Kingdom Death Character Art Style - Pony / Flux / Illustrious

Style of Kingdom Death Monster

Anima Users: please read the passages for those models below. There might be some interesting information for you there!

The board game Kingdom Death Monster miniatures come with small art cards. This lora tries to replicate this style. This style is trained on the sketch style (and not the smooth anime one).

Anima v2 After finding out what I did wrong (to many steps, and a reminder not to blindly trust LLMs even if it is Claude). I now have a version that better captures the style. It works without the quality tags, but I find myself that it adds to the image (But not really needed when using Euler A, except maybe sketch to get stronger lines).

The Trigger is: kdmstyle (Flux version has no trigger!)

Advised strength: 0.7-1.0

Anima v1 (best used with Eualer A) version is the first Anima lora that I have made. It feels a bit different, but this could be that I'm also new to Anima prompting and settings. I thought it was good enough to post, but not long after I decided to make a v2. (Future me: this one works well when using Euler A (schedular normal) sampling, I would not recommend ER SDE with this one unless you want that style)

Illustrious 2.0 version is now life. It is stronger and tends to create image with more expression thanks to the extended dataset, but it is a bit unstable, because of it. It is good with upper bodies and close ups.

The Flux version is made with the sponsoring from AikaKittie https://civitai.com/user/AikaKittie .Use natural language with this one for best effect!

Pony 2 is made with the same dataset, but any text and logos are removed from the images. (With help from AikaKittie https://civitai.com/user/AikaKittie)

Prompting Anima

Works without these, but using masterpiece, best quality, traditional media, sketch, graphite \(medium\), hatching \(texture\), in the prompt enhances the image in my opinion. For the negatives I use the following: worst quality, low quality, score_1, score_2, score_3, bad hands, mutated hands, extra fingers, missing fingers, extra digits, fewer digits, fused fingers, deformed hands, poorly drawn hands, bad anatomy, deformed fingers, hands with too many fingers, signature,

Prompting Illustrious

For all versions I recommend you use: masterpiece, best quality, amazing quality, very aesthetic, high resolution, sketch, in your prompt and I myself am using the following negative prompt: text, abstract, glitch, deformed, mutated, ugly, disfigured, long body, lowres, bad anatomy, bad hands, missing fingers, extra digit, fewer digits, cropped, lowres, (worst quality, bad quality:1.2), bad anatomy, jpeg artifacts, signature, watermark, username, signature, simple background, conjoined,

Prompting Pony

For all version I recommend you use: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, in your prompt and 3d, realistic, (english text, signature:1.5), censored, mutated, bad hands, deformed hands, missing fingers, multiple fingers, multiple digits, extra digits, in your negatives. It will greatly improve the results.

With the way this lora is trained you need to use `(english text, signature, Watermark, artist name, logo:1.5)` in the negatives. Because the dataset had the English text and logo stil in it. I will try to clean this up in a future release.

Advised strength: 0.7-0.8

For more information about the prompts you can take a look at the showcases of the models. These settings worked for me to get the best result, but may vary for others!