saluk wrote on Mar 19, 2019, 01:38:
Your terminology is bogus. A "raycast" is just the general process of sending a ray from a point in a direction and doing something when it hits something. Raytracing uses raycasts as part of the math.
The raytracing in dx12 and currently supported by RTX is true raytracing in every sense. If you tell it to produce a ray for every pixel on the screen, and feed in the proper materials for the objects, and allow each ray to bounce and refract as much as it should, you will get exactly the same kind of render that say blender or max produce.
Or if you are making a quake2 mod, just do it all
First off, the Quake 2 mod doesn't do it all, it actually shows off how hard it is to convert over current games. Missing player weapon flashes, sparks, player and opponent shadows. Though in fairness, they say player weapon flashes should be easy to add. There is even light artifacting caused by the de-noise filter.
My terminology comes from the graphics community, which is extremely picky about terminology because of how different the implementations are between the approaches.
I had however not kept up on the terminology, as what was originally called "ray cast ray tracing" is now called "path tracing."
RTX works a lot differently from what Blender natively does. Blender does real ray tracing. RTX does path tracing. I say Blender natively, as Blender supports external rendering engines and a path tracing engine could be plugged in.
Quoting one of the blender developers when asked about RTX,
It is a path tracing engine. Ray tracing engine can not do caustic, soft shadows, global illumination, etc... while a path tracing engine will give you all of these without any cheating. Both of these engine types shoot rays from the camera but the similarities end there.
Some people argue that path tracing is superior. From Wikipedia:
Path tracing naturally simulates many effects that have to be specifically added to other methods (conventional ray tracing or scanline rendering)
While the majority of the graphics community are still behind ray tracing because even though they need to cheat a little bit, path tracing creates grainy images, no matter the number of passes used. The only way around this is to smooth out the image, but then you are introducing other graphic artifacts. Also path tracing suffers from not properly scattering light from volumetric things like heat from fire, something ray tracing has no problems with.
Though some people still point out that when trying new cheats in ray tracing, they generally use a path traced image as reference.
Ray tracing is still used in the majority of films and TV where CGI elements blend with filmed elements. However, many fully CGI movies and TV shows have now gone with various forms of path tracing.
Depending on who you talk to, the real future for TV and movies is volumetric path tracing. This solves the problems of light scattering caused by non-visible elements, but also currently has quite a speed cost to the solution. It still doesn't address the how grainy the image is.