 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Peter Popov wrote:
>
> You've missed one. A zero-thickness box (yes it works). Try something
> like box { <-1,0,-1>, <1,0,1> } and see what happens. It should render
> faster than a regular box.
>
I didn't know, it is possible ( similar to cylinder ) but definitely
worth trying.
Christoph
--
Christoph Hormann <chr### [at] gmx de>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Mike Williams wrote:
>
> I'd guess that the vast majority of the tracing time would be consumed
> by looking for the bounding boxes which have been automatically
> generated. Each ray has to be checked against all 1000 bounding boxes,
> but once the bounding boxes have been located, the ray only has to be
> checked against at most one actual object. (Because the object exactly
> fills the bounding box, if there's no reflection or transparency the ray
> will always terminate at the first object). So the construction details
> of your object would only be expected to have a tiny influence the trace
> time.
>
your argumentation sounds quite reasonable, even though objects are
partly transparent in this case :-) I also looked more on the memory
point, where differences are much larger.
Christoph
--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Christoph Hormann <Chr### [at] schunteretctu-bsde> wrote:
: AFAIK with large meshes, mesh { triangle... } will also be much better
: than union { triangle...}
A mesh takes a lot less memory than a union of triangles.
Perhaps a bit surprisingly a union does not render considerably slower than
a mesh (although one could think that a mesh could have some optimizations).
--
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Have you tested if there's any difference in render time between putting
all those 1000 rectangles into a union or separately in the scene?
Povray may be able to make some bounding box optimizations with unions,
but I may remember wrong.
--
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Question:
is the majority of the memory is used by the rectangle or not?
Because one mesh of an object or 1000 mesh (copies) of the
same mesh use almost the same amount of memory. So if the
answer of my question is yes I dont understand why there is no
more difference between mesh and box for the memory peak.
Fabian.
>
> Because I needed some simple rectangle shape for a scene I am working on
> and because it is used about 1000 times in the scene, I tried some
> different possibilities building that shape and checked
> memory-use/tracing time.
>
> It is just a very quick test on a concrete scene, differences in time
> are quite low, so you probably can not say anything definitive from
> these values (did not want to wait longer :-). BTW, scene also contains
> many other objects.
>
> Any comments/ideas ?
>
> Christoph
>
> Objects I tested:
>
> //------------------------------
>
> mesh {
> triangle{<0.0,0.0,0.0>,<0.0,0.0,1.0>,<1.0,0.0,1.0>}
> triangle{<0.0,0.0,0.0>,<1.0,0.0,0.0>,<1.0,0.0,1.0>}
> }
>
> //------------------------------
>
> union {
> triangle{<0.0,0.0,0.0>,<0.0,0.0,1.0>,<1.0,0.0,1.0>}
> triangle{<0.0,0.0,0.0>,<1.0,0.0,0.0>,<1.0,0.0,1.0>}
> }
>
> //------------------------------
>
> box {
> <0, -0.00001, 0>
> <1, 0.00001, 1>
> }
>
> //------------------------------
>
> plane {
> y, 0.0
> hollow on
> clipped_by { box
> { <0, -0.1, 0>
> <1, 0.1, 1> } }
> }
>
> //------------------------------
>
> polygon {
> 5, <0, 0>, <0, 1>, <1, 1>, <1, 0>, <0, 0>
> rotate 90*x
> }
>
> //------------------------------
>
> Time/memory data:
>
> ###################################################################
> -------------------------------------------------------------------
> mesh
> -------------------------------------------------------------------
> Peak memory used: 47840394 bytes
> -------------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 10.0 seconds (10 seconds)
> Time For Trace: 0 hours 3 minutes 7.0 seconds (187 seconds)
> Total Time: 0 hours 3 minutes 17.0 seconds (197 seconds)
> ###################################################################
> -------------------------------------------------------------------
> union
> -------------------------------------------------------------------
> Peak memory used: 47803698 bytes
> -------------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 11.0 seconds (11 seconds)
> Time For Trace: 0 hours 3 minutes 3.0 seconds (183 seconds)
> Total Time: 0 hours 3 minutes 14.0 seconds (194 seconds)
> ###################################################################
> -------------------------------------------------------------------
> box
> -------------------------------------------------------------------
> Peak memory used: 47920778 bytes
> -------------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 10.0 seconds (10 seconds)
> Time For Trace: 0 hours 3 minutes 2.0 seconds (182 seconds)
> Total Time: 0 hours 3 minutes 12.0 seconds (192 seconds)
> ###################################################################
> -------------------------------------------------------------------
> clipped plane
> -------------------------------------------------------------------
> Peak memory used: 48296618 bytes
> -------------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 10.0 seconds (10 seconds)
> Time For Trace: 0 hours 3 minutes 0.0 seconds (180 seconds)
> Total Time: 0 hours 3 minutes 10.0 seconds (190 seconds)
> ###################################################################
> -------------------------------------------------------------------
> polygon
> -------------------------------------------------------------------
> Peak memory used: 47882238 bytes
> -------------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 10.0 seconds (10 seconds)
> Time For Trace: 0 hours 2 minutes 57.0 seconds (177 seconds)
> Total Time: 0 hours 3 minutes 7.0 seconds (187 seconds)
> ###################################################################
>
> --
> Christoph Hormann <chr### [at] gmxde>
> Homepage: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Fabian BRAU <Fab### [at] umhacbe> wrote:
: Because one mesh of an object or 1000 mesh (copies) of the
: same mesh use almost the same amount of memory.
