Regarding the llustrious versionI think this Lora has met my requirements, replicating the artist's style of <a target="_blank" rel="ugc" href="https://x.com/kafun">かふん @ kafun</a> paintingAll images are directly generated without the use of ADetailer modificationRegarding the pony versionI have adjusted and trained this dataset many times, but I still cannot achieve the style of the original artistHowever, there are quite a few outsiders in this dataset, and the generated image style is also satisfactoryPay attention to the lack of coordination in the limbs, it may be due to the artist's exaggerated style

Files with SHA256: c6250f1b1f26776376d4318a7ba124453ce71ebef8cc4c1d25187d8b0519e613