Hello everyone, this is my first attempt at training a LoRA — would love to hear feedback and see anything you generate with it!<hr />Base Model: Anima 1.0 — Base checkpoint (<code>anima-base-v1.0</code>). Not the Aesthetic or Turbo variants.Recommended Weight: 0.9<hr />Trigger Word &amp; Locked Appearance<pre><code>kushinabcgo, long red hair, grey eyes, hairclip, blue bracelet</code></pre>Default Outfit (optional)<pre><code>green pinafore dress, sleeveless white collared shirt with zipper</code></pre>Optional add-on -- if the art style drifts too far from her trained look:<pre><code>(anime screencap)</code></pre><hr />Known Quirk: Some NSFW images come out with stockings For bare legs, add:<pre><code>bare legs</code></pre>For full-body shots where you want her feet/footwear visible:<pre><code>toeless footwear</code></pre>Heads-up: there's a small chance of a stray, fake watermark-style text appearing somewhere in the image. Keeping <code>watermark, text, logo</code> in your negative prompt is a safe habit.<hr />Suggested Generation Settings<ul><li>Sampler: Euler A, Simple scheduler</li><li>Steps: 12 / CFG: 2.5 (with Turbo LoRA --raise CFG/steps if not using it)</li><li>LoRA weight: 0.9</li></ul><hr />Trained on a small, hand-captioned 20-image dataset - my first LoRA, so there's plenty of room for a v2. Comments and generated samples are very welcome!

Files with SHA256: b7ac4a62ea662967fd1f9cde76104ec7c950a065fa20b875517016005aad9fb1