Software Development Kit

Why MxPEG?

MxPEG seems to be "just another video codec".

In fact, MxPEG is a video codec. And MxPEG is the name of a simple audio/video container format.

MxPEG was invented by MOBOTIX in 2000. One practical design goal was to have an encoder that works on a standard Intel StrongArm CPU running at 206 MHz, and without any dedicated video compressing hardware. Other design goals were low latency and high accuracy. For these reasons, the MPEG family of standards wasn't considered for implementation into the first and second MOBOTIX camera generations, M1 and M10/D10.

MxPEG was designed to be closely related to JPEG. Basically, it should add interframe compression where Motion JPEG only makes use of JPEG's intraframe compression. This typically yields a bandwidth reduced by two thirds, compared to Motion JPEG.

The advantages of the MxPEG codec are:

uses less computing power than MPEG
this is particularly true for decoding, too (try to decode and display 25 MPEG4 streams simultaneously on your computer...)
latency < 150 ms
good still image quality (no motion estimation -- image quality only dependent on selected JPEG quality)
the codec implementation in the camera is able to deliver multiple streams with different frame rates simultaneously, e.g., for live viewing and for storing. No MPEG encoder can do that.
the MxPEG implementation in the camera is able to deliver lip-synchronous audio interleaved with the video frames

Of course there are application areas where MxPEG might not be the codec/streaming format of choice, for one of the following reasons:

not (yet) a standard format
currently no UDP/multicast support (TCP/HTTP only)
no efficient compression if large parts of the image are changing, e.g., when the camera is mounted on a constantly moving PTZ head.
sceneries that change very slowly might make the MxPEG encoder think they're static and keep the encoder from sending updated tiles

// next: How does it work?