VRay 3.0 - towards full path tracing

Vray 3.0 - bit unexpected change of direction, for me at least. I found their biased integrators were a strong point of Vray, i guess the involvement into film industry change the game a bit.
Surprised they introduced Vertex Merging

-Brute force rendering speed improvements
-Max ray intensity
-Automatic asset transfer for DR – No more missing maps EVER!
-Render Subset
-Trace sets
-Opens Shading Language
-Metaballs (isosurfaces)
-Deep images and OpenEXR 2.0 support
-Maya Viewport in Maya viewport
-Vertex merging
-Vismat materials
-Progressive rendering

http://vray.info/news/article.asp?ID=553

Yes seems like a very strong release. I can’t think of any scene that vray 3.0 won’t be able to handle, it’s so feature rich and it has so many GI algorithms that the “strong and weak points” will be more likely “only strong points”… cause which are the weak points of this release compared to other render engines?

read this in the article…

This is still an ongoing development, but examples shown already display significant speed improvements. The classroom scene below took 15m44s in V-Ray 2.4, and 5m55s in the 3.0 beta using Embree. That is 2.66 times faster brute force rendering!

Ahhh, there running the Embree raycaster on intel’s.
Wow, that’s a huge gain in render time.

Most people rendering animation use Brute Force, so it shouldn’t be that much of a surprise they focus on improvements there.

Seems to me that they are betting more on GPU rendering from now on. That is interesting to say the least, I was under the impression that biased techniques rendered faster.

I was very surprised by this announcement but as long as it renders faster and retains the level of control we had before then it’s all good.

These techniques loose advantage as we keep increasing geometry density, eg grass, trees and hair.

Biased vs. Unbiased, still have to know how to light. Vray was fast in the first place anyway.

Will interesting to see how it holds up against Arnold. Happy to see vray is going for BF. I was never much a fan of it having too many render settings.

Does it have to hold up against anything?

Yes, especially when the field of production path tracers expanding as quickly as it is. Renderman offers a path tracing solution, Arnold is leading the pack, things like Appleseed and Cycles are both close to being production ready AND open source. VRay will have to show something impressive to make this feature worth upgrading for.

the lights would behave the same unbiased and biased tells you more about the kind of errors you get in a render. With an unbiased engine you deal with noise and that’s about it with photon mapper or something like BI you can get all sorts of other artifacts. If you are using physically based lights and physically based materials the render should look more or less the same.

this book helped me finally understand the differencehttp://www.amazon.com/Realistic-Image-Synthesis-Photon-Mapping/dp/1568814623

@m9105826 and tyrantmonkey:
WHAT?

@m9105826: hasn’t impressed you thus far?
@tyrant monkey: check my sketchbook, please find some noise

As far as I know, the algorithm they implemented for their unbiased rendering (the one that combines path tracing and progressive photon mapping) is currently one of the best available due to its handling of specular-diffuse-specular paths.

Granted, Cycles can also handle these paths to an extent when ‘filter glossy’ is used, but it still requires a lot more samples and more time than if you just used that algorithm.

@VickyM72 perharps I misunderstood your post but the when I read it it seemed to me that you meant the would be difference in a render done with vray with biased integrator versus unbiased one in terms of lighting. My point was that if the both used the same material system and the lights were physically based the only difference would be in the nature of the artifacts you would have to deal with

@tyrant monkey: When it comes down to a cool render, it’s not because the engine was ‘biased’ or ‘unbiased’, or ‘physically accurate’, it’s usually because the person knows how to light. Sounds simple, but it is as simple as that.

Do I read correctly that we could potentially support the “vismat” material format for Blender?