Tal wrote:
> Hello All,
> 
> For research I'm doing in computer vision, I'm interested in modifying
> POVray so that it can output an image that is a depth (i.e.  range)
> map of the scene.  In other words what would a laser range-finder
> report back when scanning the scene.  I have several questions about
> this task.
> 
> 1. Has this been done?  I'd hate to be replicating someone else's
>    work.  I've looked in the obvious places and haven't seen that
>    anyone has done this.
> 
> 2. Is the only way to do this through modifying the source?  I thought
>    that there might be some crazy way of implementing this using the
>    programming capacities of the scene language, but I can't quite see
>    how.
> 
> 3. Can someone give me a quick roadmap and time estimate for the
>    project?  More specifically how long do you think it would take?
>    What's the basic approach that you would take?  Do you see any
>    obviously difficult stumbling blocks?
> 
>    I've looked over the code in render.cpp and it seems like I would
>    just need to make an alternate rendering routine that returns
>    ranges instead of RGB values.  In other words it doesn't seem like
>    it would be terribly difficult because I'd just be combining
>    together pieces of code that are already written in a new way.  Am
>    I totally off base?
> 
> Thanks,
> Tal

MegaPOV has this feature, and it can also be done with the gradient 
pattern applied to all objects.

i.e. pigment { gradient z scale <1,1,maxrange-minrange> translate 
z*minrange }

-- 
~Mike

Things! Billions of them!

Post a reply to this message