<h1 id="vodyanitsa-(genshin-impact)">ヴォジャニーツァ / Vodyanitsa (Genshin Impact)</h1>Character LoRA for Vodyanitsa — Water Imp and Prima Donna of the Korolevskiy Troupe.Trained on waiIllustriousSDXL v170 (Illustrious XL family).<h2 id="how-to-use">How to use</h2>Trigger word: <code>vdy</code> — that is all you need.<pre><code>1girl, solo, vdy, cowboy shot, standing, looking at viewer, simple background,
masterpiece, best quality, amazing quality, absurdres, newest</code></pre>Recommended weight: 0.8You do not need to write her hair color, eye color, head fins, hair flowers or her default outfit. All of that is bound to the trigger word. Just describe the pose, the framing, the background and the mood you want.<h2 id="what-is-baked-into-the-trigger">What is baked into the trigger</h2>Deep teal hair, blue eyes, hair over one eye, head fins, large white flowers in the hair, and her default white dress with the dark teal panel, detached sleeves and choker.Pose, camera angle, expression and background stay fully controllable from the prompt.<h2 id="measured-behaviour">Measured behaviour</h2>Verified with 20 generations per base model, trigger word only (no character tag, no outfit tags), identity checked against Danbooru reference images:| Base model | Identity match | Anatomy breakage | Aesthetic | |---|---|---|---| | waiIllustriousSDXL v170 (trained base) | 100% | 0.00 | 0.95 | | prefectIllustriousXL | 100% | 0.05 | 0.95 | | rinFlanimeIllustrious | 100% | 0.00 | 0.94 |Trigger-only robustness: 100%. NSFW identity retention: 100%. Outfit swapping (bikini / school uniform / gothic / santa): 4 of 4 succeeded while keeping her face and hair, so she can be dressed in something else without losing identity.<h2 id="weight-guidance">Weight guidance</h2><ul><li>0.8 — recommended. Best balance of likeness and clean linework.</li><li>0.6 — softer, likeness starts to drift toward the base model.</li><li>1.0 and above — likeness is strong but fine detail begins to smear.</li></ul>On other Illustrious fine-tunes, start at 0.6–0.8 and adjust.<h2 id="caveats">Caveats</h2><ul><li>Pose and composition variety leans toward upper-body framing, because the source</li></ul>artwork for this character is still mostly portraits. Ask for <code>full body</code>, <code>sitting</code>, <code>from side</code> etc. explicitly if you want more variety.<ul><li>Pony / SDXL-base checkpoints are not supported (architecture mismatch, untested).</li></ul>

Files with SHA256: 37a1c7db8b88a6474dce2d03b345f5f8495f7eaee4dabb6af526a117a3c5de19