stable-diffusion
stable-diffusion copied to clipboard
Question regarding random seeds
I've been writing an interactive version of txt2img.py that minics the behaviour of the discord bot. What I can't figure out is how to retrieve the individual random seeds for multiple sampled images? Does anyone know how this is done?
I notice that there are two ways of generating multiple images in the original txt2img.py, either using --n_iter or --n_samples. The latter is faster, but with the first I can set a new random seed between each image, which I think allows me to reproduce the behavior of the bot. Can someone advise on what the difference between these options is?