Welcome dear readers. After 2.5 years, we met again with the creators of the V-Ray renderer and talked with them in anticipation of release of newest the third one version of the renderer.
RENDER.RU: Hi Vlado, nice to meet you again at our on-line magazine. It has been about 2.5 years since our last interview and we have many questions about the achievements of your company. Tell us briefly what happened with Chaos Group during the last two years?
Vlado: The last two and a half years were very exciting as we continue developing state-of-the-art rendering solutions for the media, entertainment and design industries.
In 2011, we acquired ASGVIS and consolidated the development of V-Ray for Rhino and V-Ray for SketchUp. This is when we established our office in the US.
The next year we released V-Ray 1.5 for Rhino and V-Ray for Softimage and opened an office in Japan.
So far 2013 is also very successful. A few months ago we released version 2.0 of V-Ray for SketchUp and the beta program of V-Ray 3.0 for 3ds Max. V-Ray 2.0 for Rhino is coming in the following weeks and we are also working on V-Ray 3.0 for Maya.
RENDER.RU: With the release of V-Ray 2.0 it became clear that this rendering system can be used not only to create beautiful architectural design and visualization, but also proved to be excellent as a tool for animation and visual effects in movies (VFX). What do you think is the reason for this? Has this wide use given any new ideas to implement in V-Ray 3.0?
Vlado: The reason for this is in V-Ray's versatility – the renderer’s features are powerful enough to create stunning architectural visualizations and it is also a great tool for the animation and VFX industries. V-Ray can be easily used on a large-scale production, but it is also simple enough for an individual user to wield it to great effects.
RENDER.RU: Today many famous studios are already using V-Ray. As far as we know, such a legendary studio like ILM has used V-Ray in their work. Can you tell us in which projects V-Ray has proved its expert visualization features?
Vlado: V-Ray touches some of the most creative projects in the world. International studios like Digital Domain, Pixomondo, Zoic, ILM, Method Studios have used actively V-Ray to create the special visual effects for movies like Star Trek Into Darkness, Pacific Rim, The Lone Ranger, Game of Thrones, The Avengers, Battleship, Snow White and the Huntsman, Spiderman, Looper, etc.
RENDER.RU: What part of the functionality and features that were developed for major studios and key partners are you planning to include in V-Ray 3.0?
Vlado: Because V-Ray is being used more and more for creature work, we had to pay special attention to rendering hair and fur effects efficiently, as well as to improving our sub-surface scattering material. We already had support for various hair and fur systems (Hair&Fur, Ornatrix and HairFarm in 3ds max, and Maya hair, Shave&Haircut and Yeti in Maya); for V-Ray 3.0 we focused on optimizing the performance and memory usage. Many improvements were also added to the sub-surface scattering material, including a raytraced method for calculating multiple scattering. We also implemented support for deep OpenEXR2 output – in fact, V-Ray is one of the first renderers to include this capability out of the box.
V-Ray Feature Films Showreel 2013
RENDER.RU: The performance is greatly optimized in the V-Ray 3.0. What are the main features of the renderer that have been optimized or developed?
Vlado: We looked at many customer scenes and profiled them to find out the bottlenecks in our code. We tried to minimize redundant calculations and to optimize the most common use cases. These optimizations affect V-Ray materials, lights, and the V-Ray core itself. We fine-tuned V-Ray's sampling routines to make them better-suited to brute-force raytracing. Finally, we also implemented some new algorithms like probabilistic light sampling for scenes with many light sources.
RENDER.RU: New version provides users a new approach to constructing the rendered image. When using multi-core processors have completed the core rendering their region will no longer stand and take part in the visualization of smaller regions. How did you come to that decision, and is it true that this is one of the first implementations of modern solutions for imaging?
Vlado: This is only the initial implementation of our dynamic bucket splitting algorithm. We implemented this because a very common situation for our users is that towards the end of the image, as only a few buckets remained, the full CPU power was not utilized efficiently. We will continue to improve the implementation throughout the V-Ray 3.0 product cycle.
RENDER.RU: Besides optimization of classical algorithms, in the V-Ray 3.0 you decided to implement library Embree support. Tell us more about the other tools in V-Ray which can activate the Embree option and how this option optimizes the performance?
