PDXL/SD1.5 previews are bad, i don't sure loras bad as previews(i'm too lazy to try found good negatives/positives tags for tune it(now i use illu myself) -> possible it works better than on preview).
"training images" is fake, archive contain wildcards file with outfits, for you can easilly automatize Emilia generation process.(Civitai support zip files only for train data->no way to upload wildcards except this one)
GUIDE HOW TO GET IMAGES LIKE MY PREVIEWS
Collection of my multicharacter PDXL LoRAs
Commisions open
Support My Work:
Your support helps me create more amazing content and keeps this passion project alive. Every contribution, no matter how small, makes a big difference. Thank you for being awesome!
Description
legacy support/ it not tuned for sd1.5, just same dataset as pdxl/illustrious one -> Quality not top of SD1.5