By popular demand here's Sesshomaru from Inuyasha. Excuse the delay, I had technical problems and I had to do a lot of redoes because the AI didn't like his face markings. Same as <a target="_blank" rel="ugc" href="https://civarchive.com/models/20584">Inuyasha</a> if you would like a closer to anime look, use <code>official style</code> in your prompts. If he loses any of his characteristics you can help him restore them with: <code>long hair </code>,<code>white hair</code>,<code>fur trim</code>,<code>yellow eyes</code>,<code>white kimono</code>, <code>armor</code>, etc. I recommend to use him at 0.8~1 and to use hires with 10+ steps, as his face is complicated and ends up like a blurry mess a lot. If he has too many markings, use higher weight or just use very low seed variation to get back to two per cheek. I'll probably keep making more characters and other kinds of LoRAs, if you want to support me, you can go to <a target="_blank" rel="ugc" href="https://www.deviantart.com/paintingpandaart ">my wife's DA </a>where she has a subscription and is the same as me having a kofi or similar.

Sesshomaru - Inuyasha | Character

Trained on 60 images, for 12 epochs with Adam8bit optimizer and 15 repetitions. I might try in the future to use Adafactor and/or LoHA.

sesshomaru_v2.safetensors

By popular demand here's Sesshomaru from Inuyasha. Excuse the delay, I had technical problems and I had to do a lot of redoes because the AI didn't like his face markings. Same as <a target="_blank" rel="ugc" href="https://civitai.com/models/20584">Inuyasha</a> if you would like a closer to anime look, use <code>official style</code> in your prompts. If he loses any of his characteristics you can help him restore them with: <code>long hair </code>,<code>white hair</code>,<code>fur trim</code>,<code>yellow eyes</code>,<code>white kimono</code>, <code>armor</code>, etc. I recommend to use him at 0.8~1 and to use hires with 10+ steps, as his face is complicated and ends up like a blurry mess a lot. If he has too many markings, use higher weight or just use very low seed variation to get back to two per cheek. I'll probably keep making more characters and other kinds of LoRAs, if you want to support me, you can go to <a target="_blank" rel="ugc" href="https://www.deviantart.com/paintingpandaart ">my wife's DA </a>where she has a subscription and is the same as me having a kofi or similar.

&lt;p&gt;Trained on 60 images, for 12 epochs with Adam8bit optimizer and 15 repetitions. I might try in the future to use Adafactor and/or LoHA.&lt;/p&gt;

Files with SHA256: 7d1e63ad9d1b6e49333cd3842d3b586b57e90640935615a17b1c91579886d1f6