Vlado: Embree is a high-performance open source raycasting library developed by Intel. When Embree was released, we tested it against our own raycasting implementation and found out that Embree is significantly faster (although with somewhat lower precision than our code). As an experiment, we released a helper plugin for V-Ray 2.x that replaces the default V-Ray raycaster so that we could test the performance in real-world use cases. The results were very promising bringing up to 25% speed improvement, depending on the scene. For V-Ray 3.0, we integrated Embree in the V-Ray core so that now it is available as an option in the V-Ray settings. For the moment, we still use our old code for some types of geometry, but we will gradually implement Embree for more and more of our code.
V-Ray VFX & Game Cinematics Showreel 2013
RENDER.RU: Intel has a special solution - Xeon Phi. With the release of the new version of V-Ray is it possible to expect the ability to use this co-processor (accelerator) in the rendering process and computing? Or will there be support for this hardware solution?
Vlado: We have looked at the Xeon Phi coprocessor card, and while we did compile a version of V-Ray for it, we came up against some issues that need additional work to resolve. It is definitely easier to program for the Xeon Phi compared to GPUs, but specialized code still needs to be written for best results. Additionally, the development tools for the Xeon Phi need some improvement to help us get maximum performance from the card.
RENDER.RU: Now, a number of classic rendering tools began implementing GPU rendering support. For example, Ambient Occlusion or Indirect Illumination. Do you plan to implement these features in the upcoming versions of V-Ray, and how do you see the further development of the GPU technology in the modern rendering solutions?
Vlado: NVidia would of course want to do this for mental ray at any cost, but for the moment we are not considering using the GPU for the regular V-Ray production renderer. There are just too many things going on, including custom shaders and geometry, volumetric effects, 3rd party plugins etc., and it is not possible for the moment to support these on the GPU. Instead, we continue to focus on bringing V-Ray RT GPU as close as possible to a full production renderer.
V-Ray Advertising Showreel 2013
RENDER.RU: Besides optimization of algorithms and the renderer engine, you will implement a lot of new solutions into the new version, in particular support for Alembic, OSL and OpenColorIO. Why did you choose these technologies and how they will be integrated into V-Ray 3.0?
Vlado: OpenColorIO and Alembic support were in fact requested by our users. OpenColorIO is integrated both in the V-Ray frame buffer as a display color correction, and as a texture map for applying OpenColorIO profiles to other textures. Alembic support is added as an additional file format in the VRayProxy object. For the initial 3.0 release we have focused on supporting polygonal meshes and hair primitives from Alembic files; we will add support for other Alembic features like particles and material assignments later on in the V-Ray 3.0 product cycle. As far as OSL is concerned, we have been looking for a suitable shading language for V-Ray for a long time now. Of course, it has always been possible to write shaders for V-Ray in C++, but this is a complicated process with many technicalities. We looked at OSL initially when it was first released, however back then it had some implementation specifics that we didn't like and instead chose to implement support for GLSL on our own and also to license MetaSL from nVidia. Unfortunately, MetaSL didn't develop in the way we expected and for V-Ray 3.0 we decided to drop that and instead implement support for OSL, as it seemed to be getting some more traction in CG industry and our original concerns were already addressed. At the same time we are also keeping and developing our GLSL implementation, as it turned out very useful for some of our customers – they could take their existing GLSL shaders and use them directly in V-Ray with little or no modification.
RENDER.RU: Recently, we met with the representative of the PIXAR Animation Studio, about OpenSubdiv technology. Do you plan to implement support for OpenSubdiv in V-Ray? For example, in the V-Ray for Maya version.
Vlado: In fact we already have support for OpenSubdiv in our internal builds of V-Ray for Maya and this will be included officially in V-Ray 3.0 for Maya. We are still keeping our own subdivision code in parallel to that as it has some advantages over OpenSubdiv right now, but I'm hoping that OpenSubdiv will eventually be developed further and be more suitable for high-quality offline rendering.
RENDER.RU: V-Ray has become a universal tool for visualization; there are versions for a variety of computer graphics packages, including the free Blender. Tell us a little more about the development of V-Ray for Blender, when do you plan to release it and whether the plug-in for Blender will be free, and you only need to purchase V-Ray Standalone?
