-
Notifications
You must be signed in to change notification settings - Fork 585
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
Improved ordering of channels for EXR files. #904
Conversation
Now we very robustly cluster "layers" together (with the unnamed layer first, if it exists), and within each layer have a canonical RGBAZ order.
Any feedback on this, @sebastianelsner ? |
Both images submitted on the mailing list still make iv crash on windows. |
These changes just address the "channels in the wrong order" problem. I don't yet have any theory about why those files would crash. If you do a 'oiiotool broken.exr -o out.exr', does that work? That is, can you coax OIIO core into reading and writing the file if iv (OpenGL, Qt, etc.) is not in the mix? |
using oiiotool the way you suggested works without crashing on win + linux. Also with your changes applied iv displays without crashing on CentOS, while it still does on Windows. I will try to investigate on Windows a bit more ans open a new ticket if necessary. |
After a bit more investigating I found that the crashes are still happening on both platforms under certain circumstances: It seems if I load the small file first, the large file also loads. After that I can do the usual stuff like toggling images etc. |
Can anyone reproduce this with the steps I gave? The images are: https://dl.dropboxusercontent.com/u/82588645/oiio_exr/test.exr and https://dl.dropboxusercontent.com/u/82588645/oiio_exr/test_small.exr |
I still have been unable to reproduce on any platform. Can you compile a debug version, run iv with -F (keeps in the foreground), run in the debugger such as gdb, and get a stack trace showing exactly where it's crashing? |
Oh, and if it's still a reproducible problem, let's open a new ticket on it. |
I will test and report in a different ticket. |
Now we very robustly cluster "layers" together (with the unnamed layer
first, if it exists), and within each layer have a canonical RGBAZ
order.