Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Out of memory Issues #36

Closed
Ristellise opened this issue Dec 6, 2019 · 4 comments
Closed

Out of memory Issues #36

Ristellise opened this issue Dec 6, 2019 · 4 comments

Comments

@Ristellise
Copy link

Ristellise commented Dec 6, 2019

I noticed that DAIN doesn't handle HD video sizes [1280x720] with a 8gb of GPU memory. Running on a GTX 1070. Is there anyway for me to reduce the memory usage on the GPU side?

@laomao0
Copy link

laomao0 commented Dec 8, 2019

There is a method which costs more time.
You can first split the image into multiple patches, then run the interpolation process for each patch.
Finally, you need to combine those patches to the whole image.

@Ristellise
Copy link
Author

There is a method which costs more time.
You can first split the image into multiple patches, then run the interpolation process for each patch.
Finally, you need to combine those patches to the whole image.

Thanks for replying, so Is it a simple case of splitting the image into a smaller resolution and then rendering each individually with the next frame?

@laomao0
Copy link

laomao0 commented Dec 10, 2019

https://github.com/laomao0/EDVR_4K_HDR/blob/master/test_4k_hdr_with_GT_stitching.py
EDVR: https://github.com/xinntao/EDVR
You can refer my case.
I use the EDVR to upconvert the image resolution from 540p to 4k.
The model costs too much gpu memory.
So I split the 540p into 9 sub-images,and runs for each sub-image.
At last, I combine them to form a whole pic.

@Ristellise
Copy link
Author

Thanks :)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants