Hello, I'm working on a brush stroke renderer. I have a number of particles in world space and I want to render a brush stroke at each particle location. A brush stroke is just a simple rectangular RGBA image. My first idea was to use a billboard for each brushstroke. But this leads to different brush stroke sizes for different depth values (because of perspective projection). I could solve this by scaling each brush stroke to ensure a constant size in screenspace, but I don't know how to derive the correct scaling factors. But maybe this is not the way to do it. My question is: how can I render an image at a certain position in worldspace with an fixed screen-space size?

you should use the orthographic mode which effectively makes the openGL viewport a 2D area. then you don't have to worry about scaling or any of that stuff.

there's a tutorial on this page about using ortho mode:
http://www.gametutorials.com/Tutorials/OpenGL/OpenGL_Pg2.htm

-me

Quote:Original post by Palidine
you should use the orthographic mode which effectively makes the openGL viewport a 2D area. then you don't have to worry about scaling or any of that stuff.

there's a tutorial on this page about using ortho mode:
http://www.gametutorials.com/Tutorials/OpenGL/OpenGL_Pg2.htm

-me

I'm quite new to OpenGL so I could be mistaken, but I think this doesn't solve my problem because the locations of the brush strokes should be perspective projected. For example: I have two equal sized spheres (in world space). Each sphere is rendered by a number of brush strokes. The two spheres should be perspective projected: the sphere furthest away should be smaller (in screen-space), but the brush strokes that make up the spheres should all be the same size.

[Edited by - crocomire on September 23, 2004 1:23:58 PM]

Quote:Original post by Hellsp4wn
An idea would be to use the gluProject and gluUnproject functions.
You can then convert a 3d coordinate into screen coordinate, and then render the quad in orthographic view.

Don't ask me how to use them thought, I never remember...

Is this the idea?:
1) gluProject using a perspective projection matrix
2) gluUnProject using a orthographics projection matrix
3) render the resulting coordinates using a orthographic projection matrix