme," made public. It does one very specific job: hand you a random model photo and its matching mask, from a folder of your own images, so the rest of the workflow can generate that model wearing a garment. If that's your use case, it saves you an image picker and a file-loader every single run. If it isn't, it's basically a randomized image loader you'll only reach for because the rest of the pack makes sense around it.

How it works

The node doesn't know anything about any model weights. It knows two things: your gender and your ethnicity, and it turns those into folder paths under your ComfyUI install:

ComfyUI/faishme/models/{gender}/{ethnicity}/   ← the model photos
ComfyUI/faishme/masks/{gender}/{ethnicity}/    ← one mask per photo

It globs both directories, picks a random index, and loads that image/mask pair. The mask output is just the alpha channel of the matching mask image, ready to feed an inpainting or conditioning step. Two details matter: the files are paired purely by sort order, so the Nth model photo must correspond to the Nth mask; and the node's IS_CHANGED returns a random value, which means every queue re-rolls to a different model and re-runs whatever is downstream. That's deliberate - it's the "surprise me with a different model each time" behavior you'd want for batch product shots.

Inputs and outputs

Only two inputs, and they're just folder selectors:

• gender - female or male
• ethnicity - asian, black, caucasian, or indian

Outputs are an image (IMAGE) and a mask (MASK). Wire the image into your conditioning / img2img path and the mask into whatever does the garment replacement or background work.

The catch: the data is your job

This is where beginners trip up. The node loads your images from ComfyUI/faishme/models/... - it ships with none. Point it at empty folders and it'll blow up with a NumPy error about an empty array, not a friendly message. Before you use it, drop a set of model photos and matching masks into the right subfolders, restart ComfyUI, and confirm the paths (note it's faishme, not the usual models directory, and it's at ComfyUI's root, not under custom_nodes). This is squarely the "internal tooling made public" situation: the folder convention is whatever their product pipeline uses, and it's not documented anywhere except the source.

Installing it

Same story as the rest of the pack - search "ComfyUI_faishme" in ComfyUI Manager, or:

cd ComfyUI/custom_nodes
git clone https://github.com/AkashKarnatak/ComfyUI_faishme

Then restart ComfyUI. The pack's requirements.txt pulls in a grab-bag (numpy, torch, pandas, pillow, pyvips, gpustat, google-genai, vertexai) - most of it you won't need for this node, but Manager installs it all anyway.

Worth it?

If you're building a batch fashion/garment pipeline with a folder of pre-masked model photos, yes - it's the one node in this pack that's hard to replace with a couple of built-ins. If you just wanted a random image loader, the built-in Load Image plus a random file picker gets you there with fewer moving parts. Fair warning: this pack is brand new and essentially unmentioned in the community, so treat it as the author's own pipeline exposed to the public, not a polished, documented product.