miracle-ViceCity Nostalgic Film Simulation LORA Model SD1.5 (Portrait Version)

File Name: miracle-ViceCity.safetensors or miracle-ViceCity_v20231209.safetensors

Trigger Word: sincity (optional)

Recommended LORA Weight: 0.5~0.8, the weight varies in different film styles and models,

0.6 is the median value, you can adjust the LORA weight according to your needs. (Excess weight in some models may affect facial structure)

Recommended Sampling Steps: 25~30     CFG Scale: 7

Recommended Sampling Method: For realistic models, use DPM++ 2M Karras, for artwork models, use Euler a

Feel free to try other sampling methods, LCM and other solutions have not been tested.

Recommended Resolution: from 384x768 to 576x1152    Only for vertical screen resolution, suitable for mobile wallpapers

You can use 384x768 resolution, enable 1.5x Latent HD restoration, direct output for the best effect, redraw range 0.3-0.5

Recommended Image Enhancement Settings: 8x_NMKD-Superscale_150000_G with built-in film grain

                               4x-UltraSharp or R-ESRGAN 4x+ Anime6B also work

Recommended for Use with Large Models: The Useless Master-Dream Illusion, AWPortrait, majicmixHorror

Feel free to try other large models, please test them, feedback is welcome.