Project Zeus SD3

0
0
0
0
CGI styleAnime SceneGirlObject Enhance
최근 업데이트: 첫 번 게시:
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info
CGI style,Anime Scene,Girl,Checkpoint,SD 3Image info

Project Zeus SD3 V1


After many weeks of training and testing a lot of things, I can finally release this V1 version, and there is good news and bad news.


I must say that SD3 is a very capable model, and what it does well, it does very well, it even does some incredible things that sometimes amaze me.


The serious problem is that it is a model that has been released broken, rushed and half-finished and what it does badly, it does too badly and what is worse, it has no solution.


It is impossible to fix the anatomy, the hands, the limbs of animals in general, not with training, not with LoRa, not with embedding, not with anything. The model is completely broken and bugged. It is simply half-done.


I hope that Stability AI gets back on track and really releases a version 3.1, but at this point, I personally highly doubt it. As the model is, it would be like making a practically new model again. But that will be seen, we will always have Flux1.


It's a real shame, because SD3 makes incredible landscapes, very detailed portraits, recreates styles very well (the anime and cartoon are especially very effective), but if you try to make a ****ing decent hand at it, you can spend the whole day without achieving it.


Well, nothing more to add, I hope that this v1, which will obviously be the last, is to your liking and that you can get something positive out of it. A hug to everyone and see you in the next updates of my Flux1 models, project Gaia (Dev) and project HaDeS (Schnell).


P.D. It's a no clip verson.

토론

가장 인기
|
최신
보내기
곧 오픈
다운로드
(0.00KB)
자세히
유형
온라인 생성 횟수
0
다운로드
0
추천 매개변수
Sampler method
CFG
4.5
VAE
없음

갤러리

가장 인기
|
최신