prompt: ((((best quality, masterpiece)))), (detailed, 1girl), a girl eating a hamburger
negative_prompt: young, child, loli, ((((low quality)))), (((normal quality))), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glans,extra fingers,fewer fingers,strange fingers,bad hand, bad leg, smile, bad hands, big eyes, big eyes, makeup, bad eyes, extra leg, extra hand , extra foot, bad toes, extra toe, nsfw
seed: 9129090
width: 640
height: 896
num_outputs: 1
num_inference_steps: 40
guidance_scale: 4.0
sampler_name: dpmpp_2m
use_stable_diffusion_model: lyriel_v13.safetensors
use_vae_model:blessed2
I wrote it twice mistakenly. I compose my prompts by copy pasting from other prompts I use or steal. I don’t know how many I made but over 100 that session. I have several dozen gigabytes of SD images saved up already. My computer can make a 30 steps 512x512 image in maybe around 15-22 seconds. It takes maybe a minute to make a 768-1024 45 step image. I’m not sure of the numbers. I don’t pay too much attention to that. I usually leave them generating and go do something else. I also test so many different configurations that the numbers mix up in my memory. RTX 2060 with 6gb VRAM.
Nice hobby, nice tools, much more productive (~1 000 000 x ?) than oil painting !
… so can you ask the computer to make a 768x1024 as a better version of a 512x512 (same image yet better quality) ?
yes. It doesn’t really “make a better version of X”, but “make a version based on X”, which depending on your preferences might come out as better or completely different. But yes, in summary, if you’re not picky about the meaning of words, yes, you can do that.