compression.ru	in cooperation with		Lomonosov MSU Graphics & Media Lab (Video Group)	Video group head: Dr. Dmitriy Vatolin Project head: Dr. Dmitriy Kulikov Measurements, analysis: Dr. Mikhail Erofeev, Anastasia Antsiferova, Sergey Zvezdakov, Denis Kondranin
			Dubna State University
			Institute for Information Transmission Problems RAS

Welcome to 2018 HEVC video codecs comparison page! If you want to receive notifications about our reports, please

• 2018 report parts
• High Quality (AV1) Report results (Released!)
• 4K Report results
• Subjective Report results
• Main Report (objective) results
• Video codecs that were tested in 2018
• Objectives and testing rules
  • HEVC codec testing objectives
  • Test hardware characteristics
  • Video sequences selection
• Codec analysis and tuning for codec developers and codec users
• Thanks
• Contact information

High Quality (AV1) Report Comparison on FullHD video sequences with ultra slow/high quality presets Released on April, 4	7 codecs AV1, VP9, SIF, sz265, x265, sz264, x264	Free version Enterprise version You will receive all Enterprise Reports (High Quality, 4K, Subjective and FullHD) Alternative payment method
	5 FullHD video sequences
	Special Use Case 120 seconds per frame, encoding presets determined by codecs developers
	10 objective metrics YUV-SSIM, Y-SSIM, U-SSIM, V-SSIM, YUV-PSNR, Y-PSNR, U-PSNR, V-PSNR, Y-VMAF(v0.6.1), Y-VMAF(v0.6.1 Phone)
	HTML and PDF documents 150+ interactive charts and 48 pages
4K Report Comparison on high-resolution video sequences Released on January, 15	6 codecs HW265, sz265, Tencent Shannon Encoder, x265, sz264, x264	Free version Enterprise version You will receive all Enterprise Reports (High Quality, 4K, Subjective and FullHD) Alternative payment method
	10 4K video sequences
	Special Use Case 25 fps, encoding presets determined by codecs developers
	5 objective metrics YUV-SSIM, Y-SSIM, YUV-PSNR, Y-PSNR, Y-VMAF(v.0.6.1)
	HTML and PDF documents 535 interactive charts and 46 pages
Subjective Report Subjective comparison conducted on Subjectify.us platform Released on November, 19	10 codecs Kingsoft HEVC Encoder, SIF Encoder, sz264, sz265, Tencent Shannon Encoder, UC265, VITEC HEVC GEN2+ Encoder, VP9, x264, x265	Free version Enterprise version You will receive all Enterprise Reports (High Quality, 4K, Subjective and FullHD) Alternative payment method
	473 unique observers 22542 valid answers
	5 video sequences Short fragments from Crowd run, Ducks Take Off, Mountain bike, Playground, Red Kayak
	Ripping Use Case At least 1 FPS
	6 metrics Subjective score and 5 objective: YUV-SSIM, Y-SSIM, YUV-PSNR, Y-PSNR, Y-VMAF(v.0.6.1)
	HTML and PDF documents 118 interactive charts and 34 pages

Main Report (objective) Fast, Universal and Ripping use cases measured on FullHD videos Released on September, 4		Free version	Enterprise version
	Use cases	Universal (partially)	Fast, Universal, Ripping
	Per-sequence-results	2 of 28 sequences (only Universal use case)	All 28 sequences for all use cases (in interactive charts)
	Metric: YUV-SSIM
	Description of video sequences
	Codec info (developer, version number, website link)
	Other objective metrics (in addition to YUV-SSIM)		Y-VMAF, Y-SSIM, U-SSIM, V-SSIM, YUV-PSNR, Y-PSNR, U-PSNR, V-PSNR
	Per-frame metrics results	Only YUV-SSIM for 2 sequences, universal use case (14 charts)	All metrics for all sequences and use cases (5000+ charts)
	Relative quality analysis
	Download links for video sequences
	Encoders presets description
	PDF report	62 pages	119 pages
	HTML report	28 interactive charts	7000+ interactive charts
	Price	Free	$950
			You will receive all Enterprise Reports (High Quality, 4K, Subjective and FullHD) Alternative payment method

	Codec name	Use cases	HEVC	Hardware/GA
1	AV1 Alliance for Open Media	Ripping (in Ultra-ripping report)	(AV1)
2	HW265 Huawei Technologies Co., Ltd.	Fast, Universal, Ripping
3	Intel MFX (GA) Intel Corporation	Fast, Universal
4	Intel MFX (SW) Intel Corporation	Fast, Universal, Ripping
5	Kingsoft HEVC Encoder Kingsoft	Fast, Universal, Ripping
6	SIF encoder SIF Encoder Team	Ripping	(SIF)
7	sz264 Nanjing Yunyan	Fast, Universal, Ripping	(H.264)
8	sz265 Nanjing Yunyan	Fast, Universal, Ripping
9	Tencent Shannon Encoder Tencent	Fast, Universal, Ripping
10	UC265 Ucodec Inc.	Fast, Universal, Ripping
11	VITEC HEVC GEN2+ Vitec	Fast
12	VP9 The WebM Project (Google)	Ripping	(VP9)
13	x264 x264 Developer Team	Fast, Universal, Ripping	(H.264)
14	x265 MulticoreWare, Inc.	Fast, Universal, Ripping

The leaders of high quality comparison:

• First place: AV1
• Second place: VP9, x265 and sz265
• Third place: sz264 and x264

Below is a short summary with 3 of graphs from general report.

