removing useless information in eps/ps files

Hi,

I think I already ask this question in the past but I cannot convince
myself that it remains without answer.
I create eps images with Matlab (let's say bar3 diagrams for instance)
for which most of the bars
are behind other bars. For some reason, the eps format keeps
everything as does the pdf format in case
of conversion of such diagrams. Is there a way to remove all the
information in the original eps file that is not necessary for
visualization.

Thank you,

M.

In article
<a51778e6-4fe3-4375-aef7-b1806c36368d@a22g2000hsc.googlegroups.com>,
pluton <plutonesque@gmail.com> wrote:
...
> Is there a way to remove all the information in the original
> eps file that is not necessary for visualization.
No, not easily and not using solely Postscript.

I am not a Matlab user, but what you describe reminds me of 3D rendering
using the "painter algorithm" (that is, the scene is broken down into
surface elements and those are ordered from furthest to closest to the
viewer, projected on a plane and then their 2D projection filled). If
the rendering primitives already provide polygon-filling (like
Postscript or PDF), this is easy to implement, works reasonably well for
non-intersecting convex polyhedras and so is probably good enough for
rendering 3D bar charts.
Of course, as you noted, this is wasteful, since fully hidden surface
will be projected and painted, only to be overwritten by other projected
surfaces...

What you are asking for is called "hidden surface removal" (HSR). This
is a lot harder to achieve and harder to explain :
http://en.wikipedia.org/wiki/Hidden...e_determination

For HSR to work, it must be done by the tier that has knowledge of the
3D objects, (Matlab in this case). Attempting to do it at the (final) 2D
drawing is too late. Hence your question is a Matlab question, not a
Postscript question.

 ________________________________________
________________
François Robert

Thank you, I know a software that performs this kind of operation :
DeepExploration.
It works pretty well for Finite Element Meshes fro instance but I am
not sure if it can handle
colors or not. Anyway, its possible inputs are vrml or equivalent 3D
formats and the process
from matlab is not very easy to implement. As you mentionned, maybe I
should ask
Matlab for an improvment of their tool in that direction.

Pluton

On 31 Mrz., 17:37, pluton <plutones...@gmail.com> wrote:
>
> I think I already ask this question in the past but I cannot convince
> myself that it remains without answer.

The old thread is here: http://groups.google.com/group/comp...e7b47e66f5ef784

> The old thread is here:http://groups.google.com/group/comp...
/thr...

thank you. The thing is that what I mentionned in this thread is not
what I want anymore.
When I consider matlab objects to be saved in eps format (as real
vector object and not raster images
embedded in eps files), they are just either : surfaces constructed
with triangles (with a unique color
per triangle), or 3D bars or patches. I do not think it is a very
difficult task in this situation
to calculate which object belongs to the foreground. Anyway, it looks
like it is a matlab task
more than an .eps task. I'll send mathworks an email to ask them what
they think....

Thank you,

Pluton