The Lora is trained on MidJourney Images that use my "Personalised Style" + Stained Glass.Works without the trigger words ArsMJStyle, Stained GlassHave a good effect in the 0.4 - 0.8 rangeStarts to melt a little bit at 1 in some checkpoints. 0.8 is a good starting point.SD1.5, SDXL, and PonyV2 are trained on a 100% larger dataset. Flux update: Same dataset as PonyV2 , but with different capture. Check the version description for recommended settings.

Stained Glass

SD 1.5 Version of the Lora. Trained as Semi Realistic with Clip Skip 1Works without the trigger words ArsMJStyle, Stained GlassThe Lora Colors are intense, so it can produce bright images with lower CFG scales. I use 4-7.Have a good effect in the 0.4 - 1 rangeStarts to melt a little bit at 1.2 in some checkpoints. 0.8 is a good starting point.

Files with SHA256: 139dbd91855247e2637a4d171ec0c785fd980f712d873f631577b543720e9fbd