Přehled a porovnání standardu Essential Video Coding

This thesis provides a comprehensive overview and comparative analysis of the MPEG‑5 Essential Video Coding (EVC) standard. After tracing the historical evolution of video coding standards and overviewing the technical foundations and toolset of EVC, the work evaluates its encoding efficiency and computational complexity. Objective quality assessments were conducted on three representative test sequences (Sintel, Crowd Run, Park Joy) using VMAF, MS‑SSIM, and PSNR‑HVS‑M metrics. EVC’s performance — via its XEVE main and baseline profiles — was compared against other codecs including libaom AV1 (2019), SVT‑AV1 (preset 4), libvpx‑VP9, x265 (HEVC), and VVenC (VVC). The results demonstrated that EVC’s main profile provides competitive visual quality, particularly effective at low and medium bitrates, effectively positioning itself between royalty-free next-generation codecs and those hindered by patents. Additionally, the baseline profile offers a viable alternative characterized by significantly reduced computational demands, albeit accompanied by a compromise in visual quality. Computational complexity was quantified by analyzing encoding time and peak memory usage, with findings showing consistently low memory usage at all bitrates, low encoding time for baseline profile and high encoding time for main profile.