POV-Ray : Newsgroups : povray.binaries.images : [Geometry] Orthogonal projection (61k) Server Time
16 Nov 2024 10:28:21 EST (-0500)
  [Geometry] Orthogonal projection (61k) (Message 1 to 3 of 3)  
From: Simon
Subject: [Geometry] Orthogonal projection (61k)
Date: 18 May 2005 11:06:14
Message: <428b59e6@news.povray.org>
Hi there,
  I made some tweakings to my point to plane projection algorithm, now 
the normal can move and this way I had been able to clearly understand 
that a projection of any point on a given plane, is, an orthogonal 
projection - like what the ortho camera does in povray.  Sorry if you 
had got it before me, i'm still learning all this geometry stuff...

  So you may ask "Well, it's nice, how can it be useful inside my 
scene?" and I should say it really depends on your scene, but for 
anybody that would like to stamp the vertexes of his 3D construction 
onto a plane and make it part of his 3D scene, then this is for you.

  Examples of what you could do: Make a 3D drawing with blobs, and stamp 
it on a wall, etc...

  This algorythm is very simple and can be simplified a little more, I 
do not know how the trace() function of povray performs, but it should 
be as efficient or better I think.  The computation of the destination 
vertex uses only 9 multiplications and 7 additions per vertex, making 
calculations simple.

  For anybody that is interested in this, I may release the C++ code 
(very easy to read) that calculates the destination point from a source 
point with a normal (the plane is always on the origin in my algo).  Or 
I may translate it into a povray macro where it should work like this:

dstPoint = ortho_projection(srcPoint, NormalVector)

  If you're interested, let me know, if now, let me go on to my next 
project, which is to create a super efficient algorythm for displaying 
height fields in opengl (can effectively be used in povray to reduce 
triangle count in mesh), using various algorythm such as dynamic LOD 
(Level of Detail), etc...

Hope you like it!
  Simon


Post a reply to this message


Attachments:
Download 'ortho_projection.jpg' (61 KB)

Preview of image 'ortho_projection.jpg'
ortho_projection.jpg


 

From: scott
Subject: Re: [Geometry] Orthogonal projection (61k)
Date: 18 May 2005 15:52:34
Message: <428b9d02$1@news.povray.org>
>  If you're interested, let me know, if now, let me go on to my next
> project, which is to create a super efficient algorythm for displaying
> height fields in opengl (can effectively be used in povray to reduce
> triangle count in mesh), using various algorythm such as dynamic LOD
> (Level of Detail), etc...

Google for "ROAM".  Note that because 3D cards a so fast now, in some 
instances it's actually faster to just give the 3D card all the triangles 
rather than trying to use some clever adaptive method.


Post a reply to this message

From: Simon
Subject: Re: [Geometry] Orthogonal projection (61k)
Date: 21 May 2005 20:47:03
Message: <428fd687@news.povray.org>
I already have lots of documentation on the subject and I don't quite 
agree with you.  It is true that hardware is better (and getting even 
better) today, but optimisation will always be necessary.

The other reason this algorythm will be necessary is that by reducing 
the number of triangles used to display the same scene, i can actually 
increase the quality of the scene, thus increasing the number of 
triangles up to what my hardware (new or old) can support.

Simon


scott wrote:

>> If you're interested, let me know, if now, let me go on to my next
>>project, which is to create a super efficient algorythm for displaying
>>height fields in opengl (can effectively be used in povray to reduce
>>triangle count in mesh), using various algorythm such as dynamic LOD
>>(Level of Detail), etc...
>>    
>>
>
>Google for "ROAM".  Note that because 3D cards a so fast now, in some 
>instances it's actually faster to just give the 3D card all the triangles 
>rather than trying to use some clever adaptive method.
>
>
>
>  
>


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.