This is a v1.5 model based training of the Xiaomi su7 lora

Recommended parameters:

Model: misu7_lora_sd.safetensors

Steps: 25-35,

Sampler: DPM++ 2M Karras

Weight: 0.6-0.8

Hd Fix:

Number of steps: 12, amplitude: 0.6

Trigger word: misu7

Positive:

misu7,a car,protruding front,from_side,tall building,masterpiece,best quality,official art,extremely detailed CG unity 8k wallpaper,highly detailed,<lora:misu7_lora_sd-000003:0.6>,

Negative:

(worst quality, low quality:1.4), lowres, text, (nsfw:1.3), (normal quality:2), watermark, cropped,