Used to not like the design on this particular costume, trained the model just for completion's sake, and ended up starting to like it while generating images to use for the model's post LOL I do need to warn that there are hardly any good images of her (which is by itself not surprising, considering she started out as an antagonist), so there would be the following issues:<ol><li>Generating her with helmet, considering not just limited images in training data with helmet (should have grabbed some screenshots from the television series) but also the fact that the helmet has a mouth confusing the AI. Try "covered mouth" prompt and see if it helps. Admittedly, it does helmet with face visible extremely well though, maybe thanks to the helmet having a freaking mouth</li><li>I also didn't even try with just bodysuit and leotard, and without armor, since there were no images in the training data set at all for this.</li><li>There's four diagonal golden parts in her armor at the front which are not in the back, but AI can't seem to distinguish front from back and reproduce the four golden parts in the back anyway *shrugs*</li></ol>Would greatly appreciate it if you could read this before posting! <a target="_blank" rel="ugc" href="https://civarchive.com/articles/7060/look-both-ways-before-you-crosspost">https://civarchive.com/articles/7060/look-both-ways-before-you-crosspost</a>Main prompts:<pre><code>sazervelsou, red bodysuit, purple leotard, belt, armor, gauntlets, black gloves</code></pre>Add for anime images:<pre><code>source_anime</code></pre>Add if doing "full body" or "feet out of frame":<pre><code>boots</code></pre>For helmet (issues listed above):<pre><code>helmet, visor, wearing helmet</code></pre>For weapon:<pre><code>shield, buckler, dual wielding</code></pre>

Files with SHA256: e4740c56ba6b33b4fa30fba8233158d8b93bc8896b54260371657df71c85f8a6