Most games consider the collision artifacts of semi-transparent particles and the environment acceptable. This is due to the abundant hardware that has been available that performs the basic interpolation functions.

In Software Development Kits (SDKs) such as OpenGL or DirectX, there have been render states defined that simply do the blend function with one function call and have been available for many years. Unfortunately, art departments have been working overtime trying to get basic particle effects in the game and looking correct to account for this one line of code.

In recent years, there has been a significant leap forward with the software developer's ability to define the functions used to render a scene in the hardware: shaders. Even on cheap consumer graphics hardware, the software developer can define almost any function imaginable (usually only limited by available registers and functions made available to the shading language).

Using pixel shaders, render-to-texture technology and a little bit of math, I claim that we can more correctly simulate the volumes that the particles were intended to represent.

The problem with current methods of rendering particles to represent volumes is that, like most geometry in games, they are not volumes. They are camera facing planes rendered with the intention to never show their depth.

Now we all know that their depth does indeed show itself if that simulated depth ever comes close to other geometry defined with real depth. The problem arises due to the fact that we are aliasing the depth into those camera facing polygons. We are taking the visual characteristics of a volume in space and flattening it to a plane, then rendering the plane back in a world with real volume characteristics.

Let's assume that we don't actually want to do this. Let's also assume that we don't want to significantly change the art pipeline the artist would perform on their first pass.

What do we need to better represent these volumes? Assumptions and math. For the purposes of this article, the following assumptions are used:

1. That the contents of each particle are intended to be uniform density. That is if we were to take a sample in the actual volume that the particle texture was rendered from, it would be the same density as every other sample within the volume.
2. That there are exactly two real-world boundaries implicitly defined for each pixel in the source texture. The boundaries are the near and far extents of the volume that the pixel is intended to represent.
3. That there exists a plane that defines a mirror image of the boundary of the volume for each pixel. (This assumption will disappear in a later example).
4. That the user has access to the actual depth values rendered for the scene occluded by the particle within the pixel shader.