・実験的VAE
・エンコーダーは調整していません
・画像にRefinerは使用していません
・NansException
Description
・茶色い！
・ぼやける！！
・アップスケール推奨！！！
・使い難い
・F0.51 = F1 + (c2 - F1) x 0.51_trainDifference
FAQ
Comments (5)
The versions aren't very clear. Going by the samples alone, some others look better than this one, but this one is newest. There doesn't appear to be any descriptions.. And since they're not like v1, v2, v3 ,etc. One cannot simply assume that they are version based.
Aye, I don't understand why some creators can't version their works in some easy to grasp way without having to consult a manual. In such cases I just download the latest version and call it a day. If you decided to name your versions "AF22", "GH32" and "extra-IT34" in chronological order, then yeah, not gonna solve a riddle just to figure out which one I might want.
@jtabox I just want whichever will put out the best images. But, I suppose it depends on model used , prompts and such as well. idk..
Just a supposition but looks like each major version has a new letter, and each subversion as a number
@bleakdragoon302 One way of looking at it, but without any explanation from the uploader, it's hard to say. each letter may include different additions to the dataset for all we know. Although it is safe to say that, at the very least, the highest number of each letter, is the best of that subset.