A GPU Based Font System

I've read this paper before and it is really nice. This is flexible, fast and has lower memory requirements than typical font rendering. My understanding is that WPF uses a similar approach if you have a DX9+ GPU.

However, there are three significant issues with this approach:

1. You lose hinting support (good luck reimplementing FreeType from scratch!)

   This is generally not a problem with larger point sizes (around 18px depending on the font and the use), but any smaller than that and text becomes essentially unreadable. WPF was universally panned for its hinting implementation and the hinting engine is being rewritten for .Net 4.0 (especially important now that Microsoft is using WPF in Visual Studio).

2. You need shaders.
3. This area is a patent minefield.

There are newer papers with different techniques that are worth looking into (no links, sorry - my main workstation is offline now). My favorite is Valve's distance maps-based rendering: it has similar memory requirements to VTM, lower quality, but degrades gracefully without shaders and is simpler to implement.

In any case, OpenTK is a great way to experiment with text rendering techniques. Much easier to explore an idea than using plain C/OpenGL.