The Video Electronics Standards Association (VESA) announced today that it has made one of the largest sources of high-resolution, high dynamic range (HDR) image and video content available for unrestricted public use. More than 28,000 files (500-plus Gigabytes) of uncompressed HDR image and video content are now free to download from the VESA website. The content comes courtesy of a multiyear, multiphase research contract that VESA initiated with York University (Toronto, Canada) led by professors Robert Allison and Laurie Wilcox. The research encompassed the validation and improvement of the visually lossless performance of VESA’s Display Stream Compression (DSC) and VESA Display Compression-M (VDC-M)video compression codecs for stereoscopic 3D use cases, including for augmented/virtual/extended reality (AR/VR/XR) applications.
While the potential applications for the published material are nearly limitless, university students, faculty and researchers can leverage the content for visual science research projects, while companies can use the material for experimenting with or developing AR/VR/XR display technology and devices. The video and image files as well as instructions for downloading them can be found at: https://vesa.org/vesa-display-compression-codecs/#tab-video-and-image-files-download.
“Until now, there has been very little high-resolution HDR 3D content in uncompressed format available to the public or the research community,” stated Professor Allison, director of the Centre for Vision Research at York University. “We believe these images and image sequences will be useful for those working on image compression, image quality and other applications. VESA’s decision to make this content available to the public after commissioning York University to test and validate their own compression standards reflects their commitment and leadership in developing future display-related standards, and will benefit vision researchers and companies developing stereoscopic 3D products.”
Video compression has become increasingly popular in enabling improved performance of consumer display products by addressing bandwidth and power limitations associated with streaming video content at higher resolutions (4K and beyond), higher refresh rates and higher picture quality. However, for compression to be acceptable to the consumer, it must be visually lossless, meaning that it must be indistinguishable from uncompressed video. Over the past several years, VESA has supported research in assessing the visually lossless performance of video compression, which has played an important role in the development and refinement of VESA’s DSC and VDC-M standards.
“For the past several years, VESA’s display compression standards have helped hardware developers provide a high-quality visual experience and add new capabilities in monitors and televisions,” stated Greg Stewart, director of system architecture at Analogix Semiconductor and chair of the AR/VR Task Group at VESA. “This recent work between VESA and York University has fine-tuned VESA codecs even more. Now it’s possible to use these compression standards for the more demanding AR, VR and XR display applications without creating visible artifacts. These test images and videos were a critical part of this work, and enabled VESA member companies to perform their own tests and validate their own products. With these test images and videos now out in the public, this should benefit the entire display industry and ultimately the end-user experience.”
The first two phases of the York University research project, which were completed in November 2019, focused on the subjective validation of stereoscopic 3D image compression performance targeting AR/VR/XR head-mounted display (HMD) use cases. Areas investigated included to what extent compression artifacts, such as flicker, that are visible in a 2D image would still be visible in a stereoscopic 3D image.
The third phase of the project, completed in March 2020, extended this research by examining the impact of bypassing the color transform for images that have undergone chromatic aberration pre-correction. In stereoscopic HMDs, powerful lenses placed in front of the displays can cause significant image distortion. The traditional way of addressing this image distortion is by pre-distorting the image so that the image looks perfect after the lens distortion. However, this pre-distortion can disrupt the spatial correlation of the red, green and blue pixels in the displays, and thus impact compression performance and overall display quality. The third phase of the York University research examined whether adjusting the color processing in the codecs by turning off the color transform could improve the performance of DSC and VDC-M for these use cases. The research also examined how well VESA’s video compression codecs worked in managing chromatic aberration in 3D imaging in general.
The results of the York University research project indicated that DSC and VDC-M performed very well in testing during all three phases. The research also led to refinements in the codecs to further improve their performance. A fourth phase of the research project is currently underway to evaluate the performance of the compression codecs for motion sequences, including camera tracking of stationary objects and the rotation and movement of objects.
According to Bill Lempesis, executive director of VESA, “York University is a world-class institution dedicated to research training in visual science and its applications. This multi-year research study that VESA commissioned with them is an important undertaking to address the industry’s need for solutions to overcome the increasing bandwidth and power demands of mobile AR and VR devices. Through rigorous testing, this research has enabled us to verify the visually lossless performance of our DSC and VDC-M compression codecs – giving product developers and researchers the confidence that adopting VESA’s compression standards will give them high-quality results for their head mounted display products.”