<h3 id="heading-594">V1.1</h3>Small update that should make the model more flexible. For further details check V1.0.To make this any better I'd need a larger dataset and manga pages/anime screenshots aren't ideal for that.For prompt inspirations check the examples or the sidebar.<h3 id="heading-595">V1.0</h3>Props to holostrawberry for the lora guide <a target="_blank" rel="ugc" href="https://civarchive.com/models/22530/guide-make-your-own-loras-easy-and-free">[Guide] Make your own Loras, easy and free - colabs | Stable Diffusion Other | Civitai</a>I recommend a weight of 0.85-1. Most examples including other artstyles were created at 0.9. For comparison check the XYZ plots.This model of Suzuna from Eyeshield 21 - the first major manga illustrated by Yusuke Murata - was made with only 22 images but is quite flexible. It works with different clothes, art styles, realism and NSFW.Depending on the weight and prompts it can have artifacts at 12 steps using DPM++ SDE Karras but gets more consistent towards 20. When in doubt, hires fix.Please leave a review if you enjoyed this Lora or have problems with it. I'd also be interested in how the lora behaves with other base checkpoints so why don't you give it a try and throw some images down below!

Suzuna Taki - Eyeshield 21 | Character Lora

22 images trained in batches of 2 with 10 repeats each

Files with SHA256: cb94a6dba68002642224924835bf21b85ff68360cf3cb160373276c0f9f12a69