VrayRT how it works

[Guest]

I am looking for resources on how Vray RT works. Much like this http://www.spot3d.com/vray/help/150SP1/gimethods.htm. An explanation behind the rendering technology. Is there such an article?

[Guest]

I could be asking the wrong question here as VrayRT could be one of those mentioned din the article above, how ever I am not sure. I cant seem to find any clear information on it.

[Alan Iglesias]

In my (admittedly very basic) understanding, as far as fundamental rendering equations go RT is very much like the original Vray, which is discussed in the article you refer to.

The primary differences would be that first, RT uses a Progressive Path Tracing technique that employs no "biasing" (much like the "Brute Force" technique in the article), and second, with RT/GPU the renderer compiles all of the scene rendering information to a file that the hardware can understand directly, in this case the "cores" on the GPU card. The cores, which are like little separate parallel computers process the data, and send back rays-per-pixel image information to the display. Generally, and for both RT/CPU and RT/GPU the more rays computed per pixel, the less noise in the image.

That's a pretty simple description, and if I blew it somewhere I hope that Vlado corrects me...

-Alan

[Guest]

Thanks Alan, my understanding was that brute force doesn't use PPT? it uses something else which I am not sure of either. It would be good to get a simple explanation of each For example:

1) Irradiance map traces rays from the camera
2) Light cache traces rays from the camera
3) Photon mapping traces rays from lights
4) Brute force...
5) Vray RT...
etc...

What uses bi directional path tracing? I have even heard tri directional path tracing as well. How does Vray and VrayRT differ from say iray, bunkspeed and Maxwell? Finding a write up on all these rendering solutions is proving difficult. I am trying to piece together bits form here, there and everywhere just to get a general understanding.

[vlado]

Algorithmically, V-Ray RT and iray are unidirectional path tracers (paths always start from the camera); bunkspeed uses iray so same thing; Maxwell is a bidirectional path tracer (paths start both from the camera and the lights and are connected in some suitable way).

The production V-Ray itself can work in many ways depending on the settings; it can be both unidirectional (when using brute force/irradiance map/light cache), and bidirectional (when using the VRayBPTracer, or e.g. photon mapping).

However, even if two renderers use the same base algorithm, there are many details that may differ between them. What is important is not so much what particular algorithm is used, but that they all solve the same general problem - the rendering equation. The different methods just offer certain advantages when computing certain effects, or some speed improvements. All methods are "correct" in the sense that, when properly implemented, you will get a correct (or very close approximation to the correct) solution to the rendering equation.

Another important thing to remember is that "unbiased" is not the same as "correct". A renderer may be unbiased, but still produce a completely wrong result (perhaps a very precise and detailed result, but still wrong). On the reverse, a renderer may be biased, but still produce quite accurate results.

Best regards,
Vlado

[Alan Iglesias]

Brute Force and PPT are basically the same in that they both render "exact" data with no biasing like IR Maps and Light Caching and such. The difference is that Brute Force performs an adaptation and then renders a bucket, whereas PPT just continuously renders the whole image in a progressive fashion, until you tell it to stop via a time limit or paths-per-pixel limit. Vlado has the good details...

-Alan

[Guest]

Thanks both for a detailed explanation. It has helped a lot.