Basler Camera with Pylon SDK Errors #16
Comments
|
Hey, ok so camera id is probably a poor name. It's actually the USB id (nth device on bus). If you only have one usb device connected, src_fn should be 0. If you have more than one device, src_fn might be 1, 2, ... depending on device enumeration. If you've tried setting src_fn to 0, 1, 2.. and still not getting it to work, then this suggests something else is wrong. You should make sure the camera is properly configured and streaming before starting fictrac. You should probably be also close basler viewer app prior to starting fictrac. |
|
Thanks for getting back to me. We tried 0,1,2 etc but had no success. |
|
Hmm, ok. Basler cameras are a little poorly tested unfortunately, so there could be an API change in the newer versions that's causing an issue. Basler code was originally added June 2019 I think, so that's probably the last version that was tested properly. I will try and get hold of a basler camera to test with, but if there's API issues you should be seeing other error messages, at least in the console. In the meantime you could try posting in the forum (Reddit) to see if anyone has more recent experience with basler and fictrac. You could also try contacting directly those that have posted basler questions in the past. You should also try just using the basler camera without the SDK. Just build without those options. You probably won't be able to use advanced camera modes, so try setting to some default standard res/FPS/bitdepth first and see if it works! You'll still need to find which USB id your camera is using (should be zero if you have no other cameras). |
|
Okey so I did the following and it still doesn't work: I therefor put src_fn = 5 . Does this help you to debug? I am still lost to be honest. I will reach out on Reddit to the others and thanks for the help so far! |
|
Actually upon building with SDK 5 specified we get the following: So of course we closed the Pylon viewer to not have another client using the camera but this will result in the following error: |
|
Hmm, ok good debugging. Device 5 with SDK looks the way to go.
It also looks unfortunately like it's an SDK problem now. It's not
something I'm going to be able to fix myself in the next week or so as I
don't have access to a Basler camera right now.
I suggest you try installing pylon 5.2 - I think this was the active
version last time I tested. Let me know how you go and I can update the
docs if that works.
…On Thu, 22 Apr 2021, 17:19 lumyus, ***@***.***> wrote:
Actually upon building with SDK 5 specified we get the following:
Error! Unable to set process priority!
Looking for config file: config.txt ..
Config file parsed (21 key/value pairs).
Error opening capture device! Error was: Failed to open device '2676:ba02:14:4:5' for XML file download. Error: 'Device is exclusively opened by another client.'
Error! Could not open input frame source (5)!
So of course we closed the Pylon viewer to not have another client using
the camera but this will result in the following error:
Error! Unable to set process priority!
Looking for config file: config.txt ..
Config file parsed (21 key/value pairs).
Opening Basler camera device: �*�p�
Error opening capture device! Error was: NULL pointer dereferenced
Error! Could not open input frame source (5)!
Closing sphere tracker
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#16 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA5KTA46EICVRZNH2KG6SZTTKA46PANCNFSM43IALQRA>
.
|
|
So I verified that Pylon SDK 5.2 is not working either. The errors remain. Any other ideas on what might cause the issue? How exactly are we supposed to 'open' the camera stream? Do I have to edit anything at all in the Pylon Viewer? |
|
Actually while comparing with: https://github.com/basler/pylon-ros-camera/blob/ed094fad02eed38af830a052c7420befc6483ef3/pylon_camera/src/pylon_camera/write_device_user_id_to_camera.cpp I noticed that we actually never open the camera? |
|
Hmm sounds odd. The Basler code was contributed by another user but has
been working before. I will have to look into it. I don't have a Basler cam
right now, but if you can help test it will help!
…On Fri, 23 Apr 2021, 13:57 lumyus, ***@***.***> wrote:
Actually while comparing with:
https://github.com/basler/pylon-ros-camera/blob/ed094fad02eed38af830a052c7420befc6483ef3/pylon_camera/src/pylon_camera/write_device_user_id_to_camera.cpp
I noticed that we actually never open the camera?
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#16 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA5KTA5DCL2YIKEOIPQ3LQLTKFOCHANCNFSM43IALQRA>
.
|
|
Yes we can do that together. We have a bunch of Basler Ace at hand in our lab. How do you want to proceed? |
|
I did some more testing today and it seems that first of all: and I get the correct camera name as output (which is already a good sign). I also saw that you defined Let me know if this helps you. Thanks again for your awesome work! |
I got to make some progress. This is the line I changed to get to this point: |
|
Fictrac now works with the Ace cameras! The last issue was the following: I will make a proper pull request once I cleaned everything up. |
|
Now it's my turn to try to get fictrac up and running with an ace camera (on a Windows machine). I modified the cmake file to point to the most recent library (PylonBase_v6_3.lib) and everything compiled nicely. When I try to run the executable, however, I get the issues that @lumyus was having (a garbled camera name and |
|
As suggested above, I changed: |
Closes rjdmoore#16 using the suggestions there.
|
Had to make two additional changes:
|
|
Hello, @DanTurner-Evans @lumyus Below is what I received from the fictrac file. If we pay attention to the row of
In addition, I expected to see two types of timestamp saved in the .dat file. One at column 22 timestamp (since epoch) and the other one at column 25 alt. timestamp (since midnight) as described in the date_header.txt. However, in the .dat file, the first timestamp remained to be the first value of default timestamp(472843433.523000) and the second timestamp remained to be 0 throughtout the entire recording. |
|
Our .dat files are also missing the timestamp information. Luckily for our case, we send the data over the socket to a VR that runs in Unity and save the time information that way. |
Hey,
I built Fictrac with Cmake and the following command:
cmake -D CMAKE_TOOLCHAIN_FILE=/home/vcpkg/scripts/buildsystems/vcpkg.cmake -D BASLER_USB3=ON -D BASLER_DIR="\opt\pylon5" ..I am using PylonSDK 5 (I also tried PylonSDK 6 but the build would not work) and we are running everything on Ubuntu.
This is the error I get when running Fictrac:
I also tried to change the src_fn from 40022761 to 0 but the error stayed the same. (40022761 is the camera id in PylonViewerApp)
I also tried to build without the Pylon SDK (
cmake -D CMAKE_TOOLCHAIN_FILE=/home/vcpkg/scripts/buildsystems/vcpkg.cmake) but the error stayed the same.Somehow we are not getting any output with Fictrac, neither with the ConfigGui.
The config.txt used is the same as the one provided in the /sample folder. (Except src_fn has been changed)
Any idea on what is going wrong?
Thanks for your help!
The text was updated successfully, but these errors were encountered: