Saw the bounty from <a target="_blank" rel="ugc" href="https://civarchive.com/bounties/4939/diana-allers-pony-xl">https://civarchive.com/bounties/4939/diana-allers-pony-xl</a>Diana Allers from Mass Effect 3.<h2 id="multi-concept-lora-5p7bsevl8">Multi-concept LoRA</h2>The LoRA is able to output the original 3d art style, GTA gamesplash art style (the ones you see of her in fan arts, fandom etc) and photorealistism. As per the bounty request, the LoRA is also able to do 2 outfits: default sci-fi dress and BDSM.See how you can prompt this in the version details.As always, let me know what went well for you by posting your generations in the gallery below.

Diana Allers ME3

Trained on Pony Diffusion V6 XL Prompting styles to get the different concepts<ul><li>Photorealistic<ul><li>Prompts: realistic, supermodel, extremely detailed</li><li>Pony scores: score_9, score_8, etc, will better prompt the generation for photorealism.</li></ul></li><li>GTA Art Splash<ul><li>Prompts: photorealistic (for some reason the LoRA distinguishes "photorealism" differently from "realism")</li><li>Negative prompts: 3d</li></ul></li><li>Original in-game 3d style<ul><li>Prompts: 3d</li><li>Negative prompts: realistic</li></ul></li><li>Default sci-fi dress outfit<ul><li>Prompts: tank top, science fiction, dress, bodysuit, cyberpunk</li></ul></li><li>BDSM<ul><li>Prompts: bondage, bdsm, ropes</li></ul></li></ul>Character Trigger Words: diana allers, brown eyes, black hair, large breasts, cleavageIt also helps to include "missing eyes" in the negative prompts or include a VAE, as the LoRA occasionally outputs the eyes poorly (mostly due to the original 3d polygon art style dataset).

Files with SHA256: 84b8670f64e78af23e16e877bfbfa4c9a332848ec04557967171e779692c8b47