With that small mesh, not really. There has to be a way to say "a copy
of this mesh is located here". This 'handle' with information about the
copy may take even more memory than that mesh with two triangles.
--
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thanks for the clarification.
Fabian.
>
> Fabian BRAU <Fab### [at] umhacbe> wrote:
> : Because one mesh of an object or 1000 mesh (copies) of the
> : same mesh use almost the same amount of memory.
>
> With that small mesh, not really. There has to be a way to say "a copy
> of this mesh is located here". This 'handle' with information about the
> copy may take even more memory than that mesh with two triangles.
>
> --
> main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
> ):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Warp wrote:
>
> Have you tested if there's any difference in render time between putting
> all those 1000 rectangles into a union or separately in the scene?
> Povray may be able to make some bounding box optimizations with unions,
> but I may remember wrong.
>
The rectangles are already structured in unions.
I tested different arrangements concerning this point (larger/smaller
blocks of rectangles) and there was only small difference in memory use,
but time was the same (+/- 1 sec)
Christoph
--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I made a more specific test with a scene containing 11620 rectangles
(and only those rectangles :-).
These are the results:
-----------------------------------------------------------------
mesh
-----------------------------------------------------------------
Peak memory used: 20078124 bytes
-----------------------------------------------------------------
Time For Parse: 0 hours 0 minutes 3.0 seconds (3 seconds)
Time For Trace: 0 hours 1 minutes 13.0 seconds (73 seconds)
Total Time: 0 hours 1 minutes 16.0 seconds (76 seconds)
-----------------------------------------------------------------
triangle union
-----------------------------------------------------------------
Peak memory used: 20313423 bytes
-----------------------------------------------------------------
Time For Parse: 0 hours 0 minutes 4.0 seconds (4 seconds)
Time For Trace: 0 hours 1 minutes 35.0 seconds (95 seconds)
Total Time: 0 hours 1 minutes 39.0 seconds (99 seconds)
-----------------------------------------------------------------
polygon
-----------------------------------------------------------------
Peak memory used: 20260648 bytes
-----------------------------------------------------------------
Time For Parse: 0 hours 0 minutes 3.0 seconds (3 seconds)
Time For Trace: 0 hours 1 minutes 9.0 seconds (69 seconds)
Total Time: 0 hours 1 minutes 12.0 seconds (72 seconds)
-----------------------------------------------------------------
box 0 height
-----------------------------------------------------------------
Peak memory used: 20450420 bytes
-----------------------------------------------------------------
Time For Parse: 0 hours 0 minutes 2.0 seconds (2 seconds)
Time For Trace: 0 hours 1 minutes 7.0 seconds (67 seconds)
Total Time: 0 hours 1 minutes 9.0 seconds (69 seconds)
-----------------------------------------------------------------
clipped plane
-----------------------------------------------------------------
Peak memory used: 22095452 bytes
-----------------------------------------------------------------
Time For Parse: 0 hours 0 minutes 3.0 seconds (3 seconds)
Time For Trace: 0 hours 1 minutes 9.0 seconds (69 seconds)
Total Time: 0 hours 1 minutes 12.0 seconds (72 seconds)
--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Ahhh! The zero-height box wins :) I'll be using those for my clouds now
:)
~Sam
Christoph Hormann wrote:
> -----------------------------------------------------------------
> polygon
> -----------------------------------------------------------------
> Peak memory used: 20260648 bytes
> -----------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 3.0 seconds (3 seconds)
> Time For Trace: 0 hours 1 minutes 9.0 seconds (69 seconds)
> Total Time: 0 hours 1 minutes 12.0 seconds (72 seconds)
>
> -----------------------------------------------------------------
> box 0 height
> -----------------------------------------------------------------
> Peak memory used: 20450420 bytes
> -----------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 2.0 seconds (2 seconds)
> Time For Trace: 0 hours 1 minutes 7.0 seconds (67 seconds)
> Total Time: 0 hours 1 minutes 9.0 seconds (69 seconds)
>
> -----------------------------------------------------------------
> clipped plane
> -----------------------------------------------------------------
> Peak memory used: 22095452 bytes
> -----------------------------------------------------------------
> Time For Parse: 0 hours 0 minutes 3.0 seconds (3 seconds)
> Time For Trace: 0 hours 1 minutes 9.0 seconds (69 seconds)
> Total Time: 0 hours 1 minutes 12.0 seconds (72 seconds)
>
> --
> Christoph Hormann <chr### [at] gmxde>
> Homepage: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
