<h2 id="don't-expect-it-to-be-perfect-that's-it.-iog5g8y2n">Don't expect it to be perfect, that's it.</h2><p>Support me: <a target="_blank" rel="ugc" href="https://ko-fi.com/elevatedone">https://ko-fi.com/elevatedone</a></p><h3 id="recommendations-pony:-ofz2z04nt">Recommendations Pony:</h3><ul><li><p>LoRA weight: 0.6-1.0</p></li><li><p>Negative prompts: Use negative embedding for pony.</p></li><li><p>If you use "red cape", it will be there, if not, then not. The same for the nail (use "sword" in prompt, although it's not really stable).</p></li><li><p>Since it was trained on the same dataset as SD 1.5 (r34 mostly) and Pony can do porn much better</p><ul><li><p>If you want her to be naked, use "naked, black body" in the positive</p></li><li><p>If you want her to be not naked, use "black bodysuit" in positive (or something alike) and "naked, nude" in the negative</p></li></ul></li></ul><h3 id="recommendations-sd-1.5:-ks6w75j5l">Recommendations SD 1.5:</h3><ul><li><p>LoRA weight: 0.6-0.8, 0.8 is the most stable</p></li><li><p>Do not use long prompts, it creates disfigurement</p></li><li><p>Negative prompts: <a target="_blank" rel="ugc" href="https://civarchive.com/models/4629/deep-negative-v1x">Deep Negative V1.75T </a>and <a target="_blank" rel="ugc" href="https://civarchive.com/models/116230/bad-hands-5">Bad Hands 5</a> TIs, prompts like: signature, multiple views are also recommended</p></li><li><p>Positive: can be anything, just not too much</p></li></ul>

[LoRA] Hornet (Hollow Knight) (SD 1.5 & Pony)

<p>Trained on around 300 images which were picked by hand from multiple image boards</p>

Files with SHA256: 2a093451b8590bf3ce67287568daba110a5df08c05c70502a61cdc3c6fd82fec