The Dirac algorithm and specification is designed to support arbitrary bit depths. We had in mind the need for professional broadcasters to use 10 bit video (routinely) and higher bit depths for some applications. Furthermore we also wanted to accomodate significantly greater bit depths for the film and post production community. So the algorithm is unconstrained. But the implementation is designed to run efficiently for the most common application which is 8 bit video. So internally it uses 16 bit shorts rather than 32 bit integers in places. This means that it is currently restricted to 8 bits (I think). The restriction is principally in the motion estimation/compensation I think. It should be possible to modify the code by changing some types to accomodate greater bit width.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
From Dirac README I see the example coder supports "raw 8-bit planar YUV". Is there any way to encode bit depths greater than 8?
The Dirac algorithm and specification is designed to support arbitrary bit depths. We had in mind the need for professional broadcasters to use 10 bit video (routinely) and higher bit depths for some applications. Furthermore we also wanted to accomodate significantly greater bit depths for the film and post production community. So the algorithm is unconstrained. But the implementation is designed to run efficiently for the most common application which is 8 bit video. So internally it uses 16 bit shorts rather than 32 bit integers in places. This means that it is currently restricted to 8 bits (I think). The restriction is principally in the motion estimation/compensation I think. It should be possible to modify the code by changing some types to accomodate greater bit width.