Last year we did an experiment and tried to see if we could get V-Ray to run inside of Motionbuilder with V-Ray RT by using our App SDK when working with Kevin Margo on CONSTRUCT. It was a strange and interesting experiment. At the same time, it got a lot of people’s attention, including us showing it at Siggraph Real Time Live. The issue has always been that the system we had is tied to the hardware you bring with you. In general a desktop workstation can only hold around 4 GPUs. While that should be enough cores for anyone, the resolution and quality of our demo pushed us to see what we could do with more. A lot more.
Just like last year, it was an opportunity for us to try and push some new technology we have been developing. Where do we go from here? Well, to the cloud of course. V-Ray cloud is an infrastructure that allows you to take V-Ray App SDK and scale it through distribution over any network, including a public Internet network. In this particular case, what we did was to feed our motion builder stream of data, send it through the Internet to NVIDIA’s data center in Santa Clara, distribute the task of rendering to multiple VCAs on the Cluster, recompose the image and send it back to use through a standard Web page. The results can be seen in the video we made where we build a makeshift system in the garage.