Hi Clipka,

I had changed void operator()(DBL x, DBL y, DBL width, DBL height, Colour&
colour)
this function.

Yes, you are right! There was something wrong with the way I was displaying it.
Because the displaying algorithm had only limited grey value resolution, it
quantized everything and created bands of depth. How just such a simple thing
confounded me.

Thanks for your message,
Ankur.

clipka <ano### [at] anonymousorg> wrote:
> Am 21.09.2012 19:10, schrieb handos:
> > Hi everyone,
> >
> > I had written a povray patch for obtaining depth per pixel of an image some time
> > ago. I had used povray-3.7 and modified the tracetask.cpp in
> > (source/backend/render/tracetask.cpp) in the function TraceTask::Run() where I
> > was using all the time SimpleSamplingM0() and in this function I replaced the
> > line
> >   trace(x, y, GetViewData()->GetWidth(), GetViewData()->GetHeight(), col);
> > by
> >   trace(x, y, GetViewData()->GetWidth(), GetViewData()->GetHeight(), col,
> > depthVal); to obtain the depth.
>
> That change alone wouldn't do much good, because it wouldn't compile;
> you must have changed other stuff besides, at least the function
> operator in TracePixel:
>    void operator()(DBL x, DBL y, DBL width, DBL height, Colour& colour)
>
> > I have been using Jaime's office scene for
> > rendering and obtaining depth-map but I only observe that there are
> > discretization artifacts I observe in the depth. I have kept the image and the
> > depth file here at www.doc.ic.ac.uk/~ahanda/scene_and_depth.html . I am
> > wondering if this is due to very simple sampling method used for tracing? I am
> > not using any AntiAliasing options and jitter etc. to render that scene and
> > simply picking up the depth values returned by the trace function that's called
> > by SimpleSampleM0().
>
> This is interesting: It isn't just some color banding - each band is a
> gradient of its own. I'm pretty sure you're doing something wrong there.
>
> You're obviously translating the depth value into a color; are you sure
> that algorithm works properly? An error in some code intended to
> interpolate between colors from a table might fit the bill.

Post a reply to this message