Update 6/19/2010

Further information on the driver and using the card can be found on Ben Collins blog (here).  Check back next week for an update on this site for sample code for saving MPEG-4 streams to the disk.

As previously mentioned we have been working hard over the past few months on the driver for version 2 of our Linux hardware compression card.  Since version 2 is a complete rewrite of our DVR software we naturally wanted to start with a clean driver, one that was written from scratch.  So, we leveraged the Linux kernel’s API for Video (Video4Linux) and Audio (ALSA).  This effectively puts Bluecherry as the first company to produce a multi-input MPEG-4 hardware compression GPL driver written around the Linux kernel’s API.

As mentioned in Ben Collin’s personal blog today we published version 2 of this driver on git.kernel.org.  While we have a few features we still plan to add to the driver along with the Video4Linux API, but it’s close enough that we feel comfortable releasing the driver to the community.  The driver was originally intended for version 2 of the Bluecherry surveillance application, however we felt that other open source applications could make use of the time and effort we put into the driver.

This driver will load up to 16 MPEG-4 and JPEG encoders, 16 audio (ALSA) devices, along with a uncompressed YUV display port.  This display port allows you to change the composite video output, for example you can have a quad view (or more) of attached cameras, also known as a “4-UP”.  This can actually save hundreds of dollars, since technically you would not need an external video quad or mulitplexer to combine separate cameras into one video output.

The driver is really meant for those with a bit of technical knowledge in using the Video4Linux API, since (currently) most Video4Linux expect to receive MPEG-2 / JPEG / YUV from the Video4Linux API.  This means that the majority of open source applications will only be able to access the JPEG feed off the card.   It should be noted that the FPS on each port in JPEG and MPEG-4 are dedicated.  This basically means that there’s no switching of inputs, so applications can expect to receive a full 7.5FPS @ 704×480 (Using 16 channels).  Making use of the MPEG-4 stream would require some modification in the Video4Linux support in the user-space application, or piping the MPEG-4 stream into an application that supports MPEG-4 streams.

Note:  This currently only covers our MPEG-4 cards.  We will be adding support for a seperate line of H.264 in the next month, so check back for an updated driver.

An example of the kernel log when the driver is loaded is below:

[   10.867392] solo6010 0000:03:01.0: PCI INT A -> GSI 22 (level, low) -> IRQ 22

[   10.868562] solo6010 0000:03:01.0: Enabled 2 i2c adapters

[   14.328870] solo6010 0000:03:01.0: Initialized 4 tw28xx chips: tw2864[4]

[   14.328943] solo6010 0000:03:01.0: Display as /dev/video0 with 16 inputs (5 extended)

[   14.382318] solo6010 0000:03:01.0: Encoders as /dev/video1-16

[   14.382666] solo6010 0000:03:01.0: Alsa sound card as Softlogic0

03:01.0 Multimedia video controller: Bluecherry BC-16480A MPEG4 16 port audio video encoder / decoder

Expected retail price of these cards, without our software, is $200 for the 4 port, $230 for the 8 port, and $375 for the 16 port card.  Quantity discounts will apply

Each card comes with a one year warranty, a thirty day no hassle return, and access / support of our GPL driver. Bluecherry is a registered licensee of the MPEG-4 Visual patent license (MPEGLA)

More information:

Bug tracker: http://improve.bluecherrydvr.com/projects/solo6010-v2

Bluecherry is proud to start gathering applications for beta testing of version 2 of it’s software. If you are interested in beta testing please complete the version 2 beta testing application

Note:  Bluecherry does not anticipate opening version 2 beta testing until the first part of September, 2010. If you are selected we will contact you and provide further more information.

There’s well over one hundred new features that is going into version 2, here are a few:

• Cross platform client (Windows, OSX, Linux)
• Network camera support (ACTI, Axis, Arecont Vision, Pixord, Panasonic, TRENDnet, Vivotek)
• Significantly increased performance
• Lower memory footprint
• Better storage handling routines
• 64 bit support
• Headless server option

For those of you who may not know version 2 is a complete rewrite from version 1. Version 2 is also focused on using, and enhancing existing Linux kernel APIs. Ben Collins has done an excellent job in writing a Video4Linux 2 compliant MPEG driver for handling the video encoding, along with what comes out of the display / spot out on the cards. By writing the driver we’ve been able to reduce the CPU usage during encoding to only around 10%, opposed to around 30% in version1 (for 16 inputs at 480FPS (at 320×240)).

Update: For kicks, we tested encoding 44 channels of audio and video @ 704×480 (4 channels at 30FPS, 8 channels at 15FPS, and 32 and 7.5fps) and the CPU usage was 10% with a load average of 1.5 on a dual core processor. This would factor down to about 20% CPU usage on a single core processor.

Ben has also done an excellent job in using the ALSA library to handle the audio inputs from the current MPEG-4 capture cards. He talks a bit about this in his blog posted here:

http://ben-collins.blogspot.com/2010/05/writing-alsa-driver-basics.html
http://ben-collins.blogspot.com/2010/05/writing-alsa-driver-setting-up-capture.html
http://ben-collins.blogspot.com/2010/05/writing-alsa-driver-pcm-hardware.html
http://ben-collins.blogspot.com/2010/05/writing-alsa-driver-pcm-handler.html

We are also changing container formats in version 2 to the Matroska (.mkv) format and encoding the audio and video into the same stream. This is a huge change from the way version 1 works, which stores the video in a (non-compliant) video container and the audio in a separate audio file.

Check back for more updates on version 2. We are estimating a release towards the middle to end of July, 2010.

We are excited to announce Brandon Holtsclaw has joined our development team. Brandon is a core developer for Ubuntu, focusing on general packaging and the Kubuntu project. Besides Brandon’s background as a Ubuntu developer he has freelanced in numerous positions creating and enhancing the user interface software for several fortune 500 businesses.

Brandon’s primary responsibility will be developing the new user interface in version 2, which includes the Windows, OSX and Linux remote clients. Brandon will also assist with the web interface and API connections.

For those who are following the Ubuntu release schedules you know that Lucid Lynx’s release date was today. While Ubuntu had a small offset in the morning of the release (An issue with dual booting) it appears to be resolved and the release is out.

With that in mind we are still keeping version 1 on Jaunty (9.04). Version 2 is in the works and progress is being made very rapidly which will have a base release of 10.04 and we will also support 9.04 (Jaunty), 9.10 (Karmic). Version 2 will be a much cleaner code base and once it’s released we will have more development time free to add support for different Ubuntu releases and different architectures (specifically a 64 bit release).

We had wanted a version 2 release in June, however this may be moved back to July. Stay tuned for news on beta testing!

Links:

https://lists.ubuntu.com/archives/ubuntu-announce/2010-April/000133.html