This lora was trained with an experimental approach to training using flux kontext to generate multiple images with different angles and scenes from a single image, and using Joycaption to do the initial prompting. The quality and versality of the lora is not great but it's probably has to do more with my knowledge of how to train a lora :( Main Prompt : cathcivv,long_hair, blue_eyes, drill_hair,twin_drills, plus the lora.As always I recommend using <a rel="ugc" href="https://civarchive.com/models/443821/cyberrealistic-pony">CyberRealistic Pony</a> or <a rel="ugc" href="https://civarchive.com/models/709404?modelVersionId=1871475">Cyberrealistic semi real Pony</a> to generate image, in addition to adetailer for faces (and hands !) Since I didn't prompt as well as it should be done, the Lora baked the character's background and dress from the game, it seems to be easily removable but if you don't specify a different settings for background/ scenes and clothes it might interfere.

Files with SHA256: 5a3aeef449e11d8e62e906f7a75e029303e5704ba52b31e2f8f89ef12ab0c888