Fast Multi-View Rendering for Real-Time Applications

For a long time, GPUs have primarily been optimized to render more and more triangles with increasingly flexible shading. However, scene data itself has typically been generated on the CPU and then uploaded to GPU memory. Therefore, widely used techniques that generate geometry at render time on demand for the rendering of smooth and displaced surfaces were not applicable to interactive applications. As a result of recent advances in graphics hardware, in particular the GPU tessellation unit's ability to overcome this limitation, complex geometry can now be generated within the GPU's rendering pipeline on the fly. GPU hardware tessellation enables the generation of smooth parametric surfaces or application of displacement mapping in real-time applications. However, many well-established approaches in offline rendering are not directly transferable, due to the limited tessellation patterns or the parallel execution model of the tessellation stage. In this state of the art report, we provide an overview of recent work and challenges in this topic by summarizing, discussing and comparing methods for the rendering of smooth and highly detailed surfaces in real-time.

ACM Transactions on Graphics, 2015

We describe our successful initiative to accelerate Adobe Illustrator with the graphics hardware pipeline of modern GPUs. Relying on OpenGL 4.4 plus recent OpenGL extensions for advanced blend modes and first-class GPU-accelerated path rendering, we accelerate the Adobe Graphics Model (AGM) layer responsible for rendering sophisticated Illustrator scenes. Illustrator documents render in either an RGB or CMYK color mode. While GPUs are designed and optimized for RGB rendering, we orchestrate OpenGL rendering of vector content in the proper CMYK color space and accommodate the 5+ color components required. We support both non-isolated and isolated transparency groups, knockout, patterns, and arbitrary path clipping. We harness GPU tessellation to shade paths smoothly with gradient meshes. We do all this and render complex Illustrator scenes 2 to 6x faster than CPU rendering at Full HD resolutions; and 5 to 16x faster at Ultra HD resolutions.

2008

We present an architecture for rendering multiple views efficiently on a cluster of GPUs. The original scene is sampled by virtual cameras which are used later to reconstruct the desired views. We show that this image-based approach can be very scalable and support rendering at interactive rates.

IEEE Visualization 2005 - (VIS'05)

Existing parallel or remote rendering solutions rely on communicating pixels, OpenGL commands, scene-graph changes or application-specific data. We propose an intermediate solution based on a set of independent graphics primitives that use hardware shaders to specify their visual appearance. Compared to an OpenGL based approach, it reduces the complexity of the model by eliminating most fixed function parameters while giving access to the latest functionalities of graphics cards. It also suppresses the OpenGL state machine that creates data dependencies making primitive rescheduling difficult. Using a retained-mode communication protocol transmitting changes between each frame, combined with the possibility to use shaders to implement interactive data processing operations instead of sending final colors and geometry, we are able to optimize the network load. High level information such as bounding volumes is used to setup advanced schemes where primitives are issued in parallel, routed according to their visibility, merged and reordered when received for rendering. Different optimization algorithms can be efficiently implemented, saving network bandwidth or reducing texture switches for instance. We present performance results based on two VTK applications, a parallel iso-surface extraction and a parallel volume renderer. We compare our approach with Chromium. Results show that our approach leads to significantly better performance and scalability, while offering easy access to hardware accelerated rendering algorithms.

2010 3rd International Conference on Computer Science and Information Technology, 2010

Spatial-multiplex autostereoscopic systems display images interleaved from multi-view to provide adequate viewing zone sections for multiple users. The traditional methods render multiple views in multiple passes. In this paper, we present a GPU-based multi-view render (GBMVR) for spatial-multiplex autostereoscopic displays. It generates multiple views as textures in only one pass through the geometry shader and then interleaves the images through the fragment shader in another pass. It is implemented noninvasively in Chromium so that most traditional OpenGL applications like the game Quake III, without any source-code modification and re-compiling, can run directly on the Chromium enhanced with GBMVR. Configured as the sortfirst parallel rendering architecture, our method can also render images for high-resolution spatial-multiplex autostereoscopic systems that display high-resolution images with tiled projectors.

ACM Transactions on Graphics, 2018

In this paper, we present a real-time graphics pipeline implemented entirely in software on a modern GPU. As opposed to previous work, our approach features a fully-concurrent, multi-stage, streaming design with dynamic load balancing, capable of operating efficiently within bounded memory. We address issues such as primitive order, vertex reuse, and screen-space derivatives of dependent variables, which are essential to real-world applications, but have largely been ignored by comparable work in the past. The power of a software approach lies in the ability to tailor the graphics pipeline to any given application. In exploration of this potential, we design and implement four novel pipeline modifications. Evaluation of the performance of our approach on more than 100 real-world scenes collected from video games shows rendering speeds within one order of magnitude of the hardware graphics pipeline as well as significant improvements over previous work, not only in terms of capabilit...

2004

Graphics hardware is optimized for rasterizing and solving the visibility of points, lines and polygons. In this paper, we propose a GPU implementation of new graphics primitives (e.g. spheres, cylinders, ellipsoids), which are compatible with the standard pipeline. Our technique consists of two steps, both executed on the GPU:

Vis 05 Ieee Visualization 2005, 2005

We describe OpenGL Multipipe SDK (MPK), a toolkit for scalable parallel rendering based on OpenGL. MPK provides a uniform application programming interface (API) to manage scalable graphics applications across many different graphics subsystems. MPKbased applications run seamlessly from single-processor, singlepipe desktop systems to large multi-processor, multipipe scalable graphics systems. The application is oblivious of the system configuration, which can be specified through a configuration file at run time. To scale application performance, MPK uses a decomposition system that supports different modes for task partitioning and implements optimized GPU-based composition algorithms. MPK also provides a customizable image composition interface, which can be used to apply post-processing algorithms on raw pixel data obtained from executing sub-tasks on multiple graphics pipes in parallel. This can be used to implement parallel versions of any GPUbased algorithm, not necessarily used for rendering. In this paper, we motivate the need for a scalable graphics API and discuss the architecture of MPK. We present MPK's graphics configuration interface, introduce the notion of compound-based decomposition schemes and describe our implementation. We present some results from our work on a couple of target system architectures and conclude with future directions of research in this area.

The fundamental drawback of current stereo and multi-view display techniques is the necessity to perform multi-pass rendering (one pass for each separate view) and subsequent image composition-and-masking for generating multi-stereo views. Thus the rendering time increases in general linearly with the number of rendered views. An increase in frame-rate can be achieved by sub-resolution rendering and extrapolation before display, at the expense of degraded image quality. To overcome these disadvantages, we introduce a new method for rendering volumes, points and triangles meshes on multi-view auto-stereoscopic displays. Our method exploits the programmability of modern graphic processing units (GPUs) for rendering multi-stereo views in a single rendering pass. The views are calculated directly on the GPU including sub-pixel wavelength selective views. We describe our algorithm precisely and provide details about its implementation. Experimental results demonstrate the performance adv...

2008

In this paper we present a new approach for interactive view-dependent rendering of large polygonal datasets, which relies on advanced features of modern graphics hardware. Our preprocessing algorithm starts by generating a simplified representation of the input mesh. It then builds a multiresolution hierarchy for the simplified model. For each face in the hierarchy, it generates and assigns a displacement map that resembles the original surface represented by that face. At runtime, the multiresolution hierarchy is used to select a coarse viewdependent level-of-detail representation, which is sent to the graphics hardware. The GPU then refines the coarse representation by replacing each face with a planar patch, which is elevated according to the assigned displacement map. Initial results show that our implementation achieves quality images at high rates.