A style lora from the work of criis-chan (no, not the person you're probably thinking of). The artist's works cover a wide swath of Danganronpa lore and preserving that was one of my main concerns for the 1.x to 2.x branches of this lora. WithBasically my hope for the lora is to balance the style options from the artist with the ability to inherit some danganronpa character design and theming.3.0 Notes (mainly for Krea 2):With the 3.0 branch of the lora I wanted to allow for the fact that the artist has multiple individual styles in addition to some main thematic and stylistic elements. The standard start of "ccdanganart , In the ccdanganart style." will provide the basic foundation. But adding one of the following phrases should allow you to steer it a little further into the desired direction."Rendered in a muted dusty palette.A high-quality anime illustration in a stylized digital painting style with crisp line art and soft cel-shading.A loose, painterly anime illustration rendered with sketchy watercolor textures and muted sepia line work.A cel-shaded anime illustration rendered with clean black line art and soft flat colors.Clean digital anime illustration with crisp line art and soft cel-shaded pastel colors.A cinematic anime illustration rendered with soft cel-shaded gradients and filmic lighting.Anime illustration rendered in soft cel-shaded tones with painterly gradients.Cel-shaded anime illustration rendered with crisp dark linework and flat, smooth color fields.A loose ink-heavy anime illustration with expressive linework and selective flat color.Sketchy semi-realistic anime illustration with textured cross-hatching and muted digital shading.Anime-style digital illustration rendered in soft, painterly manga shading.A cel-shaded anime illustration.Rough sketch anime illustration in a loose pencil-drawn style with visible construction lines and soft muted cel-shading.A cel-shaded anime illustration in the distinctive Danganronpa style.Anime cel-shaded illustration with thick dark lineart and vibrant saturated colors.Soft anime-style digital illustration.A dynamic cel-shaded anime illustration in a clean digital painting style.A painterly anime-style illustration." <h1 id="krea-2-notes"> Krea 2 Notes</h1>The Krea 2 lora marked the start of a new 3.0 branch of my underlying dataset. So far that aspect of it seems to be working fine in my testing. The support for Danganronpa characters is surprisingly serviceable as well. Krea 2 on its own isn't up to the level of anima or illustrious. But it does have a rough idea of some of the most popular characters.This lora seems to do a good job of helping to push that further into a more faithful direction while still not 'forcing' adherence to any outfit, hair style, etc. Monomi/Usami tends to be fairly rough even with this lora, as seen in the gallery she winds up slightly disquieting, but all in all I'm happy with the nice danganropa boost it provides. The Krea 2 lora also seems to work quite well with my danganronpa character lora. If you're using the character lora at 1..0 strength there can be some light mixing of styles. So you might want to push a character lora down to 0.7 or 0.8 while keeping this style lora at a 1.0 strength. However, for the most part I'd say that the results of just tossing a character lora in with this at 1.0 is quite good. I actually like the mix of this style and the light Rui Komatsuzaki influence that comes from using my character loras at 1.0 with this. <h1 id="chroma-notes">Chroma notes</h1>I wasn't satisfied with my Chroma 1.0 lora so I tried again and the result is the chroma 1.1 lora. I do think the 1.1 version is a big improvement over the original chroma 1.0. However, it does have some issues. The biggest is that text will often appear if you're not using a negative prompt. Of course using negative prompts will also constrain you a bit in terms of lowered step loras. If you want to use one of these loras I'd suggest the <a target="_blank" rel="ugc" href="https://civarchive.com/models/2032955?modelVersionId=2300950">rank 48 Chroma Flash Heun</a> lora. Then go for a cfg of 1.3 or at most 1.4. And for the negative prompt use at least the following: "signature, watermark, signature, twitter_username, artist_name,watermark, subtitle,tumblr_username". I've had acceptable results with a sampler of Heun/Beta and step count of 10 though obviously more is better. Aside from style improvements, the 1.1 Chroma lora also has improved ability to prompt for Danganronpa characters over the 1.0 version. Still not great but a nice push in the right direction at least. You can push the style of the 1.1 lora into a more polished form by adding something like "A modern digital anime illustration in the style of a visual novel key image. The lighting is dramatic. The art features colored line art soft cel-shading mixed with airbrushed gradients and highly detailed expressive eyes with multiple highlights." to the prompt.<h1 id="usage">Usage</h1>Illustrious and Anima already have a strong knowledge of danganronpa characters so compatibility there is already a given. The chroma lora comes close to being able to replicate characters from the dataset rather than just the style alone. Not perfectly, there'll inevitably be issues, but close. For example when using 'only' the chroma lora Chiaki Nanami's distinctive backpack might just be a normal backpack, Junko Enoshima's bear hair orniments might not come out correctly, etc. The gallery should have some examples of how mixing in even a standard flux character lora with this style lora can work in tandem with this one. But in general 0.1 to 0.3 strength for a character lora, even a flux one, should help there.However, in general, this lora really is best seen as a style lora and not something specifically for danganronpa characters in isolation.<h1 id="tags:">Tags:</h1>The main tag you'll need is ccdanganart. Using criis-chan as an artist tag isn't required, it might help or it might hurt. The arist has made a number of pieces emulating the official Danganronpa style, and you can push in that direction with a tag of komatsuzaki_rui_(style). Whether you use the rui tag will heavily influence how "danganronpa" the results are.<h1 id="compatibility:">Compatibility:</h1>It seems to do well with my Danganronpa character loras. When doing so I'd suggest lowering the character lora strength to around 0.3 while keeping this lora at 1.0.<h1 id="negative-tags:">Negative tags:</h1>The Illustrious and Chroma (but not Anima) lora does have a tendency to overlay text on top of the images. You can lower the chances of that happening by using these tags as negatives: signature, watermark, signature, twitter_username, artist_name,watermark, subtitle,tumblr_username

Ccdanganart (style and Danganronpa booster)

Ccdanganart_Krea2_krea2_3344811_epoch_20.safetensors

Files with SHA256: 6c069bcc4af33fa497a74f17921ea6e8fb312b658332ded071abdec934442f14