A quick first test of training a lora for Z-Image. Considering Z-Image is not primarily an anime model it turned out quite well.This was trained on images generated by the Illustrious checkpoint Dvine 6.0.The trigger word is only 'Holo Wisewolf', but I find it works more consistently when also describing some of her features.<pre><code>Holo Wisewolf has wolf ears, a wolf tail, and red eyes. Holo Wisewolf has long light brown hair with sidelocks and blunt bangs.</code></pre>The model picked up on a couple of her outfits as well as her pouch, and will reproduce them somewhat faithfully if you call them out. Don't take these descriptions as gospel or trained in this specific form, they are just what I used that worked.<pre><code>She has a small red pouch hanging from her neck.</code></pre><pre><code>She wears a black vest with black fur trim over a blue long sleeve shirt, with a black fur-trimmed black skirt and a red sash tied around her waist. She has a small red pouch hanging from her neck.</code></pre><pre><code>She wears a long purple dress with long sleeves and a red sash tied around her waist. She has a small red pouch hanging from her neck. She wears simple brown shoes.</code></pre>

Files with SHA256: c306e8b4229e3a50a36a07eea425b74f71b7ac7694d3b9205ecd9b66cb75f36a