I just want to share the process of ai rendering, hope you guys would like it.
recommend models: https://civitai.com/user/aji1/models , https://civitai.com/user/EDG/models , https://civitai.com/user/Keith/models
stage1:
step1:
#write the prompts. usually in the first line, I will write the prompts of picture quality and lens control. in the second line, write the prompts for character characteristics and scene description. the last line is used to put the lora model to be used and the trigger words of the model.
#sampling method i will used the "DPM++ 2M KARRAS Test" this one in here.
https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/8457
#the first draft will be rendered after setting the size of the picture and the number of sampling steps, when i get a satisfactory draft, i proceed to the next step.
step2:
#click the Hires. fix, select the upscaler that i will used the 4x-UniScaleV2_Sharp, Hires steps is 15~20, Denoising strength will not over 0.5, reuse seed from last generation, check out the Adetailer and start Generate. send to img2img when it finished.
stage2:
#select the same Sampling method, decrease the sampling steps, check out the script of Ultimate SD upscale, start generate. send to extras when it finished.
stage3:
#set the resize, select the upscaler 1&2, start generate. Finish~