I am trying to get the "*_mono16" video modes working with camera1394.
If I just copy the two-byte pixels, image_view of camera_raw fails
with an opencv exception. I can't tell if this is a bug in the camera
driver or in image_view.

I've looked at docs and code, but I'm still not clear how 16-bit
monochrome image data should be encoded. The two-byte per pixel copy
is modeled on cameradc1394. Is this snippet of code wrong?

    case DC1394_VIDEO_MODE_640x480_MONO16:
    case DC1394_VIDEO_MODE_800x600_MONO16:
    case DC1394_VIDEO_MODE_1024x768_MONO16:
    case DC1394_VIDEO_MODE_1280x960_MONO16:
      if (!DoBayerConversion_)
        {
          image.step=image.width*2;
          image_size = image.height*image.step;
          image.encoding = enc::MONO16;
          image.data.resize(image_size);
          memcpy(&image.data[0], capture_buffer, image_size);
        }

Could this be a little-endian problem?
-- 
 joq