Releases
v1.2.0
SDK
Features and Improvements
Substantial code and API refactor
Added support for Vulkan
Added support for Linux (x86 and ARM)
Added support for bindless resource access models
Infinite Scrolling Volumes (ISV) feature is now available (no longer beta)
Improved multi-volume support, including GPU workload batching
Improved shaders with define driven feature and resource declarations
Shaders can now be used directly in any codebase that supports D3D12 or Vulkan, regardless of resource binding model
Improved Probe Classification accuracy and increased performance gains when classification is enabled
Moved DDGIVolume constants to structured buffer(s)
Reduced the size of DDGIVolume constants by 2x
Merged per-probe relocation and classification data into a single texture resource
Removed the R8 UINT texture resource used with classification
Substantial CMake and build system improvements
Removed unnecessary third party dependencies
Refactored and simplified the defines system
Refreshed and improved documentation and converted to Markdown
Bug Fixes
Infinite scrolling volumes now properly clear irradiance and distance data for probes that are scrolled
Probe classification now accurately tests each probe ray's hit distance against the probe voxel's bounding planes
Fixed probe rays no longer perform unnecessary shading and lighting
Fixed regression related to distance blending and sampling
Test Harness
Features and Improvements
Substantial code redesign and overhaul
Added support for Vulkan
Added support for Linux (x86 and ARM)
Added support for a bindless resource access in all shaders
Added support for compressing textures to BC7 format on the CPU or GPU (using DirectXTex)
Added support for creating texture mipmaps (using DirectXTex)
Added support for using BC7 compressed textures on the GPU
Added support for using mipmapped textures on the GPU
Added support for an arbitrary number of lights
Added hot shader reloading
Added primary ray texture level of detail with ray differentials
Added a binary scene cache system for fast runtime scene loading
Added cross-platform windowing and input with GLFW
Added CPU and GPU timestamp instrumentation
Moved RTXGI DDGI indirect lighting pass to compute
All GPU resources moved off of the upload heap to the device (default heap)
Substantial CMake and build system improvements
Third party dependencies now included as proper Git Submodules
DXC binaries now delivered through packman
Refactored and simplified defines system
Added new "Tunnel" and "Cornell Boxes" test scenes
Refreshed and improved the configuration file system
Refreshed and improved UI
Bug Fixes
Fixed vsync and fullscreen crashing issues
You can’t perform that action at this time.