DAINAPP
patreon
Dain-App Alpha 0.1
🕑 Added 2020-01-15 11:12:57 +0000 UTCThis can be considered truly the first version of the app. The previous release i didn't even thought it would work in other people computers, imagine my surprise once it did!
So this version there is pretty much the basic stuff that should have the first release.
Change log:
- Fix to the path with white space
- Fix to the progress bar
- Watermark is optional
- New option to interpolate 2X/4X/8X
- Three new options to interpolate down to 60fps
- Most CUDA versions should be fixed, but i'm almost sure it won't work for now in CUDA 10. Will try to fix it soon.
Comments
GRisk
It should, yes
will it work with a super rtx 2060?
GRisk
Sadly no, maybe in the future, but it will take a long even that happen.
The program is very heavy on VRAM usage. So you have to tell DAIN to resize the video with the checkbox options. You'll have to keeo lowering it from the original resolution until it works. For my GPU with 8GB of VRAM, I have to keep the width below 1100 if it's a 16:9 ratio video. The program doesn't seem to use Unified Memory for CUDA to utilize your system RAM.
Aw this won't work without an NVidia GPU? :<
RuntimeError: CUDA out of memory. Tried to allocate 1.68GiB (GPU 0; 6.00GiB total capacity; 3.47GiB already allocated; 1.12GiB free;; 44.05MiB cached)
GRisk
I seen some people with this problem, still trying to figure out what is the problem, hope i can fix it soon.
I also get the python warning, but more serious I get a "no kernel image is available for execution on the device" error and it won't run. ;( This is on NVidia GTX 750 Ti with CUDA 9.0.
GRisk
This can be really slow if the resolution is big. I suggest trying in a small gif first. The link is only showing a blue background, weird site indeed.
Pastebin is having issues at this time so here's this site I've never heard of. https://hastebin.com/wihupotiti.sql Edit: Reading it now it actually looks like it's doing something? Is it this slow per frame or is that maybe because it keeps trying to do whatever that error is? I've only ever tried algorithm based interpolates.
GRisk
Hey there, this message is only a warning and should not impact the end result. Can you share a print of the console where the app hangs?
So Torch is spitting errors at me about autograd's python function being Legacy and no longer supported. Do you think it'll work for the AI to update that on your end or should I just install an older version of Python? I think I have 3.7 installed right now. Edit: I had 2.7.13 (64bit) and 3.7.3 (32bit) installed. I removed both and installed 3.6.8 as said from the Github with no success.
