Home Artists Posts Import Register

Content

  • You can now create a output with a manually set FPS, this is so you can have an output of 60fps if the final interpolations is a value bigger that that.
  • You can create a slow motion output, just set the speed you want it.
  • [BETA] You can interpolate the input in 1X, this will only create interpolations if the frame is duplicated.
  • Improved "Smooth Interpolation" code a bit.

Files

RIFE-App 2.01.7z

Comments

Snake Plissken

Oo I like that beta feature, good for "fixing" false 60fps videos.

DAINAPP

Yeah, the code still need some tweaking, but it should more or less be working =)

Snake Plissken

Main issue I have is the algorithm trained is pretty good at animations but for real videos, if its a 30fps video, a high motion scene will have a lot of blur frames that I would hope get “deblurred” but usually just get interpolated so the blur remains.

Snake Plissken

I have not had time but considering training the algorithm more using real 60fps videos that I downgrade to 30fps somehow but even that presents an issue since a 30fps video that came from a 60fps video will still have “less blurry” frames I think, it will just be less smooth.

Sergey Tokarev

Wanted to test CPU mode, but keep getting this error. Traceback (most recent call last): File "my_design.py", line 74, in run File "my_DAIN_class.py", line 1256, in RenderVideo File "torch\cuda\__init__.py", line 259, in set_device device = _get_device_index(device) File "torch\cuda\_utils.py", line 31, in _get_device_index raise ValueError('Expected a cuda device, but got: {}'.format(device)) ValueError: Expected a cuda device, but got: cpu

DAINAPP

Must be something I had to change for the model 3.0, will take a look, thanks

Sergey Tokarev

I also checked version 1.7.4, since it's the oldest one I have installed on my PC. Pretty much the same error, except 'my_DAIN_class.py' line is different: "File "my_DAIN_class.py", line 1121, in RenderVideo" So yeah, I don't think it's actually working on my CPU. I'm using Ryzen 9 3900X, so maybe it's somehow unsupported? Also, should I check my CUDA version, just in case, and if that's causing issues which one I should install?