Online generation
Recommended Settings:
Sampling Method: DPM++ 2M SDE
Schedule Type: SGM Uniform
CFG Scale: 6-4
建议设置:
采样方法：DPM++ 2M SDE
Schedule type：SGM Uniform
CFG Scale：6-4
If you like this model, you can support me here.
If you'd like me to develop a personalized LoRA model tailored to your needs, check out my custom service offerings on Patreon!
Description
Art Style Tag List:
- https://huggingface.co/rim0/GENESIS/blob/main/Art%20StyleV4.txt
Nji Style Tag List:
- https://huggingface.co/rim0/GENESIS/blob/main/niji%20StyleV4.txt
Lighting Effect Tag List:
- https://huggingface.co/rim0/GENESIS/blob/main/Lighting%20Effect.txt
Character Tag List:
- https://huggingface.co/rim0/GENESIS/blob/main/CharacterV3.txt
FAQ
Comments (24)
Great! It's updated! Thank you for sharing.
群友来支持了喵，很强
holy shit I have to test this <3 What's new with V2?
Wider Style Variety: V2 incorporates a much larger and more diverse training set, so it can handle a significantly broader range of art styles and concepts.
Better Prompt Comprehension: It should have a slightly better understanding of natural language prompts compared to V1.5, making it a bit easier to get what you want.
Think of V2 less as a direct 'upgrade' and more of a 'sidegrade' with a new aesthetic balance. So while the overall range is much wider, you might find some styles feel a bit different compared to V1.5.
Thanks for the enthusiasm, hope you have fun with it! (｢`･ω･)｢
@EWIGKEIT0 Sounds awesome, can you give us an updated txt file on the art styles? Or is this not affiliated to new trigger prompts?
@deitychaser To be honest, the main reason I haven't updated the style list is that after some testing, I felt a lot of the new triggers weren't fitting/responding as well as I'd like yet, so I didn't want to release them "officially."
But since you asked, I'll go ahead and organize the triggers. I can probably post a public list in a few days.
@EWIGKEIT0 I see, yeah, take your time. I can rather say that some older artist prompts are now way stronger though. I can say that it reacts way better to joycaption "art critic" descriptions than NTDMIX 1.5 - like really a lot of improvement in terms of understanding how coloration should be. Character knowledge is a double edged sword, it seems. I have problems with falin touden \(chimera\) from dungeon meshi, which worked really well in 1.5 but in v2.0 its almost as if it doesn't know her although she has lots of fan art on danbooru. Chibaolin from artstation seems to be much stronger now, he has a dramatic influence on 3d-ish facial features that I really don't like while in NTDMIX 1.5 i'd get the cool drawing style but still remain a somewhat anime-esque 2d facial structure. When I tried to merge Genesis v1 and v0.4 on some of my models I noticed as well how fast the model can be overblown and default more in the 3d direction. Maybe this happened with NTDMIX 2 to some extent as well? Of course, if this was your intendend goal then this is just my personal feeling.
V2 - is it EPS or V-pred?
@homoludens EPS
正常图片很强，不过 我生成nsfw的图就质量一般了，当然我还不是很会用这个模型，
我感觉是nsfw方面可能没有特化，
需要搭配lora
在不加lora的前提下，我跑出来的图基本都是偏向真实风格的，就算加了artist标签也是，不知道怎么回事😔
可以在prompt里加一句This is a digital illustration in an anime style
我想问下好的自然语言提示词怎么写，我词语太贫乏了，想不出好的画面
I don’t know why I never tried this model. Unusual gens and unique styles. 10/10
为什么设置clip-1就会出错误图，奇怪
You model aren't allowing to high fix or anything else in site... (After gen)
It is best to use it in comfyui
@qian_Q i use site generator (sadly)