Be sure to specify the medium you wish to be used, in your prompt, eg (a watercolor painting).
V2 - use DragonfruitGT SD model
V1 - use ORIGINAL SD 1.5 SD model
you can experiment with other models, but reesults WILL vary
This is a LoRA extracted from a current WIP checkpoint model named "MINT"
The goal of "MINT" is to increase accuracy when generating images that use art mediums such as watercolor, charcoal, pen, pencil, sketches, oil, paint etc.
I use this at a strength of 1, with low sampling steps