I feel like a broken record responding to this thread with basically the same info, but the reality has not changed.
Essentially, most renderers that are now being accelerated for GPUs are largely raytracing-driven. Raytracing is inherently easily computed in parallel at a massive scale and also well adapted to use on GPUs. Since raytracers were some of the first things to be implemented with general purpose GPU computation experiments, there are many example implementations and things to learn from or base your own work off of, and this is why there are now tons of these renderers.
While TG2 does use some raytracing elements, the majority of rendering works differently, more akin to Renderman's REYES or other micropolygon approaches. And while other micropolygon approaches are *similar*, they are not the same, so even though other systems may be adapted to GPUs, there is still a lot of work and forging of new ground for us to convert TG2's renderer to be able to work on GPUs. We can't just take off-the-shelf techniques and implement them and get a sudden boost in speed.
The important thing is we do keep up on the research and competition and would love to have this kind of technology implemented in Terragen. Hopefully there will be an opportunity to do this in the future.
- Oshyan
