Used a different training strategy than my previous Illustrious models. SilvermoonMix-Illustrious-Evolved is good at:
Stable hands
Neutral art style to work with artist tags
Multiple characters
For instructions on how to use the V-pred model, please refer to https://civarchive.com/models/833294/noobai-xl-nai-xl
Recommended settings:
Sampling method: Euler a
CFG: 4-5
Start with no negative prompt. Sample images were generated without negative prompt.
Use ADetailer extension to enhance facial details: https://github.com/Bing-su/adetailer
推荐设置：
采样方法：Euler a
CFG: 4-5
先尝试不使用负面提示词。示例图片都没有使用负面提示词。
使用 ADetailer 扩展来增强面部细节：https://github.com/Bing-su/adetailer
Change notes:
v6.0: Further enhanced lighting, prompt obedience and compatibility with LoRAs that are trained on other models.
v5.0: Lights looks more nature now. Improved logic.
v4.0: Improved compatibility with LoRAs that are trained on other models.
v3.1: Improved logic, especially background. Adjusted saturation.
Description
FAQ
Comments (13)
Evolved和前面的01、02、03有什么区别
是不一样的训练理念。01-03是通过我人工手选的训练素材来强化某一个方面，是以我的喜好、设计为主的训练。
Evolved是不怎么筛选训练集，在训练中让AI动态调整，人工只在最后阶段干预epoch的选择。从结果上来说是尽可能减少了人为喜好，显得没有特点，但是综合肢体、逻辑会有很大提升，对画师串的适配也会更好。
Does this model have Illustrious XL 1.0 as a base?
This model is trained on Noob V-Pred v1.0.
No Epsilon for V2? If so that's a shame.
Hi @Giangiotto , because of how this model is trained, retraining the E-pred version is very time-consuming, and I'm not planning to do it for now.
This came at the perfect time for me, going for a more neutral-style v-pred workflow has been my focus for the last couple weeks! Can't wait to try it out.
After a few hours of testing, I really enjoyed experimenting with this model! It has its ups and downs, but in the v-pred scene, variation is a good thing. I tested it both as a base model and as a refiner layered over the NAI v-pred, swapping at 50% steps. The model adds a ton of detail to images and works especially well with certain artist tags (leaning more toward painterly styles while struggling to adapt to sketchier, pen-based styles.)
One of its biggest drawbacks is that it didn't handle extremely dark or bright prompts which is a big draw for v-pred models. That said, it’s one of the best plug-and-play v-pred models I’ve tested so far. As a base it gave really cool (albeit less stable) compositions and backgrounds and as a refiner it enhanced details really well.
@randomlygenerated Hey, thank you for the tests. I really appreciate.
Image post by aertt6950 | Civitai
I've downloaded version 2.0 twice, and it doesn't generate images properly.
----- Edit update: I just downloaded v1.0, and v1.0 works fine.
Hi @aertt6950, for instructions on how to use V-pred models, please refer to https://civitai.com/models/833294/noobai-xl-nai-xl
@silvermoong Well, thanks for the heads up. I just realized that.
It belongs to the NoobAI series.
But since 2.0 isn't in the Illustrious series, why would you bother locating it within the SilvermoonMix-Illustrious-Evolved thread?
I want to add this resource with images I made, but CivitAi can't find SilvermoonMix-Illustrious-Evolved in resources.