Rate-distortion results of the competitors at Fire sequence:

The leaders of comparison on 4K videos:

• First place: HW265 (Huawei) and Tencent Shannon Encoder
• Second place: sz265
• Third place: x265

Below is a short summary with 3 of 535 graphs from general report.

Rate-distortion results of the competitors at House Demolition sequence:

Below is a short summary with 3 of 118 graphs from general report.

According to the results of subjective comparison, the best codecs are the following:

• First place: Tencent Shannon Encoder and VITEC HEVC GEN2+ Encoder
• Second place: VP9 and x265
• Third place: Kingsoft HEVC Encoder

The winners determine from mean quality scores, but there is no absolute winner in the comparison, since different encoders take first place at different test video sequences: for example, on Crowd Run (short) three encoders show Pareto-optimal results: VITEC HEVC GEN2+ Encoder, x265 and VP9.

Tencent Shannon Encoder shows the best quality option for Red Kayak (short) sequence.

According to just quality scores (YUV-SSIM), the best codecs (among those we evaluated for all three use cases) are the following:

• First place: HW265
• Second place: Tencent Shannon Encoder
• Third place: Intel MSDK HEVC (SW) and Kingsoft HEVC Encoder

We tested three encoded use cases (see the description in section Test Hardware Characteristics). The universal-encoding use case has five Pareto optimal encoders in terms of mean speed and quality: UC265, Intel MSDK HEVC (SW), Intel MSDK HEVC (GA), Tencent Shannon Encoder and HW265. Nevertheless, the differences emerge for particular sequences and use cases.

Free report contains the results for two of 28 sequences, and results of all sequences and use cases are available in enterprise version.

Here is one of the RD-charts:

HEVC codec testing objectives

The main goal of this report is the presentation of a comparative evaluation of the quality of new HEVC codecs and codecs of other standards using objective measures of assessment. The comparison was done using settings provided by the developers of each codec. Nevertheless, we required all presets to satisfy minimum speed requirement on the particular use case. The main task of the comparison is to analyze different encoders for the task of transcoding video – e.g., compressing video for personal use.

Test Hardware Characteristics

• CPU: Intel Socket 1151 Core i7 8700K (Coffee Lake) (3.7Ghz, 6C12T, TDP 95W)
• Mainboard: ASRock Z370M Pro4
• RAM: Crucial CT16G4DFD824A 16GB DIMM DDR4 2400MHz CL15
• OS: Windows 10 x64

For this platform we considered three key use cases with different speed requirements:

• Fast/High Density – 1080@60fps
• Universal/Broadcast VQ – 1080p@25fps
• Ripping/Pristine VQ – 1080p@1fps and SSIM-RD curve better than x264-veryslow

See more on Call-for-codecs 2018 page

Video Sequences Selection

We have updated video database from which we choose sample videos for encoders' comparison. In this year, we analyzed 539765 videos hosted at Vimeo looking for 4K and FullHD videos with high bitrates (50 Mbps was selected as a lower bitrate boundary). This enabled us to find and download 942 new 4K videos and 2346 new FullHD videos.

We also completed list of selected sequences with high-quality videos from media.xiph.org.

This year test data set consists of 28 sequences: 5 from the old data set, 16 new ones from Vimeo and 7 from xiph.org. 25 sequences from the old data set were excluded. The average bitrate for all sequences in the final set is 449.72 Mbps, median – 192.02 Mbps. "Hera" (90 Mbps), "Television studio" (92 Mbps) and "Foggy beach" (93 Mbps) sequences have minimal bitrates. The complete list of sequences and description of selection process appears in PDF report.

Computer Graphics and Multimedia Laboratory of Moscow State University:

• 14+ years working in the area of video codec analysis and tuning using objective quality metrics and subjective comparisons.
• 27+ reports of video codec comparisons and analysis (H.265, H.264, AV1, VP9, MPEG-4, MPEG-2, decoders' error recovery).
• Methods and algorithms for codec comparison and analysis development, separate codec's features and codec's options analysis.

Strong and Weak Points of Your Codec

• Deep encoder parts analysis (ME, RC on GOP, mode decision, etc).
• Weak and strong points for your encoder and complete information about encoding quality on different content types.
• Encoding Quality improvement by the pre and post filtering (including technologies licensing).

Independent Codec Estimation Comparing to Other Codecs for Different Use-cases

• Comparative analysis of your encoder and other encoders.
• We have direct contact with many codec developers.
• You will know place of your encoder between other newest well-known encoders (compare encoding quality, speed, bitrate handling, etc.).

Encoder Features Implementation Optimality Analysis

Materials about MSU Codec Comparison

Other Materials