/vt/ - Virtual Youtubers

>>75860297
That's easy to do but I'm not sure everyone is keen on doing this entire prompt to get the 1st outfit instead of just outfit name though. I agree however that schizo prompting makes things more accurate. The lora is still half-half and nothing stops anyone from prompting everything. But when you want to make her naked and still have some parts of the outfit, 2-tagging makes more sense.

>minato aqua, 1girl, solo, long hair, breasts, bangs, dress, bow, ribbon, cleavage, twintails, medium breasts, blue hair, purple eyes, hair ribbon, purple hair, braid, ahoge, short sleeves, multicolored hair, frills, puffy sleeves, bowtie, nail polish, two-tone hair, puffy short sleeves, wrist cuffs, maid headdress, blue bow, frilled dress, blue nails, blue footwear, colored inner hair, anchor symbol

You can check these two, one with full prompting and one where most outfit prompts were removed on the same lora:
https://files.catbox.moe/8nacda.png
https://files.catbox.moe/o5g0hz.png

That entire Lora + Embedding training would make things soo much easier if I could easily train embeddings with each outfit AFTER the training is complete, one for each outfit, and then give them the same name of the outfit so when you use it in the prompt, it actually loads everything by itself.