Vlado: Blender is becoming an increasingly useful tool for a wide variety of applications; I myself know people that are using it very successfully for professional visualization in conjunction with V-Ray for Blender. For these reasons, we wanted to have a solid implementation of V-Ray inside Blender and we talked to Andrey Izrantsev – the original developer of V-Ray for Blender – and asked him if he is willing to join Chaos Software to focus full time on this project. We were very pleased when he agreed and currently we are working on preparing the first sort of official release of V-Ray for Blender. The connection between Blender and V-Ray continues to be free and open source; only a copy of V-Ray Standalone is needed for rendering. In addition, with V-Ray 3.0 we are moving to universal render licenses and we are including V-Ray Standalone as part of many of our products. This means for example that owners of V-Ray 3.0 for 3ds Max or Maya can try V-Ray for Blender right away.
RENDER.RU: In addition to a number of popular computer graphics programs, V-Ray is not yet available for the Houdini. Having in mind the increasing popularity of Houdini in the film and VFX industry, do you plan to develop V-Ray for Houdini and Phoenix FD for Houdini?
Vlado: Houdini is an important tool for visual effects and provides excellent tools for all kinds of dynamics simulations. From that point of view I don't think there is much point in developing Phoenix FD for Houdini, as there are already very good fluid and water simulators and renderers in that package. As far as V-Ray is concerned, we have been asked many times to look into integration. We may get to that at one point, but for the moment we don't have the resources to dedicate to that. Houdini is a very complicated product and a good implementation of V-Ray would require a lot of effort. At the same time, Houdini is really good at supporting common formats like OpenVDB, Alembic and OpenEXR2 which makes it relatively easy to take data out of Houdini and integrate it with rendering in other packages.
V-Ray Animation Showreel 2013
RENDER.RU: Do you plan V-Ray to supports any of the CAD systems, such as ArchiCAD, or Autodesk Revit? There were rumors that Chaos Group partners are working on V-Ray for these CAD systems.
Vlado: Unfortunately there is nothing that I can share publicly in that regard right now. I hope that there will be more information in the coming months.
After talking with Vlado, we did not miss the opportunity to ask a few questions to representatives of Marketing Department of Chaos Group. With the release of the third version of V-Ray, we asked a couple of questions regarding the new licensing model. They said our old friend Javid Imanov.
New Licensing Model in V-Ray 3.0
RENDER.RU: We have discussed with you almost all the major technological innovations of the new version of V-Ray and even a look into a possible future. The final version of V-Ray 3.0 for 3ds Max is on the horizon. But it is necessary to discuss the licensing model, with version 3.0, it will be available in the new format. Please tell us in detail about it. How much will change the price will be cheaper for the end user or remain the same?
Javid: With the release of V-Ray 3.0 we are introducing universal Render Nodes to support distributed rendering on multiple host applications. This is a major part of our roadmap as it includes the development of universal V-Ray assets that can be shared across multiple 3D platforms.
The new pricing system is built to open up new levels of flexibility for users at every end of the spectrum. For example knowing that a lot of our user base consists of individual artists and small teams, we decided to drop the workstation license price by more than 20%. We saw this as a way to help them cut overhead for new hires, and pay for only what they need. We also put a rental policy in place for situations where quick deadlines require immediate scaling of your system.
RENDER.RU: As now there is support for distributed rendering, will need to purchase a separate license for each of the compute nodes?
Javid: With V-Ray 3.0 there will be no distinction between render licenses for local, network, or distributed rendering. All of these rendering options are covered by the new Render Node license, making it possible to seamlessly use your license across multiple 3D platforms.
By purchasing 1 copy of V-Ray 3.0 Workstation for 3ds Max, users get the ability to use V-Ray’s Graphical User Interface (GUI) for editing their settings on one machine and also to render on one machine as a Render Node.
The rendering can be performed either on the same machine where the GUI license is used, or on another machine in the network.
In order to use additional computers for rendering though, you will now have to purchase additional V-Ray 3.0 Render nodes.
We thank Vlado and Javid for answering our questions and wish the company Chaos Group to further develop and scale new heights. And with the release of the fourth version, we will once again communicate with Vlado and his colleagues.