Ray Tracing

Nowadays, my 3D libraries are very robust. Probably they are almost so mature as my 2D vector based libraries, although they are less than two years old. Some weeks ago, I thought it was already time to create my own ray tracer, a really powerful one.

Well, it is under development. It's already fully functional and has very professional quality. Now I have going through many challenges, because I don't want it the way it is.

I need something very beyond what I already got: I want the same quality, but at least 20 times faster. 3D rendering optimization is a very hard work if you need it in a very deep level. This subject is heavily researched by many people and it has been a very important area of Computer Graphics for at least 20 years. Unfortunately, my experience shows the best algorithms are never published for good.

Anyway, I need to overcome this challenge, or will not be able to publish graphic tools based on ray tracing. My software has NO dependencies, so it uses NO libraries or even hardware acceleration. It needs to be VERY fast in its own essence and I'm working on it.

A text with 20 letters made using a complex font can have several thousands of polygons. To get professional quality, even when rendering a small image, the software needs to test dozens of millions of ray-object intersections. If everything goes as I want, I will be able to render complex shapes like 3D text in a reasonable time, so I can release it at Picture to People site.

Below I show you an image rendered using all power of my ray tracer (there is no background plane).

Subscribe in a reader