LoRa trained for mimicking the artstyle of the artist Tansau. Work well for creating cartoon shortstacks and furry characters.

FOR V3:

best prompt: 1girl, chibi

V3 is more adaptative and flexible than V1. So a weight of 1 should be good most of the time. If you get some strange result due to the Lora, you can lower it until 0.7-0.6 (with some loss of the art style of course)

FOR V1:

best prompt: 1girl, ass, breasts, thights, full body

To get a shortstack character, add short, <size> breasts and <size> ass to your prompt. <size> waists can also help.

To get a furry, add furry or furry male/female, colored skin, <animal> girl/boy, body fur, <animal> tail/nose/ears.

If you want the best results with this model (or even any LoRa or TI), I suggest you to generate the base image then using img2img with a larger resolution, (and with 4-6 batch size optionaly) and finishing by inpainting the face. That's the best and simplest workflow I use to generate those presentation images.

Can also work with semi realistic models like RevAnimated. You just need to increase the weight of the model to 1.5 or a little above.