What parts were misleading? I didn't see anything really that was misleading. Most of the images they used were from real-time demos, and later on in the presentation they even did a live real time demo, changing settings on the fly in front of the crowd.

http://www.caustic.com/pdf/...

page 5.

The images aren't equal. Both images should be equivalent to each other with regards to the technical abilities of both rendering styles. ie, the same assets, same resolutions, and if both scenes are dynamic then they should be able to run at the same frame-rate. That would also imply all proper techniques that rasterization has at hand to increase fidelity and quality in the image.

It's only a personal opinion about the subject. I'd love to see what the QuakeRT team could do with the same hardware.

What do you mean a visual cap?

Will it be impossible to produce accurate lighting and reflections in photorealistic game visuals? In other words, is there a limit to how realistic games can look using rasterization?

Oh of course. We are already pretty close to the limit of rasterisation. There's a ton of work, and programming tricks implemented to make rasterisation look like it is today in some games.

When it comes to lighting/reflections/transpare ncy, rasterisation is very limited. With Ray-Tracing and ambient occlusion, your able to simulate light in game how it would act in real life. And all you have to do is define material and surface parameters for objects and the ray-tracer does the rest. It makes creating amazing graphics extremely simple. But currently with a huge performance hit.

That's what companies like Cuastic Graphics are trying to improve is the performance, in the hopes of making Ray-Tracing a much more viable option for everyone in all industries, including gaming.