V1.0 may produce a messier scene with slightly softer lighting, but it can sometimes generate unclear, blurry, and indistinct objects. V1.1 addresses these issues with optimized scene clarity，primarily achieved through improved dataset quality, though it introduces slightly more saturated lighting.

V1.0尽管看起来更杂乱，灯光也更自然，但是有伪影问题（人像之外物体的模糊、畸形）

V1.1通过一定程度提升数据集图像清晰度来弱化伪影问题，但是灯光也更饱和了

The dataset is provided, so if you're interested and good at Lora training, you can use it to train, and then you can let me know your tips and results!

This Lora can depict scenes of China Red Light District Inside

V1.1

31imgs,18epochs,5repeats

girls in neon ghetto room,girls in  empty cheap neon room,enclosed room,neon_ghetto_room,sitting on a cheap sofa,  girls,  dim neon lit,dim lighting,dim lit,

V1.0

This is my 4th trained Lora, the faces were masked in the dataset using black squares, be careful that the Lora strength can not be too high or you will get black blocks in generated imgs.

Common Prompts

????????????, a group of young women sitting in neon lit room, whore, brothel, 3girls, multiple girls, thighhighs, couch, sitting, cleavage, various hair, realistic, various dress, indoors, looking at viewer,  bottle, ring,

((the room is in  messy,dirty and dilapidated urban village, ???????????? in urban village alley's messy room,The room is simple and dilapidated.)),

dim neon light,