• Tuesday August 16th, 2016
  • Blogs

Precompute performance improvements

Enlighten works by precomputing the visibility of static geometry (light transport) in the scene. Following the precompute, lights and materials can change at runtime and the global illumination will update in real-time.  A scene correctly set up for Enlighten should take no longer than one hour to precompute.  With 3.06, this precompute time has been reduced by up to 10% for balanced scenes and more for cases where CreatePreClustering is the longest stage.

Automatic system generation

When calculating the radiosity of a scene, Enlighten separates the environment into different areas called “systems”. They can be thought of as a sub-section of a level where each system is a collection of geometry that reads its lighting results from the same lightmap.  Systems are useful for improving both the precompute and the runtime performance as the engine can calculate radiosity for certain systems when others are not in use.

Previous versions of this automated process have required some level of manual intervention. This release has improved the voxel-based automatic system generation technology, making it more robust and ensuring that it will succeed for the majority of game scenes. Future work is planned to make this feature even more stable for users to work with.

Integration guide

A brand new set of tutorials and sample code have been included in this release that will make it easier for Enlighten partners to integrate the SDK into an internal game engine. It starts with a basic scene export and builds to more advanced features including multiple system processing. These tutorials are available for evaluators and customers on our secure documentation servers.

Forge performance

Large scenes in Forge are seeing up to three times faster rendering speeds and additional performance improvements have been developed for specific features including:

  • Shadow casting
  • Selecting multiple objects in the scene manager
  • Performing multi-edit on objects

Forge per pixel probe interpolation

Per pixel probe interpolation (PPPI) is a feature that improves the visual quality of large, static objects that are lit by probes. Rather than generating radiosity data for an entire surface from a single probe, per pixel probe interpolation allows a surface to use data from multiple nearby probes. This results in a more accurate, graduated lighting effect across the object.

PPPI was first seen in the Seastack Bay showcase and is now available for Forge users. Forge is an intuitive evaluation tool with which artists are able to quickly and easily set up a scene for Enlighten and see the value that real-time global illumination can bring to their assets and workflow; it also acts as a compelling example of how a team could integrate Enlighten into their engine. With the inclusion of PPPI, the radiosity quality seen in Forge is more representative of what a user would be able to achieve with Enlighten in their internal engine.

DirectX12 for PC support (Unreal Engine)

DirectX12 is the newest graphics API for Windows platforms which provides lower level access to hardware, helping developers directly manage system resources according to the application’s requirements.

Experimental support for DirectX12 for Windows 10 for PC was released with Unreal Engine 4.10 and this was further improved with the release of Unreal Engine 4.11. Enlighten now provides limited support for DirectX12 on PC and will continue development in this area.

_________________________________________________________________________________

For further information on these and other items included in Enlighten 3.06, please read the release notes.

If you have a specific feature request for Enlighten, please contact support@geomerics.com.

To request an evaluation, fill out the contact form or email sales@geomerics.com.