Some days ago Jaime Vives, with the suggestion of Jerry Anning, processed a
macro to turn lathes to mesh. The intention was to reduce the time to
render.

I decided to study the subject. I made some tests with different
alternatives on a simple bottle:

- Mesh with the macro of Jaime Vives
- Mesh with the help of Spatch. First DXF from Spatch and later 3DWin to
turn to pov mesh with smooth triangles.
- Bicubic patch with Spatch.
- Lathe.

I use a resolution of 320 x 240 no AA.  The results were:

                                          Time Render       File Size
Mesh (Macro)                        3:29                   16 Kb
Mesh (Spatch - 3DWin)        3:31                 604 Kb
Bicubic patch (Spatch)         6:58                 110 Kb
Lathe                                       8:44                     4 Kb


Conclusion: The macro is more efficient; smaller time to render and reduced
volume of the archives.

Post a reply to this message

Guillermo Espitia Rojas wrote:
> 
> 
>                                           Time Render       File Size
> Mesh (Macro)                        3:29                   16 Kb
> Mesh (Spatch - 3DWin)        3:31                 604 Kb
> Bicubic patch (Spatch)         6:58                 110 Kb
> Lathe                                       8:44                     4 Kb
> 
> Conclusion: The macro is more efficient; smaller time to render and reduced
> volume of the archives.

Question concerning conclusion:

You mentioned you used a very poor image resolution
for the tests. I can forgive that but I am curious
at to how each process looked in comparison. One of
the very nice features of a lathe object is the high
order math functions used in their implementaion.
The surface smoothness is equal to that of the sphere
object. Meshes, patches, polygons, and the other methods
notoriously have a problem with faceting because they
are constructed from flat surfaces.

  So I guess the question is are they all comparable
or are you trading quality for speed ?

Ken Tyler

Post a reply to this message

>Question concerning conclusion:
>
>You mentioned you used a very poor image resolution
>for the tests. I can forgive that but I am curious
>at to how each process looked in comparison. One of
>the very nice features of a lathe object is the high
>order math functions used in their implementaion.
>The surface smoothness is equal to that of the sphere
>object. Meshes, patches, polygons, and the other methods
>notoriously have a problem with faceting because they
>are constructed from flat surfaces.
>
>  So I guess the question is are they all comparable
>or are you trading quality for speed ?
>
>Ken Tyler

Hi Ken:

His observation is very opportune.  At my conclusion i not did consider the
characteristics of the lathe than you indicates.

The objective initial was check the benefits of convert a lathe to mesh,
suggestted by Vives-Anning. At second instance, i pretend know techniques to
reduce the time of render and reduce the necessities of hardware. This me
allows do one image with more amount of elements, keepping a time reasonable
of render.  Obviously is sacrificed the quality by the quickness.  However,
at one image good studied the reduction of quality it can pass unnoticed.
This can be observed at the following images:

- Lathe to mesh:      http://www.ctav.es/jaime/barrabar.jpg

- Patch to height-field:
http://oz.irtc.org/ftp/pub/stills/1998-02-28/temple.jpg

Thanks by his attention.

Guillermo Espitia

Post a reply to this message

Guillermo Espitia Rojas wrote:
> 
> Hi Ken:
> 
> His observation is very opportune.  At my conclusion i not did consider the
> characteristics of the lathe than you indicates.

I get lucky sometimes :)

 
> The objective initial was check the benefits of convert a lathe to mesh,
> suggestted by Vives-Anning. At second instance, i pretend know techniques to
> reduce the time of render and reduce the necessities of hardware. This me
> allows do one image with more amount of elements, keepping a time reasonable
> of render.  Obviously is sacrificed the quality by the quickness.  However,
> at one image good studied the reduction of quality it can pass unnoticed.
> This can be observed at the following images:

 
> - Lathe to mesh:      http://www.ctav.es/jaime/barrabar.jpg

I don't know about the conversion process but the composition
and lighting in this scene are truly fantastic.

 
> - Patch to height-field:
> http://oz.irtc.org/ftp/pub/stills/1998-02-28/temple.jpg

That was created using patches originaly ? Wow ! I had
always presumed it was done using a height field. Is the
source for this file available somewhere ?


Thank you for replying to my questions.

Ken Tyler

Post a reply to this message

Ken wrote:

> One of
> the very nice features of a lathe object is the high
> order math functions used in their implementaion.
> The surface smoothness is equal to that of the sphere
> object. Meshes, patches, polygons, and the other methods
> notoriously have a problem with faceting because they
> are constructed from flat surfaces.
> 
>   So I guess the question is are they all comparable
> or are you trading quality for speed ?

Jaime Vives Piqueres wrote meshlath.inc because the lathes, at least
when used to make glass objects, were slow enough to make the scene
unrenderable.  With meshes, the bottles were much faster.  The
triangulation method of meshlath.inc is partly based on that in my old
cylindrical height field include file (and some neat spline macros). 
With this method you can generate subpixel size triangles if you want. 
Obviously, you have no faceting artifacts if you do this.  You can
adjust the fineness to taste.  Incidentally the include generates
smooth_triangles, further reducing any potential faceting.  The downside
is that you take a parse time hit with superfine meshing.  On the other
hand, the render time improvement of mesh over lathe, in my experience
so far, still produces a net speed benefit (and avoids some potential
"issues" with lathes).  Another benefit is the ability to use Cardinal
splines, Beta splines, Tau splines and others by simply dropping in a
different basis matrix.  This allows smoother and more controllable
shapes, depending on what you are modelling.  Jaime has created a very
useful tool.

Jerry Anning
cle### [at] dholcom

Post a reply to this message

Well, this "efficiency #macro" seems to be very practical (I'd like to try
it in conjunction with mesh-pcm?).

I did try for one half hour to read all texts of Jerry Anning on tha most
of povray's forum, but couldn't be able to find this #macro "meshlath.inc".
Even not in Twisted Site. Is it possible to find it anywhere?

Thank in advance,



<3679B89E.D5090BEB@dhol.com>...
> 
>
> Jaime has created a very
> useful tool.
> 
> Jerry Anning
> cle### [at] dholcom
> 

David Etter Perere
per### [at] bofinfomaniakch
remove the "bof."

Post a reply to this message

David Etter Perere wrote:
> couldn't be able to find this #macro "meshlath.inc".
> Even not in Twisted Site. Is it possible to find it anywhere?

The macro itself was written by Jaime Vives Piqueres.  You can find on
his site at: http://www.ctav.es/jaime/ .

Jerry Anning
cle### [at] dholcom

Post a reply to this message

Thank's for the link. And beatiful images.
I'm really interested by the "POVRay accelerator board" and his soft.


> The macro itself was written by Jaime Vives Piqueres.  You can find on
> his site at: http://www.ctav.es/jaime/ .
> 
> Jerry Anning
> cle### [at] dholcom
> 

David Etter Perere
per### [at] bofinfomaniakch
remove the "bof."

Post a reply to this message

Jerry Anning wrote:
> 
> David Etter Perere wrote:
> > couldn't be able to find this #macro "meshlath.inc".
> > Even not in Twisted Site. Is it possible to find it anywhere?
> 
> The macro itself was written by Jaime Vives Piqueres.  You can find on
> his site at: http://www.ctav.es/jaime/ .

Jeesh!  Every time I go to his web site there is another stunning masterpiece
posted there..... 

http://www.ctav.es/jaime/barrabar.jpg is one of the best POV scenes I have
ever had the pleasure to view.  The incredible attention to detail clearly
made use of his macro.... I don't think there is a wasted pixel in the
scene.

Dan

Post a reply to this message