Closed breunigs closed 2 years ago
Thanks for reporting the issue and providing the test data!
Warning ! It's a bug from Gopro, and this "hack" could generate some other problems as the subsecond is wrong! I've reported this problem to Gopro here: https://community.gopro.com/t5/Cameras/subsecond-timestamp-bug/m-p/1057505
Thanks @Stefal for the information. What problem do you think it will cause? I will reopen it and investigate it later.
If the real subsec value is an equivalent of 5ms, it will be stored in the Gopro image as . 50
and converted to .50
. It is a 495ms offset. I see two solutions:
None of these solution is good.
I have my own script to fix this, it checks the length of the subsectimeoriginal value and insert some 0
if needed.
So you can insert some code in the mapillary-tools to check this:
But that's a lot of work for only one cam, with a bug which should be fix soon... I hope...
btw: @breunigs, feel free to add some pressure on Gopro, and explain that this bug is a problem for you too.
Basic information
0.6.0
Steps to reproduce behavior
I uploaded an attached image that fails to extract time taken because of the space in the subseconds field. I'm using
Expected behavior
It parses the subseconds just fine.
Actual behavior
It fails with
Warning, required geotag_process failed for image bla.JPG
, skipping the image. More generic error messages are repeated later.Additional information
I hacked in a
sub_sec.strip()
https://github.com/mapillary/mapillary_tools/blob/d0544d0475fbe9fb44edbadf7479debd0a409ac3/mapillary_tools/exif_read.py#L202 and this seems to fix the issue… well, I only take pictures once every second, so even if this is not a proper fix, I might not see it.I can provide more images if you want to test more. Images around the same area+time as the provided example one are being uploaded right now, if that's more convenient for you.