Workflows (with prompts) for creating high quality photorealism (using AI)
Model named Analog Madness v3 and based on SD 1.5
Use the Dreambooth library.
Use DPM Adaptative with a very low CFG (2 or 3), at 40 steps.
The prompt is very simple, just add terms like "good lighting", "dimly lit", "camera flash", "depth of field", and always use the LORA epiNoiseoffset_v2 to have a better contrast in the lighting.
The prompt is: "your subject" (selfie), 4k, sharp focus, living room, dimly lit, (sunrays), black shirt, camera flash, depth of field, lora:epiNoiseoffset_v2:1
Negative prompt: blur, blurry, low quality, drawing, painting, photoshop, hdr, old, b&w
Steps: 38, Sampler: DPM adaptive, CFG scale: 2, Seed: 3535069355, Size: 640x832
https://www.reddit.com/r/StableDiffusion/comments/12sbyc5/true_to_life_photorealism_v2/