TECHNICAL
REFERENCES 1 As shown in Table 1.
2 https://netflixtechblog.com/per-title-encode-optimization-7e99442b62a2, 12/14/2015 (per title)
3 https://netflixtechblog.com/more-efficient-mobile-encodes-for-netflix-downloads-625d7b082909, 12/1/2016 (per chunk)
4 https://netflixtechblog.com/dynamic-optimizer-a-perceptual-video-encoding-optimization-framework-e19f1e3a277f, 3/5/2018 (per shot)
5 https://netflixtechblog.com/toward-a-practical-perceptual-video-quality-metric-653f208b9652, 6/1/2016 (VMAF)
6 Ozer, Jan. “Report: Cloud-based Per-Title H.264 Encoding Benchmark Report” https://streaminglearningcenter.com/encoding/slc-releases- cloud-based-hevc-and-h-264-per- title-quality-and-opex-reports.html 6/6/2022
7 https://www.visualon.com/index.php/press/visualon-introduces-first-universal-content-adaptive-encoding-solution-for-video-streaming/
8 https://www.visualon.com/index.php/press/visualon-wins-2024-nab-show-product-of-the-year-award/
9 Testing performed by VisualOn on 7/25/2023. Source video: http://ftp.nluug.nl/pub/graphics/blender/demo/movies/ToS/ToS-4k-1920.mov. FFmpeg command line for x264: ffmpeg -i input -b:v 7.8M -maxrate 7.8M -bufsize 15.6M output.mp4, and for x264 with Optimizer: ffmpeg -vo_optimizer -i input -vo_vmaf 98 output.mp4. Full data available at https://www.visualon.com/wp-content/uploads/2024/09/side-by-side- comparison.pdf. 10 Testing performed by VisualOn on 9/05/2023. Source video: http://ftp.nluug.nl/pub/graphics/blender/demo/movies/ToS/ToS-4k-1920.mov. FFmpeg command line for x264: ffmpeg -i input -b:v 2.3M -maxrate 2.3M -bufsize 4.6M output.mp4, and for x264 with Optimizer: ffmpeg -vo_optimizer -i input output.mp4. Full data available at https://www.visua- lon.com/wp-content/uploads/2024/09/side-by-side-comparison. pdf. 11 Based on testing performance by VisualOn on 7/30/2024. System configuration details: Dual socket Intel Xeon Scalable 8480+ (56 cores per socket), 192 GB system memory (16x32GB DDR5 4800), 500GB and 750 GB SSDs (Intel SSDSC2KB48 and Intel SSDSC2BB80), CPU microcode ver. 0x2b0005c0, Intel Turbo Boost Enabled (up to 3.80 GHz.), Hyperthreading enabled, Ubuntu ver 22.04.1 LTS with 5.17.0-107-generic kernel patches. Configuration details and full data available at https://www.visualon.com/wp-content/ uploads/2024/09/ Optimizer-benchmarks.pdf, configuration in the “H.264 Test Setup”, results in the “H.264 top rung comparison” tab, with Optimizer’s results extracted from the “x264 results” tab.Results from other encoders are from the report referenced in footnote 6. 12 Tested by VisualOn on 8/2/2024. System configuration details: Dual socket Intel Xeon Scalable 8480+ (56 cores per socket) with Intel Datacenter Flex 140 GPU, 192 GB system memory (16x32GB DDR5 4800), 500GB and 750 GB SSDs (Intel SSDSC2KB48 and Intel SSDSC2BB80), CPU microcode ver. 0x2b0005c0, Intel Turbo Boost Enabled (up to 3.80 GHz.), Hyperthreading enabled, Intel Data Center GPU Flex 140, Ubuntu ver 22.04.1 LTS with 5.17.0-107-generic kernel patches. For full data details, see https://www.visualon.com/ wp- content/uploads/2024/09/Performance-on-Intel-SDP.pdf.
VisualOn provides an innovative, AI- powered solution that reduces bitrates up to 40 percent on average, while improving overall viewer experience. VisualOn Optimizer suite is encoder-agnostic and seamlessly integrates into encoding workflows, irrespective of the encoder being used or the infrastructure it runs on—CPUs or GPUs. The Optimizer easily integrates with Intel Quick Sync Video, dramatically reducing bitrate and allowing more encoding sessions on the same platform. With the variety of infrastructures used across media provider data centers, at the edge, and in the cloud, VisualOn Optimizer combined with Intel Xeon Scalable processors and Intel Data Center GPU Flex Series GPUs can help reduce cost of operations and storage of encoded content through lower bitrate, while enhancing visual quality.
www.visualon.com
DECEMBER 2024 Volume 46 No.4
95
Made with FlippingBook - Online magazine maker