I've finally decided to hunker down and do model some complex backgrounds.
In so doing, I'm considering which povray objects are the most efficient.

For example, I know making everything out of isosurfaces would be really
slow.   I'm assuming "modelled"  laboratory equipment then exported to
triangles would render much faster than asking povray to do the complex CSG.


Right now, I'm trying to figure out which renders & parses faster:
1) a bicubic patch
2) a really dense triangle mesh of the same object (Hamapatch export to
DXF).


Actually, I have two side questions about modellers:
i) I'm surprised at the limited number of triangles that show up in a
hamapatch export to DXF.  There's a setting of "4" subsurfaces: I'd have
liked to try 16 or more.  Any tips?

ii) Does POVLAB work with WinXP?   For about  a decade now, I've had trouble
getting the program to run at all....

Post a reply to this message

"Greg M. Johnson" <gregj;-)565### [at] aolcom> wrote in message
news:3f10696d$1@news.povray.org...

>   I'm assuming "modelled"  laboratory equipment then exported to
> triangles would render much faster than asking povray to do the complex
CSG.
>


What I meant here is:

One could make a complex object using boxes & cylinders in povray with
dozens of CSG statements.  I'm assuming for render time, using a formal
modelling program like Moray, POVLAB, or Hamapatch should win hands down in
terms of render time.  Am I correct.

Post a reply to this message

I've never used hamapatch, but I use autocad a lot... dxf format leaves much
to be desired.  If you can export as 3ds you might get better results.


--
Douglas C. Eichenberg
dou### [at] nlsnet
http://www.getinfo.net/douge

Post a reply to this message

Greg M. Johnson <gregj;-)565### [at] aolcom> wrote in message
news:3f10696d$1@news.povray.org...
>
> Right now, I'm trying to figure out which renders & parses faster:
> 1) a bicubic patch
> 2) a really dense triangle mesh of the same object (Hamapatch export to
> DXF).

Don't know about render times, guess my models aren't that complex, the
stuff I've exported from hamapatch seemed to render about the same as
modeling in POV, the biggest difference was the parse time.  I've found that
parse times on either bicubic patch or mesh is considerably longer than CSG.
Lately I find I only use Hamapatch when I want to do something I simply
don't know how to do in CSG.

>
> Actually, I have two side questions about modellers:
> i) I'm surprised at the limited number of triangles that show up in a
> hamapatch export to DXF.  There's a setting of "4" subsurfaces: I'd have
> liked to try 16 or more.  Any tips?

Have you tried Wings3d? ( http://www.wings3d.com/ ) It's quite a capable
modeler on its own, and I also find it useful for subdivision on stuff I've
modeled in other programs.   If you have something modeled in Hamapatch,
rather than subdividing the whole object, export as .obj, import into Wings
and subdivide only the parts that need fine detail. Then convert to POV
using Poseray.

RG

Post a reply to this message

In article <3f1069e2$1@news.povray.org>,
 "Greg M. Johnson" <gregj;-)565### [at] aolcom> wrote:

> Right now, I'm trying to figure out which renders & parses faster:
> 1) a bicubic patch
> 2) a really dense triangle mesh of the same object (Hamapatch export to
> DXF).

Answer: it depends. The patch can precompute a lot of data, making 
memory and speed about equal to mesh. That precomputation takes parse 
time, bringing it closer to that of a high res mesh. How close? I don't 
know, and it probably depends on what settings you use.


> For example, I know making everything out of isosurfaces would be really
> slow.   I'm assuming "modelled"  laboratory equipment then exported to
> triangles would render much faster than asking povray to do the complex CSG.

> One could make a complex object using boxes & cylinders in povray with
> dozens of CSG statements.  I'm assuming for render time, using a formal
> modelling program like Moray, POVLAB, or Hamapatch should win hands down in
> terms of render time.  Am I correct.

No. You're asking POV to replace a relatively simple boolean operation 
on a group of primitives with a search through a much larger group of 
triangles.
For some cases it could be correct, but for others it is not. Boxes, 
spheres, and cylinders are all quite fast, you would have to get a 
pretty complex CSG before a mesh equivalent would be faster. If you're 
using a complex and slow isosurface or a sweep of thousands of spheres, 
a precomputed mesh becomes a better option. Of course, the mesh will be 
much slower to parse.

You're most likely "prematurely optimizing". You should look for a 
faster way if it is too slow, but scenes are usually slow for reasons 
other than CSG and data parsing.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: chr### [at] tagpovrayorg
http://tag.povray.org/

Post a reply to this message

"Christopher James Huff" <cja### [at] earthlinknet> wrote

> You're most likely "prematurely optimizing". You should look for a
> faster way if it is too slow, but scenes are usually slow for reasons
> other than CSG and data parsing.
>

Yes, I'm trying to choose the optimum tool before I even start making
things.

Thanks for the advice all.  Maybe I'll stick with hamapatch.

Okay Chris, so what's the likely biggest contributor to slow scenes?

Post a reply to this message

In article <3f11b45b$1@news.povray.org>,
 "Greg M. Johnson" <gregj;-)565### [at] aolcom> wrote:

> Okay Chris, so what's the likely biggest contributor to slow scenes?

It depends on the scene. With isocacti.pov, it's isosurfaces. That scene 
could be a lot faster with mesh or patch geometry. (Of course, since the 
purpose of the scene is to demo isosurfaces, that's not really an option)

For newdiffract.pov, it's photons. Making the geometry in that scene out 
of meshes or patches would most likely only slow it down further.

Meshes are mainly used in modellers for these reasons: it is trivial to 
do a high-speed, low quality render using wireframe or a scanline 
renderer. If you are doing a scanline final render, you have to convert 
it to triangles anyway. And using a mesh opens up the possibility of 
doing displacements and other manipulation that you can't easily do on 
primitives. And finally, most other modellers use meshes.

If you're modelling something like a test tube, it's a simple CSG of two 
spheres and two cylinders (and maybe a torus to round the rim). You 
aren't doing anything that needs meshes, and the shape will probably 
render more quickly as a CSG. But if you're doing something like a car 
body or human figure, a mesh is the best way to go.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: chr### [at] tagpovrayorg
http://tag.povray.org/

Post a reply to this message

3f10696d$1@news.povray.org...
> i) I'm surprised at the limited number of triangles that show up in a
> hamapatch export to DXF.  There's a setting of "4" subsurfaces: I'd have
> liked to try 16 or more.  Any tips?

As you seem to work with DXF, there is a simple solution.
You only have to put the sPatch plugins in a folder called
"export plugins" under Hamapatch main folder.
When you want to export your object select from the menu:
File -> sPatch plugin -> DXF or Layered DXF (wolfie).
When it exports, it give you options for a higher subdivision
level than the Hamapatch in-built exporter.
You can get the sPatch plugins (and more utilities) here:
www.geocities.com/hamapatch/links/hamautil.html

Hope this helps.
Bye

Txemi Jendrix
http://www.txemijendrix.com

Post a reply to this message