studioGPU

Breaking out of the adjust-render-view loop

I’m going to date myself when I say this, but I remember the day when using desktop-based 3D and digital imaging for video and film production in Hollywood was considered a revolutionary concept. Times have obviously changed and products like Photoshop, Maya, and Final Cut Pro are routinely used in everything from pre-visualizations to final renders. Same thing goes for architectural visualization and product design.

But the desktop 3D workflow is still constrained by an “adjust-render-view” loop.  Basically for each and every change of camera position, character performance, texture and material or lighting, you have to adjust the settings, let it render (seconds for low res, minutes to hours for film res) and then view the results.  Change the position or lighting for even one object and to visualize it in context, you render again.

The ability to go at full-throttle and try new ideas in an organic way without disrupting your creative work flow, remains for most part, the exclusive domain of dedicated systems using expensive multi-processor hardware.

To be sure things have gotten much better then even a few years ago. Many programs now offer fast renderers (typically stand-alone) or use external renderers (e.g. Renderman, Mental Ray, etc.) Gelato (no longer being developed) even started using the GPU to accelerate several key aspects of rendering.  And with the fast fast cards form ATI and Nvidia, things are definitely improved.

But when you have a vision in your own mind of what a presentation or render should be, or if you just want to come up with it interactively, none of the above gives you what you really need. Once past the modeling stage, the visualization and rendering is still tedious and requires an enormous amount of patience (which I sorely lack).

StudioGPU looked at this problem and asked, not just how do we render faster, but how do we break out of the adjust-render-view loop?  That is what led to MachStudio Pro.

I learned about an early version of MachStudio Pro at Siggraph 2008. From what I saw, I was cautiously optimistic that StudioGPU was on to something—using the GPU to do pretty much everything related to setup and rendering.  It interested me so much that I actually contacted the company and said I wanted to work with them. So here I am today.

So what is MachStudio Pro - the condensed version of the story?  The goal in MachStudio Pro is that setup time and rendering time should be practically simultaneous.  The user can move the 3D models in real-time. Camera and character placement, material adjustment and lighting set-up are all elements that can be changed using the integrated GUI without having to program, setup, stop and wait to see the results.

How well it succeeds at letting you “Work at the Speed of Thought”  is best evaluated by checking out our industry showcase,  watching the video, or of course, seeing a live demo.  Work with the product yourself and let us know how well it achieves the “real-time” goal.

This blog will be both a place for us to share our thoughts and experiences as MachStudio Pro goes out “into the wild”, and also to provide tips and tricks from creative users as well as our own software engineers. We strongly encourage you to post comments, ask questions, tell us your own story, or refer us to other developments in real-time rendering.