2D Soul PLUS XL V2 Heaven of Drawing - 2D Soul upgraded model! Improved performance!

0
0
0
0
AnimeAnime CharacterGirl
최근 업데이트: 첫 번 게시:
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info
Anime,Anime Character,Girl,Checkpoint,SDXLImage info

After nearly a year, the most important model for the author, the Two-Dimensional Soul PLUS V2 version, has finally been released!

Its original version was a MIX model with excellent fusion effect of the generated images, so much so that it became the mother of LORA,

almost all LORA models are trained based on this model.

This V2 version is a training version,

the difficulty coefficient of the generated images has been qualitatively improved.

No matter how complex the composition is, large models have a mechanism for simplification.

That is to say, even if large models train very complex compositions, they will also relatively "optimize" the content of the training set.

Therefore, it's quite difficult to make the large models capable of generating complex compositions.

Therefore, the V2 version has thousands of appropriate generated images to counterbalance and optimize the solution.

V2 is a model that can generate complex compositions.

More possibilities need to be explored comprehensively by everyone.

This release hopes to be opened for download after a round of recommendations.

The author's preferred output resolution is generally 536X960, if you set it to 540 it will automatically become 536,

width 664, 768 are all acceptable.

All CFG values are set to 5,

using SDXL's VAE,

and enable the ADetailer plugin.

No need for high-definition restoration, but of course, there is no problem if you use it.

Not only for generating images separately, it is also recommended to use it as a basis for LORA, adding a layer of generated image buff~

토론

가장 인기
|
최신
보내기
곧 오픈
다운로드
(0.00KB)
자세히
유형
온라인 생성 횟수
0
다운로드
0
추천 매개변수
Sampler method
CFG
5
VAE
없음

갤러리

가장 인기
|
최신