<h1 id="read-this-before-using-or-rating-this-lora!-739j1iz73">READ THIS BEFORE USING OR RATING THIS LORA!</h1>This is my first ponyXL experiment. The detaset also isn't perfect so sometimes the results might be wonky or the characters will look all similar if not the same. If you want to help me improve this model, You can give me some characters (ocs are allowed) for me to recreate in VRoid Studio and modify in Blender so I can add them to the detaset. I'm willing to do this for free so please don't give anything too complicated, Thank you!I decided to train the AI to learn my 3D VRoid model style. So yeah, I assume you read the message above, so yeah!<h1 id="info:-a2zgggo6y">INFO:</h1><ol><li>Keyword is "cappyadams". but I believe it works better when typing "by cappyadams"</li><li>generating male characters is harder at the moment.</li><li>tested with PonyXL TurboDPO but it should also work with the main pony model.</li><li>recommended lora weight is 0.8</li><li>Try experimenting with this lora if the results are weird (remember, this lora is an experiment.)</li><li>Tested in SDNext. should would in auto1111 and comfyUI</li></ol><h1 id="contact:-s76hgrdl6">CONTACT:</h1><ul><li><a rel="ugc" href="https://www.deviantart.com/cappyadams">DeviantArt</a></li><li><a rel="ugc" href="https://x.com/CapdamAI">Twitter</a></li></ul>

CappyAdams / Artist style LoRA(EXTREME BETA)

First attempt at making a FLUX loRAShould be a little more stable then the Pony versiontechnically doesn't require a trigger word... but starting the prompt with "this is an animated anime image of (whatever you want)" does help. removing the word anime increases realism, keeping the word anime decreases style accuracy.

Files with SHA256: 0b10a3ead2d10d7795037bc8c6a14fe34b968f2d8d65ccc4f8efce1a7d09a421