for base v010

v2版本在v3依旧可以正常工作,但在basev1上存在偏色问题,因此重新在base v1版本以1280px分辨率训练

The v2 version still works properly on v3, but has color deviation issues on base v1. Therefore, retraining is conducted on the base v1 version with a resolution of 1280px.

实际上个人判断训练结果并不如v2上的表现,非常遗憾.

In fact, I personally think the training results are not as good as those shown in v2, which is quite regrettable.

另外,旨在为风格混合时提供特殊的质感,可以考虑使用低lora权重以及高触发词权重In addition, to create a special texture when blending styles, you can use low LoRA weights and high trigger word weights.

for preview v2 and v3

触发词@Nyte,推荐与@dino共同使用,封面2的ninomae ina'nis为混合风格,封面1的platinum \(shimmering dew\) \(arknights\)为该lora在1权重时的效果

Trigger word @Nyte,Recommended for use with @dino. Ninomae Ina'nis on cover 2 features a mixed style, while Platinum (Shimmering Dew) (Arknights) on cover 1 shows the effect of this LoRA at a weight of 1.

https://civarchive.com/models/1805424/nyte-tyde-style-or-illustrious

从该LORA模型蒸馏,在anima preview v2上以lora算法进行训练,纯tag标签,参杂了带勾线的作品,对打光做了标注

Distilled from this LoRA model, trained using the LoRA algorithm on Anima Preview V2 with tag labels only. incorporate works with line art, and annotate the lighting especially.

存在问题:Known issues:

1.ninomae ina'nis的tentacle hair会受到影响不能正常生成

Ninomae Ina'nis's tentacle hair may be affected and fail to generate properly.

2.受训练素材印象,手指可能会比较暧昧

Due to the influence of training materials, fingers may appear ambiguous.

3.模型的构图似乎受到一定程度影响,会倾向于生成近景的特写.这与我在v1上训练类似素材的经验有出入.但可以通过约束lora跳过前面60%阶段生成解决这个以上3个问题

he model's composition is somewhat affected, tending to generate close-up shots. This differs from my experience training similar materials on V1. However, all three issues above can be resolved by constraining the LoRA to skip the first 60% of the generation phase, especially when used with @Dino.