[ 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.74830960 [View]
File: 1.16 MB, 896x1152, 1710623807841940.png [View same] [iqdb] [saucenao] [google]
74830960

>>74827286

>>74821434
https://files.catbox.moe/nb5jdz.png
https://files.catbox.moe/g6472a.png
https://files.catbox.moe/823tyi.png
https://files.catbox.moe/6vfw81.png
https://files.catbox.moe/r5ofhy.png
https://files.catbox.moe/gjsl8j.png
Scuffed bonus: https://files.catbox.moe/kl2w59.png

>>74828416
I don't think NAI looks so much better than the other models that you couldn't achieve the same level of results with the other local models available now, also the dataset is 8-9 months outdated at best. You really notice how limited it is when you want to gen someone that isn't in Hololive or Nijisanji.
I don't believe that the things it does significantly better than the other models are just due to SDXL, they also didn't release their research for NAI v1 until it got leaked so we can't really say for sure if they're doing anything extra, but for me it's really hard to believe that the stuff it can pull off is just due to a bigger dataset.

>>74829189
And this also, they delivered with Anime v3 but if you've been keeping an eye on the company as a whole they're not exactly inspiring confidence right now

>>74830716
lol

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