On Tue, Dec 6, 2011 at 7:38 PM, Kevin Walchko <kevin.walchko@gmail.com>wrote:

>
> On Dec 6, 2011, at 10:16 PM, Patrick Mihelich wrote:
>
> The proposed depth image representation is *exactly* the one currently
> used by the ROS OpenNI driver. The intent of this REP is to codify existing
> practice. What code would need to be rewritten?
>
>
> ok, wouldn't you have to re-write anything you want to start using
> DepthImage with? Currently any packages that accept sensor_msgs/Image can
> read the Kinect images … that is cool. However, now everything would have
> to be re-written if you wanted it to view the kinect depth image. Maybe I
> am wrong, but wouldn't that mean rviz, image_view, other user software, etc
> would have to be modified to accept this new message? To me this sounds
> like a big change for little gain.
>

There is no DepthImage message. The REP describes how to use
sensor_msgs/Image to represent depths.

I'll update the abstract to make the purpose and recommendations more
explicit from the beginning.

Cheers,
Patrick