Kitty from Dorokei (ドロケイ) Prompts:-<ul><li>Character:<pre><code>kiti, 1girl, blue hair, side ponytail, blue eyes</code></pre></li></ul><ul><li>Headwear:<pre><code>eyewear on head, cat ear headphones, headset</code></pre></li><li>Outfit:<pre><code>pendant, sports bikini, underboob, gloves

Optional:
white belt, chain
thighhighs</code></pre></li><li>Jacket:<pre><code>jacket, bare shoulders

Optional:
black jacket (this helps if the jacket generated is not black)</code></pre></li></ul><ul><li>Mouth Expressions:<pre><code>skin fang, open mouth
skin fang, closed mouth
:3</code></pre></li><li>Game CG Style:<pre><code>anime coloring</code></pre></li></ul>Known Issues:<ul><li>Her pose and outfit might appear a bit inflexible because there is only one image (excluding alternative facial expressions) of her wearing the full outfit. I have already tried to increase the variety by including AI gens in the training dataset. Try using the NegPiP extension to remove a specific piece of the outfit if negative prompts don't help.</li><li>The jacket might be generated in the wrong color. Adding black jacket can help. I didn't tag it in my dataset mainly because of OneTrainer's 75-token limitation.</li><li>Sometimes, skin fang or other mouth expressions can result in multiple skin fangs.</li></ul>Afterword: I have been trying out OneTrainer's Prodigy Adv optimizer with Simplified AdEMAMix. According to the dev, this is good for small-batch training. Hopefully some of you will like this model.

[IL v0.1] Kitty キティ | Dorokei ドロケイ

This model was trained on Illustrious-XL v0.1.

Files with SHA256: 449abfe3c70c1bb68a59238a671464610681896a2631f225cd23faa2f6a8f099