Stochastic sampling of light transport paths is key to Monte Carlo forward rendering, and previous studies have led to mature techniques capable of drawing high-contribution light paths in complex scenes. These sampling techniques have also been applied to differentiable rendering. In this paper, we demonstrate that path sampling techniques developed for forward rendering can become inefficient for differentiable rendering of glossy materials---especially when estimating derivatives with respect to global scene geometries. To address this problem, we introduce antithetic sampling of BSDFs and light-transport paths, allowing significantly faster convergence and can be easily integrated into existing differentiable rendering pipelines. We validate our method by comparing our derivative estimates to those generated with existing unbiased techniques. Further, we demonstrate the effectiveness of our technique by providing equal-quality and equal-time comparisons with existing sampling methods.

Abstract We present a new SAH guided approach to subdividing triangles as the scene is coarsely partitioned into smaller sets of spatially coherent triangles. Our triangle split approach is integrated into the partitioning stage of a fast BVH construction algorithm, but may as well be used as a stand alone pre‐split pass. Our algorithm significantly reduces the number of split triangles compared to previous methods, while at the same time improving ray tracing performance compared to competing fast BVH construction techniques. We compare performance on Intel's Embree ray tracer and show that BVH construction with our splitting algorithm is always faster than Embree's pre‐split construction algorithm. We also show that our algorithm builds significantly improved quality trees that deliver higher ray tracing performance. Our algorithm is implemented into Embree's open source ray tracing framework, and the source code will be released late 2015.

In this paper, we present a meshing scheme for the implementation of an adaptive tessellation of triangular meshes on the graphics processing unit (GPU). Tessellation is performed according to a local test to generate primitives dynamically. The refinement procedure does not require the pre-computation of any refinement pattern. The resulting adaptive procedure is efficient and simple, and generates the tessellation pattern of each triangle dynamically. The non-recursive strategy simplifies mesh reconstruction, avoids irregular memory access patterns and uses a simple control flow to make it a good candidate to guide the evolution of tessellation algorithms on future graphics cards. To test our proposal, we have used the geometry shader of current GPUs to generate primitives dynamically.

We propose a generalized approach to decoupling shading from visibility sampling in graphics pipelines, which we call decoupled sampling . Decoupled sampling enables stochastic supersampling of motion and defocus blur at reduced shading cost, as well as controllable or adaptive shading rates which trade off shading quality for performance. It can be thought of as a generalization of multisample antialiasing (MSAA) to support complex and dynamic mappings from visibility to shading samples, as introduced by motion and defocus blur and adaptive shading. It works by defining a many-to-one hash from visibility to shading samples, and using a buffer to memoize shading samples and exploit reuse across visibility samples. Decoupled sampling is inspired by the Reyes rendering architecture, but like traditional graphics pipelines, it shades fragments rather than micropolygon vertices, decoupling shading from the geometry sampling rate. Also unlike Reyes, decoupled sampling only shades fragments after precise computation of visibility, reducing overshading. We present extensions of two modern graphics pipelines to support decoupled sampling: a GPU-style sort-last fragment architecture, and a Larrabee-style sort-middle pipeline. We study the architectural implications of decoupled sampling and blur, and derive end-to-end performance estimates on real applications through an instrumented functional simulator. We demonstrate high-quality motion and defocus blur, as well as variable and adaptive shading rates.

This paper presents a novel algorithm and architectures for perspective correct displacement of the surface geometry of a polygonal model using a displacement map. This new displaced surface geometry is passed onto a traditional rendering pipeline. The algorithm uses a multiple pass approach in which the geometry is displaced in the first pass and then the displaced geometry is rendered. The significant features of the algorithm are that the surface is displaced after its triangle mesh is transformed into screen space and that it uses only bi‐linear interpolation for calculating the displaced geometry allowing a cheap incremental scan‐line implementation. A hardware architecture based on this algorithm is presented along with possible alternative implementations. The technique presented here allows greater photorealism by using increased detail without an increase in bandwidth for geometry or calculation time for transformation.

This paper presents a hardware design that processes several rays simultaneously to improve the memory access and compositing performance for three dimensional image synthesis using ray casting. By processing multiple rays, the memory access coherence of a group of neighbouring rays can be exploited and fewer page misses in a modern SDRAM memory device are encountered. Tracing multiple rays also avoids the data hazard that occurs when tracing the path of a single ray. Implementation of the presented design in a real time volume rendering system improves the performance of the system by increasing the number of frames rendered per second.

This paper presents an architecture capable of generating two dimensional images of volume data at video rates. This architecture fits into an image synthesis pipeline and uses only simple arithmetic operations and a look-up table to generate two dimensional images in real time. The shading algorithm is the grey-level gradient algorithm for shading volume data. Hardware simulations of the critical components can process a 256/sup 3/ data set at 20 frames a second using a parallel implementation. Image synthesis from volume data in real time is an important technique in visualization and graphics systems.