Attached is a patch that seems to work for desired_encoding = 'passthrough'.

Also, from the opencv python cookbok there should be a cv.fromarray to go
from cvMat to numpy array. I don't see this in the vision_opencv-1.0.3.
What version of the vision_opencv stack does the online documentation
correspond to?


On Thu, Jun 10, 2010 at 9:00 PM, Advait Jain <advait@cc.gatech.edu> wrote:
> In the python implementation of cv_bridge the imgmsg_to_cv function
> returns a CvMat object instead of an IplImage. This is different from the
> documentation and the C++ implementation. Is this intentional or a bug?
>
> Advait
>