[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]

/vt/ - Virtual Youtubers

Search:


View post   

>> No.73795825 [View]
File: 1.84 MB, 960x1440, 20276644-20240416114959-1101468248-25-960x1440.png [View same] [iqdb] [saucenao] [google]
73795825

>>73780198
>does it do her other outfits or is it just her base outfit? if so, will you do the others?
It does the base outfit and the sportswear one (to an extent... no green laces on the shorts because artists and drawing only the upper body, and the top doesn't always include text since art doesn't always feature the text either). I had actually forgotten to check the hoodie, but it seems to be there in a limited capacity due to only about 10 applicable images, sometimes inherits qualities of the sweater so I wouldn't say it can.
I'm not aware of any other outfits other than the first one that I've never seen. Doing multiple outfits in one dataset if doable in a somewhat limited capacity if they're different enough from each other to not have overlapping features and the number of images are balanced well enough. Also, I had no exclude some fan art due to, for some reason, artists drawing the hairclips featured in her 3d model, which would have likely ruined generation of the single hair clip from her live2d design. Once again, I'm not in a position to be able to do properly segmented multiclass loras for SDXL, due to the time required.

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