Hi Tracers!

Playing around with seed() and rand(), I soon started to explore my
computer's limits of performance by programming scenes with several
hundred thousands of primitives... and I was badly surprised how few (!)
objects I can render! With 128 megs of RAM, 210.000 objects obviously
seems to be the end of the flagpole, now with 320 megs, the upper limit
should be around 525.000 or so - but instead, already at 480.000 objects
I get the message "Couldn't allocate XXXXXX bytes for building slab".

On the other hand, I often see scenes with several million (or even
several dozens of million) objects in this newsgroup, knowing they were
made on machines not excessively much larger then mine (for example,
even Mick Hazelgrove uses only about three times the RAM I own - and
nevertheless successfully renders scenes with 30 million grass leaves!),
so I begin to ask myself what I do wrong...

Is there any way (e. g. some special tracing options - I looked them up
the in Manual, but it doesn't talk about memory usage anywhere) to
optimize RAM usage?

Another question deals with the image attached here: how to do realistic
dunes? The large-scale structures of my dunes are already quite
satisfying (I used the "turbulence" function in Heightfield Lab to
generate the heightfield) - but how can I do the small ripples covering
the windward sides of the dunes? Obviously, neither ripples nor wave
normals do the trick...

And one last question: is there a way to make infinite heightfields
(like the isosurfaces in MegaPOV) in official PoV-Ray?

See you in Khyberspace -
http://www.geocities.com/electricafghan/index-e.html

Yadgar

Post a reply to this message

Attachments:
Download 'arrakis03.png' (428 KB)

Preview of image 'arrakis03.png'

On Wed, 08 Aug 2001 22:22:48 +0200, Yadgar <j.b### [at] ndhnet> wrote:

>Mick Hazelgrove uses only about three times the RAM I own - and
>nevertheless successfully renders scenes with 30 million grass leaves!),

Mick's grass is made of meshes. Copies of the same mesh only take a
few dozens of bytes (to store a pointer and a transformation, I think)
so if you make a 1000 leaves mesh and copy it 30000 times to get 30
million, it will only eat as much memory as a 1000 leaves mesh + a
little for the 30000 copies (even if it were 1k per copy, and it's
much lower than that, that would be 30 megs, and that's nothing)

>Is there any way (e. g. some special tracing options - I looked them up
>the in Manual, but it doesn't talk about memory usage anywhere) to
>optimize RAM usage?

If you're desperate, you can turn light buffers off... though this may
greatly slow down your scene.

OTOH people are buying 768 megs of RAM for less than $128, so...

>Another question deals with the image attached here: how to do realistic
>dunes? The large-scale structures of my dunes are already quite
>satisfying (I used the "turbulence" function in Heightfield Lab to
>generate the heightfield) - but how can I do the small ripples covering
>the windward sides of the dunes? Obviously, neither ripples nor wave
>normals do the trick...

How about a marble normal? Have you tried that?

>And one last question: is there a way to make infinite heightfields
>(like the isosurfaces in MegaPOV) in official PoV-Ray?

Don't think so. Also, isosurfaces in MegaPOV are not infinite, they
are always contained in an object.


Peter Popov ICQ : 15002700
Personal e-mail : pet### [at] vipbg
TAG      e-mail : pet### [at] tagpovrayorg

Post a reply to this message

Peter has it - it took me a long time to realise that a mesh could be
repeated many times. I use a file to generate a mesh and then use loops and
trace to repeat it many times and place it on a HF. I suggest a close study
of Gilles grass macro will explain all.

Mick

Post a reply to this message

could not view this.

didnt show in OE, and PSP7 choked... Not going to install the mamoth
photoshop...


--
Rick

Kitty5 WebDesign - http://Kitty5.com
Hi-Impact database driven web site design & e-commerce
TEL : +44 (01625) 266358 - FAX : +44 (01625) 611913 - ICQ : 15776037
POV-Ray News & Resources - http://Povray.co.uk

PGP Public Key
http://pgpkeys.mit.edu:11371/pks/lookup?op=get&search=0x231E1CEA

Post a reply to this message

"Rick [Kitty5]" wrote:
> 
> could not view this.
> 
> didnt show in OE, and PSP7 choked... Not going to install the mamoth
> photoshop...

It displayed fine in Netscape and Picture Publisher for me...

-- 
David Fontaine  <dav### [at] faricynet>  ICQ 55354965
My raytracing gallery:  http://davidf.faricy.net/

Post a reply to this message

The GIMP (win32) opened it, but complained "File Corrupted?"

Post a reply to this message

Hi Mick and all others!

Does that mean that I can save such an amount of memory even when I
construct my standard "primitive" (e. g. a box) from triangles and then
place and scale it randomly - or does it work only with *identical*
copies?

See you in Khyberspace -
http://www.geocities.com/electricafghan/index-e.html

Yadgar

Post a reply to this message

On Thu, 9 Aug 2001 16:52:49 -0400, Daniel Lin wrote:
>The GIMP (win32) opened it, but complained "File Corrupted?"

I just got a black image in ImageMagic. 

--
Cheers
Steve              email mailto:ste### [at] zeroppsuklinuxnet

%HAV-A-NICEDAY Error not enough coffee  0 pps. 

web http://www.zeropps.uklinux.net/

or  http://start.at/zero-pps

 11:55pm  up 27 days,  1:58,  2 users,  load average: 1.00, 1.00, 1.01

Post a reply to this message

On Fri, 10 Aug 2001 07:10:42 +0200, Yadgar <j.b### [at] ndhnet> wrote:

>Does that mean that I can save such an amount of memory even when I
>construct my standard "primitive" (e. g. a box) from triangles and then
>place and scale it randomly 

A box only takes two vectors (six floats + two pointers), a sphere
takes a vector and two floars (center, radius and radius squared) etc.
Even if you do some savings this way, they will be marginal.

>- or does it work only with *identical* copies?

Identical in the sense that they have the same base object but
different transformations, since POV will then only store the original
once and then the transformation for each copy.


Peter Popov ICQ : 15002700
Personal e-mail : pet### [at] vipbg
TAG      e-mail : pet### [at] tagpovrayorg

Post a reply to this message

Yadgar <j.b### [at] ndhnet> wrote:
: Does that mean that I can save such an amount of memory even when I
: construct my standard "primitive" (e. g. a box) from triangles and then
: place and scale it randomly - or does it work only with *identical*
: copies?

  As Peter said, you will not get too much saving from making one primitive
with a mesh and then copying it.
  The real memory saving with meshes come when you can make a bigger mesh
(eg. one containing some hundreds or thousands of "primitives") and then
copy this bigger mesh.

  Also note that this type of big mesh renders a lot faster than the
equivalent group of primitives.

  You can apply transformations (scale, rotate and translate) to the copies
with no problems.

-- 
#macro N(D,I)#if(I<6)cylinder{M()#local D[I]=div(D[I],104);M().5,2pigment{
rgb M()}}N(D,(D[I]>99?I:I+1))#end#end#macro M()<mod(D[I],13)-6,mod(div(D[I
],13),8)-3,10>#end blob{N(array[6]{11117333955,
7382340,3358,3900569407,970,4254934330},0)}//                     - Warp -

Post a reply to this message