Thea Presto is an engine running entirely on the GPU. And the development is based on Nvidia CUDA, which means that a compatible graphic card is needed to run it. All the CUDA supporting cards 1.x/2.x/3.x can be used but, unfortunately, the engine will not run on ATI/Intel graphic cards.

Thea Presto will come as a separate plugin for Thea Render. So, Thea Render continues to be the versatile platform-independent render we all know and if one installs Presto, he/she gets the extra engine as well.

Presto and Features

Engine Characteristics

Presto comes with 2 variants; one for very fast preview with ambient occlusion and another photorealistic unbiased-like mode. They both are progressive render modes.

Interactiveness

One thing that is impressive about Presto is the interactiveness rate that one can achieve. It is the most fun engine to run the interactive render on, with a fantastic response (particularly the case if you have a separate graphic card for display).

Material / Lighting System

The same quality you get from Thea CPU engines, you will get them from Presto as well. Presto has been coded on the GPU with the same principles for materials/lights like on the CPU. No loss of quality.

Multiple GPUs and Network

Multiple GPUs are supported (only in darkroom rendering currently) along with all the GPUs you can setup in your network. With the new network render mode which allows to render frames separately on each node, one can render complete product and exterior animations in a few hours (something not possible before).

Product Design

We have found that the GPU engine is particularly useful for product design and external scenery with dominant direct lighting. This is where it particularly shines.

The Next Steps

Adding Missing Features

One cannot expect that the whole list of features of Thea Render can be coded inside the GPU engine within the previous time framework, much more when there was a generic development effort in many directions at the same time (Adaptive BSD reboot, Plugins & SDK, UI adjustments). Also, one cannot expect that the complexity of the TR1/TR2 unbiased engines can be coded inside a GPU engine, not even on the most recent graphic card generations. There are certain features that we decided that we can't live without and need to be there right from the beginning and others we will be adding with time; features such as SSS, instancing, procedurals and texture layering are currently missing. A detailed list of supported features will be available with the new release.

CUDA vs OpenCL

In the beginning there was only one path for us, the standard path, that is OpenCL. But we have found out (the hard way) that OpenCL is not really "tuned" for complex staff. There is a varying performance and quite a lot of fighting with getting the code to actually run. Nevertheless, this is the path we would like to follow and this is why, the language used during programming has been made with OpenCL in mind and to exactly help the transition to the standard when it is more mature. We will be visiting this topic frequently to see how we can make Presto running on ATI/Intel graphic cards.

And for the last part, the Rolex watch animation. Patrick could add quite a few images and animations in my post here, but for the time being we wanted only a minimal announcement. The new release is coming next week, so you can all judge for yourselves.