Lustimoon — Anima LoRAA celestial/lunar-themed original character LoRA trained on Anima. Silver-blue wavy hair, crescent-moon forehead mark and hairpin, matching choker, and a signature chain-detailed lingerie-style outfit with sheer draping panels.Trigger word: <code>lstmn</code>Using the trigger word alone is enough to reproduce the full character — hair color, jewelry, and outfit are all baked into the token, so you don't need to spell any of that out in your prompt. Just add pose, background, and lighting to direct the scene.Recommended settings:<ul><li>Resolution: 1024×1024 (or other square/portrait ratios close to it — see landscape note below)</li><li>CFG/guidance: ~4</li><li>Sampler: euler_a or similar</li><li>LoRA strength: 0.8–1.0</li></ul>What it handles well:<ul><li>Plain/studio backgrounds, standing and sitting poses</li><li>Back view (<code>from behind</code>, <code>back view</code>)</li><li>Close-up and portrait framing</li><li>Night/moonlit/fantasy-forest scenes</li><li>Outfit swaps (works, but combine with familiar lighting for best results — see below)</li></ul>Known limitations:<ul><li>Aspect ratio: trained entirely on portrait/square images. Landscape/widescreen generations can lose identity fidelity — stick to portrait or square, or generate portrait and crop/outpaint afterward.</li><li>Strong directional or colored lighting (sunset, backlighting, golden hour) isn't well represented in training and may cause the character to drift toward a generic look. Flat, moonlit, or studio lighting is most reliable.</li><li>Framing can lean toward closer crops (bust/half-body) at higher LoRA strength — lower strength slightly or specify <code>full body</code> explicitly if you want consistent wide shots.</li></ul>

Files with SHA256: 02f58e4ab2900a3915a356981f286c9e7909f686fc587f7212d649e830deb88b