Develop3D is “simply astounded” by Machstudio Pro running on a FirePro card

So now I am not the only one dazzled by the MachStudio Pro/FirePro package.  The Develop3D blog just posted their initial impressions of MachStudio Pro.  A quote from the article says it all: Now I’m no rendering expert, but I have to say I was gobsmacked with the speed and quality of renders, and the control you have over scenes with near instant feedback is simply astounding. Moving rendering to the GPU is an extremely exciting development and the potential to revolutionize the design visualisation workflow is huge.

I’ve got to add one thing.  If you see the product, you get this immediately, but if you are just reading about it, most people think: “ah - GPU-accelerated final renders - seems logical and we expect to see more of this trend.” But reality is, this is a lot more then accelerated final rendering.  Just a fast final render means you still do setup and compositing in the same way.  You setup, adjust, then render then try again until you get it right or run out of patience or budget.  But the way MSP uses the GPU is that it accelerates all steps of the process, not just the final render. The creative component is the same as the rendering and viewing component.  It becomes a non-linear workflow.  So just saying renders of 500-900 times faster, doesn’t really cover how big of a change this really is.

Not sure my explanation makes it any clearer so I encourage you to take a look at MachStudio Pro in the AMD booth #2417 at SIGGRAPH. From what I have read, they are going to be showing some snazzy use of one of ATI’s unique strong suits: hardware-accelerated tessellation.  According to the StudioGPU events page the SIGGRAPH demos will show “support for displacement mapping with hardware tessellation and the ability to process more than a billion polygons in real-time. Other new features being demonstrated will include the ability to create full motion blur with velocity maps, stereoscopic camera support, configurable anti-aliasing algorithms (Box, Gaussian, Mitchell) and many more.”