KnoBuddy
[BUG]
Describe the bug When I try to set ETA : 0.8 I have an error
To Reproduce
Traceback (most recent call last): File "prd.py", line 2125, in <module> do_run(batch_image) File "prd.py", line 1127, in do_run samples = do_sample_fn(init, args.steps - cur_t - 1) File "prd.py", line 1100, in do_sample_fn eta=eta,
Desktop (please complete the following information):
- OS: WINDOWS 11
- GPU and VRAM: NVIDIA RTX3060Ti 8GB
- CUDA Version 11.7
Additional context Thank you
I believe this is the same issue as you asked about with clamp max. Both are schedules. I haven't used PRD in over a month, unless something's changed there isn't many bugs I was aware of and the ones there are shouldn't occur within normal operation of PRD. And since you are getting a PRD error it could be Visual Diffusion giving PRD bad settings, or it could be that it just passed on your bad settings and PRD is erroring out.
Also are you 100* certain that's the whole error?
Traceback (most recent call last): File "prd.py", line 2125, in <module> do_run(batch_image) File "prd.py", line 1127, in do_run samples = do_sample_fn(init, args.steps - cur_t - 1) File "prd.py", line 1100, in do_sample_fn eta=eta, NameError: name 'eta' is not defined
I have no more errors
I have eta is not defined error too if I change the eta and clamp max.
I have eta is not defined error too if I change the eta and clamp max.
Are you using PRS, or PRD? I will assume PRD in which case, if you want to change clamp_max and eta from auto, you must format the value as a schedule as seen in the fields to the right in this picture.
The total must add up to 1000, if you would like a single value to use the whole way through you can do something like [XX]*1000, where XX is some number you want to use. Otherwise if you do [XX]*400+[YY]*600 which would use XX value for the first 40% of the render, and YY for the second 60% of the render.
I use PRD: Hey thanks for the quick reply, so I copied the format of cut ic pow over to clamp max and I got an error before saying “could not convert string to float integer”. Can you tell me what format works for you in there and for eta? And when I enter in the prompt replacing the default prompt, it’s completely different than it should actually be. (I know this from disco diffusion on colab)