-
Notifications
You must be signed in to change notification settings - Fork 2.7k
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
Re-seed image if needed #97
base: master
Are you sure you want to change the base?
Conversation
If the total loss from previous iteration is 0, the image is re-seeded, otherwise a "function value changing less than tolX" message will appear and fail. Tested with many images and has worked every time. Note that you are more likely to reproduce the issue the larger the image output size is. Try 1024-1568.
I can confirm this fixes my issues with "function value changing less than tolX". |
Nice, glad to hear it. I can report that I have over 1,000 high res renders done with it now (non-stop). No added memory leaks as far as I can tell. |
I could be wrong but I think its because you are rendering it at a different scale. What do they look like the same scale and if re-seeded? |
I've only managed to get it to re-seed on large resolution outputs. I've
|
I guess I am not sure what I am looking at. You attached a really large render and a quite small one. |
Sorry - I'm struggling to find my 1750px render. I'll try to re-run and
|
I did a few more runs while adjusting the scale - definitely better. Thanks
|
Thanks for this PR! I could not for the life of me figure out why it was crashing randomly for some images and this fixed it. |
@spencermize you might also want to look at this : #171 |
Thanks, @3DTOPO, your fix is working great for me! |
If the total loss from previous iteration is 0, the image is re-seeded, otherwise a "function value changing less than tolX" message will appear and fail.
Moved the image initialization code into a reusable function and called if the total loss from previous iteration is 0 at proposed line 301.
Tested with many images and has worked every time. Note that you are more likely to reproduce the issue the larger the image output size is. Try 1024-1568.