- On the mailing list, Roberto reported on the mailing list there was no way to switch the input used for capture during capture, other than stopping the current frame grabber and re-creating another one. I have implemented a fix and it is now possible to change the current input while capturing, however, the time required to complete the switch is driver dependent, and, as Roberto reported, it can take a couple of frames for the switch to happen, and the switch might not happen between frames.
- Leonardo reported a bug here due to v4l4j not handling properly string, interger64 and bitmask controls. This is now fixed.
- An error reported by Jay here was occurring when capturing with PAC_3711 devices, which is now fixed
- Many users have reported an issue where v4l4j fails to run because of undefined symbols. I tracked down the problem to some versions of GCC not parsing the command line in the same way as others, and ignoring the list of linked libraries.
- Jeff reported a NullPointerException when stopping and starting the capture. This has been fixed.
- v4l-utils has been updated to v0.8.6 .
- v4l4j uses PixFC v0.3 to speed up the following conversions in v4l-utills: YUV422 <-> RGB24, YUV422P <-> RGB24, YUV420P <-> RGB24
- JPEG conversions have also been optimised. v4l4j now only supplies downsampled YUV frames to libjpeg, and relies on PixFC for YUV downsampling and RGB to YUV conversions.
- I have updated the Ubuntu packages to not depend specifically on OpenJDK, as request on the mailing list.
- Last, I started testing v4l4j in qemu emulating an ARMv7l architecture. While v4l4j builds fine, I couldn't actually test interactions with a video device. So ARM support is only experimental as this stage, but I am working on it actively.
That's it. v4l4j v0.9.0 can be downloaded from here. Ubuntu packages will be available shortly.
No comments:
Post a Comment