carla
carla copied to clipboard
Camera output distorted
CARLA version: 0.9.14
Platform/OS: Windows 11 using a GTX 1080ti
Problem you have experienced: Camera (tried depth and rgb) output distorted as shown below
What you expected to happen: Image showing the world without distortion
Steps to reproduce: Follow the quick start guide until camera.listen(lambda image: image.save_to_disk('out/%06d.png' % image.frame)). Output images come with distortion.
Other information (documentation you consulted, workarounds you tried): Tried playing around with camera settings or other maps to no avail.
Hi, I discovered a similar problem, which might be also your problem. When I changed my camera resolution to a aligned value (I used multiple of 64 pixels) I could solve it. It was needed to align for both, width and height:
I used width = 600 ==> stride problem Align width to 640 ==> OK
Hope this helps you also.
I have the same camera distortion problem after going to v 0.9.14. Then I went back to v0.9.13, everything works again. To test quickly you can run visualise_multiple_sensors.py in the examples folder in v13 vs.14. Does this mean it is v14 issue?
I have the same camera distortion problem after going to v 0.9.14. Then I went back to v0.9.13, everything works again. To test quickly you can run visualise_multiple_sensors.py in the examples folder in v13 vs.14. Does this mean it is v14 issue?
And is your resolution aligned?
The problem happens only in Windows (DirectX) for resolutions not aligned to 256. DirectX add some extra bytes to each row. We have made a PR to fix that.
Will you be releasing a version of this bug fix in the near future, as it is indeed a serious bug?
Yes, probably we will release soon a hot fix like 0.9.14.1. The bug is fixed in the dev branch of Carla currently. We are going to make a new package with those changes, and as usual it will be the Carla_Latest package that you can download also.
Thank you, then I’ll wait for your new build instead of building my own version or reverting to the old one:)
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
I just wrote a short blog post about this issue on the Carla simulator research blog - https://carlasimblog.wordpress.com/2023/09/16/the-64-division-curse/. And added a short code snippet to fix this.
Hi,
CARLA version: 0.9.14 Platform/OS: Windows 10 using NVIDIA Geforce RTX 3070 Problem you have experienced: RGB Camera and Semantic Segmentation merged output is distorted
What you expected to happen: RGB Camera output merged with Semantic segmentation Steps to reproduce: Follow the quick start guide and run synchronous_mode.py under PythonAPI/examples Other information (documentation you consulted, workarounds you tried): I have followed the above steps and changed the resolutions to 640x480 and 1920x1080
Solution for this given here: https://github.com/TzabarDolev/carlablog/blob/main/fix_image_distortion.py
Is this problem fixed now?