A style lora trained on a mix of various works by Rui Komatsuzaki (the Danganronpa artist). I used a general base from the highest quality work either from him or emulating his style on danbooru and then mixed in a small sample from the various Danganronpa and Hundred Line loras I've made. This also ensured a nice mix of tag and narrative caption styles so in theory at least both should be suitable for use with this lora.The most basic tags to use the style are: komatsuzaki_rui,komatsuzaki_rui_(style)However, there's a fair amount of variety in the images I used that highlight different projects he's worked on. These can be mixed and matched. Right now I feel that the best results come from prompting for the danganronpa 3 anime, the danganronpa sprite style, and the hundred line sprite style. To go for all of these at once you'll want to use this: komatsuzaki_rui,komatsuzaki_rui_(style),danganronpa_3_(anime),danganronpa_3_(anime)_(style),danganronpa_(series),anime_screenshot, official_artwork, the_hundred_line_-last_defense_academy- , danganronpa sprite styleIf you're getting a style too similar to the in-game style found in the dangarnonpa games you could try removing " danganronpa sprite style" from the tags.<h1 id="update-notes">Update Notes</h1>Chroma V1.1: I've had some time to get used to Chroma and wasn't happy with the state of the 1.0 Chroma lora. V1.1 is a lot larger in size but should have better style adherence than the Chroma 1.0 version. However, 1.1 seems to not do as well with pure tag prompts. I'd suggest either using narrative descriptions for prompts or a combination of narrative prompts and tags. If you want to push slightly away from the more in-game sprite style try adding something like "The anime art style utilizes sharp outlines and distinct shading with soft lighting."<h1 id="different-models-and-usage-scenarios">Different models and usage scenarios</h1>You'll see fairly significant differences in performance between the different models I've trained this for. Illustrious will presumably provide the best results because there's so much danganronpa material on danbooru. While use with chroma and flux models will generally benefit from the addition of character loras if you're trying to work with one of Rui's specific franchises. If you're 'only' using this lora in Chroma then you 'might' be able to get specific characters. But in general this really is just meant for style rather than as a "danganronpa character lora" in and of itself. Meanwhile you'll almost certainly need to use a character lora with the Flux lora if you want to specifically prompt for any of them. I didn't want to risk overtraining of the Flux model and erred on the side of just going with style for it without actively trying to bake in any of the characters. You 'might' be able to get some of them from the Flux model. Chiaki in particular is heavily represented in the training data due to her strong role in the despair arc of Danganronpa 3. But even then you'll be better off just using a Flux character lora alongside this style lora if you want to prompt for her in Flux. Though the combination of this Flux style lora, my Danganronpa anime style lora, and my Flux danganronpa character loras does seem to produce especially good results.

Rui Komatsuzaki Style [Danganronpa Style]

Files with SHA256: 561af5082366370c418e75d716ad0fa719ac70b16f713ed60201227adeee73f0