Closed lizhiqi49 closed 11 months ago
And the sampler I used is 50-steps DDIM, with the prompt as "an astronaut riding a horse, 3d asset", and unconditional prompt as null ("").
which script are you using and did you change anything?
Actually its on my custom diffusers code since I'm more familiar with diffusers, I converted your ldm checkpoint to diffusers version... I think it unlikely generates such reasonable results if there is something wrong in my code.
But if the issue I met is unmatched with your practice, I will go to try your official code and check mine.
I've tried your official code and everything goes well, so there must be something wrong in my code, I'll check it. hhh
Hi I have a question about the guidance scale.
I have tested the 2.1 model in my machine, and I find that the generation quality is very bad when guidance scale is relatively small (<20), and the quality was getting better and better as I added it along to 60. Below shows some of my testing results.
guidance scale = 10:
guidance scale = 15:
guidance scale = 20:
guidance scale = 40:
guidance scale = 60:
Since I notice that in your T2I script you set the guidance scale defaultly as 10, I'm wondering if I made anything wrong.
And the backgrounds in the images above are all same brown color, is it correct?