Note: I am not a native English speaker. The following content is all from a translation software. Please forgive any possible grammar errors.

Kouga, from DragonBlast SE, a fighting game that is a rarity to this day, but unfortunately has been discontinued for a long time, and there is very little information about this game on the Internet.

The training of this lora mainly relies on the original game drawings and a few pictures and is iterated. For this reason, this lora has the following characteristics.

① The model has completely different 2D and 3D representations. When generating 3D images, the output image style will be highly similar to the game. There are two 3D-style images in the uploaded preview for reference.

② Under normal circumstances, characters will be in a slightly hunched posture, which is due to the fact that the training heavily relies on game images, and in the game, the characters are in this posture. Therefore, when using it, please appropriately add modifiers for the posture to avoid this problem.

③ Because the training set has a single source and lacks high-quality images, this LORA is not good at handling overly complex movements and compositions. When the movements are too complex, obvious problems with the character's movements can be noticed.

④ Accidentally found that this model and black and white sketch style collocation effect is good. (Did you guys see the picture I posted? Oh, my God, he's cute.)

Key trigger words: [kouga, shark anthro]

Optional trigger words:

[loincloth] = The original clothing, which is the red loincloth. If this prompt word is not provided, the model is more likely to not generate this object.

[male, male focus, muscular, muscular male] = If these prompt words are not used, the output result may tend to be female. It is recommended to use certain gender restriction words to avoid unexpected outputs.

[3d model] = When you want the output result to be closer to the game scene, you can use this prompt word. Otherwise, the model will default to outputting 2D style images.

[little black eyes] = The eyes of this character are a difficult point to output. In some cases, the output eyes may appear larger. Using this prompt word can make the eyes closer to the original character, but the effect is not obvious.

[chain] = The iron chain of this character is a difficult point to output. This can be used when the iron chain on the character is not obvious enough, but using this prompt word may make the iron chain too obvious. Please adjust the weight.