<p><em>A lora trained on </em><strong><em>pony realism</em></strong><em> as a foundation, using images of the nsfw PvC figure collection produced by Shibadon Studios; and a few others to help bolster what it can do. The dataset has a very strong male bias or towards more muscular characters.</em></p><p><em>Lord knows I could never afford a real one so this will have to do!</em></p><p></p><h2 id="prompts:-shibadon-figurine-plastic-body-action-figure-on-a-desk-diorama-bara-muscular-qb5a4fvzk"><strong><span style="color:rgb(250, 82, 82)">Prompts:</span> shibadon, figurine, plastic body, action figure, on a desk, diorama, bara, muscular,</strong></h2><p></p><h3 id="additional-notes:-this-model-is-trained-on-pony-realism-so-i-suggest-using-that-for-the-desired-effect.-w0tgt0th2"><strong><span style="color:rgb(250, 82, 82)">Additional Notes:</span></strong> This model is trained on <a target="_blank" rel="ugc" href="https://civarchive.com/models/372465/pony-realism">pony realism</a> so I suggest using that for the desired effect.</h3><h3 id="there-are-three-models-listed-here-because-i-trained-five-and-these-3-gave-good-results-but-some-better-than-others-for-various-reasons-so-i-figured-i-would-post-all-three.-3pyjv52e9">There are three models listed here because I trained five and these 3 gave good results but some better than others for various reasons; so I figured I would post all three.</h3><h3 id="i-do-however-recommend-primarily-using-the-v1-model-over-the-other-two.-this-one-provides-the-best-overall-results-the-others-have-their-uses-but-overall-are-not-as-stable.-hgvdy04c5">I <em>do</em> however recommend primarily using the <strong>V1</strong> model over the other two. This one provides the best overall results, the others have their uses but overall are not as stable.</h3>

SHIBADON STUDIOS FIGURINE [PONY REALISM]

<p>V3, overtrained test. Good results when it works, but sometimes just straight up breaks and causes artifacts or weird visual things.</p>

Files with SHA256: d001331f3f86577dd093f56f9af6b8cb062d5efbbe9377ecf77d5a2da2ee9a48