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

Re-seed image if needed #97

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Re-seed image if needed #97

wants to merge 1 commit into from

Conversation

3DTOPO
Copy link

@3DTOPO 3DTOPO commented Dec 15, 2015

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.

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.
@spencermize
Copy link

I can confirm this fixes my issues with "function value changing less than tolX".

@3DTOPO
Copy link
Author

3DTOPO commented Dec 17, 2015

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.

@spencermize
Copy link

After a few runs, I'm starting to run into some coloring issues after receiving the re-seeding message. See the attached - small image (no re-seed) looks "correct" whereas the big (re-seeded) image looks off. This has happened on every re-seeded photo, I think.
re-seed
small
riverstyle
river

Thoughts?

@3DTOPO
Copy link
Author

3DTOPO commented Dec 22, 2015

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?

@spencermize
Copy link

I've only managed to get it to re-seed on large resolution outputs. I've
come close (1750px runs don't seem to error) and the scale looks okay. The
one I attached isn't massively larger (2250px) but the colors and stylings
are very far off.
On Dec 21, 2015 9:13 PM, "Jeshua Lacock" [email protected] wrote:

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?


Reply to this email directly or view it on GitHub
#97 (comment)
.

@3DTOPO
Copy link
Author

3DTOPO commented Dec 22, 2015

I guess I am not sure what I am looking at. You attached a really large render and a quite small one.

@spencermize
Copy link

Sorry - I'm struggling to find my 1750px render. I'll try to re-run and
repost.
On Dec 21, 2015 9:32 PM, "Jeshua Lacock" [email protected] wrote:

I guess I am not sure what I am looking at. You attached a really large
render and a quite small one.


Reply to this email directly or view it on GitHub
#97 (comment)
.

@spencermize
Copy link

I did a few more runs while adjusting the scale - definitely better. Thanks
for the tip!
On Dec 21, 2015 9:32 PM, "Jeshua Lacock" [email protected] wrote:

I guess I am not sure what I am looking at. You attached a really large
render and a quite small one.


Reply to this email directly or view it on GitHub
#97 (comment)
.

@dionysio
Copy link

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.

@0wnrepo
Copy link

0wnrepo commented Mar 23, 2016

@spencermize you might also want to look at this : #171

@tenorslowworm
Copy link

Thanks, @3DTOPO, your fix is working great for me!

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

Successfully merging this pull request may close these issues.

5 participants