[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/vt/ - Virtual Youtubers

Search:


View post   

>> No.43610489 [View]
File: 1.66 MB, 1536x1024, 03350-1020406241-(best quality), (ultra-detailed), masterpiece, 1girl, solo, (nekomata okayu), ahoge, blush, smirk, purple hair, purple eyes, sho.png [View same] [iqdb] [saucenao] [google]
43610489

>>43610054
Thanks for the advice! I'll definetly start with loras. Also thanks for the links to the guides and trainer, I'll check those out.
>bout training on AI-generated images: I don't know if using original prompts as captions, or running WD tagger would be better. If you will be using WD tagger, best models are swin-v2 and convnext-v2, but you still will need to check and edit generated tags.
Good to know. The vast majority of my folder is traditional fanart, but if I added my seperate AI folders images then I'd have even more.

Navigation
View posts[+24][+48][+96]