Same condition (seg), same sample results

[LoveU3tHousand2]

As the title , my ldm sample always generates almost same sample if I feed same seg-condition. Is that normal? What is the reason of it ? training time not enough? training data not enough ?

[jenuk]

What exactly do you mean by "seg-condition"? How did you generate these images? Is it the txt2img script? If you have run it multiple times to generate one image each, you need to change the --seed <some number> each time, to get different output, since the output is deterministic given a seed and conditioning.