Organisation internationale de normalisation

Regular Session 2: Compression Technology

4-1 Geert Van der Auwera and Yeong Taeg Kim

(Samsung Information Systems):
Triangular Sub-Macroblock Partitioning for Motion Compensated Prediction
4-2 Munchurl Kim (ICU), Changseob Park (KBS):
Beyond Macroblock based Predictive Coding
4-3 Kyohyuk Lee, Elena Alshina, Jeonghoon Park, Woojin Han
and Junghye Min (Samsung):
Technical considerations on new challenges in video coding standardization
4-4 Johannes Ballé, Steffen Kamp, Aleksandar Stojanovic, Mathias Wien
and Jens-Rainer Ohm (RWTH Aachen University):
Tools for Improving Texture and Motion Compression

Conclusions from Workshop (including high-level summary from various presentations):

1. 
   Next-generation Networks
   

Fixed: Open where the limits per user are … but consumption is expected to increase by a factor of 1.7 per year

Mobile: Limited data rate per user (bitrate achieved in practical services far below theoretical maximum), cost per bit transmitted will always be issue.

Complexity issue in battery operated devices

2. 
   Devices:
   

Displays: 4K x 2K will happen (2012?). Human/display relationship should be further investigated (will the viewing distance always be 3H or will it become closer?) Color, frame rate, bit-depth also increase.

Cameras: No presentation given, but 4Kx2K cameras exist. In general, quality of cameras (low noise) has largely increased recently.
Storage: Flash memory – further development no problem; compression is an issue a) because data transfer is slow b) because it has to compete in terms of capacity e.g. against optical storage. Packaged media is relevant application for FM.

3. 
   Network Protocol aspects: Impact of network QoS?
   

4. 
   Compression technology: Improvement over the gains reported in Hannover.
   

Main conclusion: Compression as much as possible is needed. The main result of the workshop is that the digital video services over networks are expected to steadily grow over time in terms of resolution, quality and volume (amount of consumption). As a consequence, the overall amount of video data rate that is to be transferred will grow at a very fast pace. Networks are already finding it difficult to carry HDTV resolution and data rates economically to the end user. Further increase in the resolution and the data rates will put further pressure on the networks. For example:

• 
  High-definition (HD) devices (displays and cameras) are affordable for consumer usage today, while the currently available internet and broadcast network capacity is not sufficient to transfer large amount of HD content economically. While this situation may change slowly over time, the next generation of ultra-HD (UHD) contents and devices, such as 4Kx2K displays for home cinema applications and digital cameras, are already appearing on the horizon.
  
• 
  For mobile terminals, video quality using resolutions such as QCIF at low frame rates and low bit rates today is largely unacceptable. While the overall data rate will increase with the evolution of 3D/LTE and 4G networks, also the number of users increases simultaneously with their quality demand. Anticipating that lightweight HD resolutions such as 720p or even beyond will be introduced in the mobile sector to provide similar perceptual quality as for the home applications, lack of sufficient data rates as well as the prices to be paid for transmission will remain a problem for the long term.
  

Analyzing this situation, video bitrate (when current compression technology is used) will go up faster than the network infrastructure will be able to carry economically, both for wireless and wired networks. MPEG has therefore concluded that next generation of video compression technology is needed that has clearly higher compression capability than the existing AVC standard in its best configuration, the High Profile. As a consequence, a study has been started on the feasibility of High-Performance Video Coding (HVC), which is mainly intended for high quality applications, in particular expecting

• 
  Performance improvements at higher resolution,
  
• 
  Applicability to entertainment-quality services such as HD mobile, home cinema and UHD TV.
  

The first steps towards HVC were made by producing the following three documents:

• 
  Vision and Requirements of HVC, starting from a document that had previously been presented on the AHG reflector
  
• 
  Call for test materials, which is urgently needed both for further evaluation of evidence of existing technical solutions, as well as for possible standards development (note: Currently available material is often captured with outdated generations of cameras; good test material for ultra-high resolutions does not exist). First responses are expected to the next meeting
  
• 
  Draft Call for Evidence with tentative timeline (depending on availability of test materials) for Call in February, responses by April.
  

Documents reviewed

m15796	Video coding in wired and wireless network applications Note: This document was presented immediately after the workshop session, and some of the conclusions stated about have been drawn from it, as far as it could seen as supplementing the workshop presentations. Video over IP with large market potential. Wired and wireless will converge in IP. In China, 94% of IP usage in 2010 will be video. Wired network access 1 Gb/s, next Gen 10 Gb/s. All-IP wireless download will be around 100 Mb/s. Current rates (necess. comp. ratios) EDGE 60 kbps(49 for QCIF), 3G 128 kbps(70 for QVGA), DSL 1.5 Mbbs(78). 3G still too expensive for users. QVGA still needs to be improved.DSL video is not yet competitive with DTV. Investigation on mobile display sizes: Could go from 480x320 up to 960x540 (QHD). For home, “theatrical presentation”, necessary view angle should go up to 90 degrees, which would result in 5400x2700 pixels to make pixels invisible. Error resilience is not sufficient in current video coding for IPTV applications. 4G: 960x540Q30p 1 Mbit/s VDSL2: 1920x1080@60p 8 Mbps. Low complexity for mobile is also important	Haoping Yu, Lianhuan Xiong
m15818	Comments on test conditions for high-quality coding experiments Propose QP from 20 … 32 (omit bad quality) Propose to use 4:4:4 mixed PSNR Offer 4Kx2K @ 30 “traffic” sequence and1080p @ 24 “cosme”	Shun-ichi Sekiguchi, Shuichi Yamagishi, Yoshihisa Yamada, Yoshiaki Kato, Kohtaro Asai, Tokumichi Murakami
m15864	Vision and Requirements for High-Performance Video Coding (HVC) Codec This document was discussed in the joint video and requirements meetings on HVC, and was used as a starting point in drafting N10175.	T.K. Tan
m15899	Technical considerations on New Challenges in Video Coding Standardization Note: This was presented in the workshop. Use adaptive interpolation and after-loop filters, larger block sizes, improved intra prediction (more directions). Also use initial implementation of decoder-side motion vector derivation by template matching. Total gain of 37 % on average for HD 1080p (50% for rolling tomatoes), approx. 25% for 720p for hierarchical B structure. Similar for IPPP structure with 1 reference frame.	Elena Alshina, Kyohyuk Lee, Woo-Jin Han, Jeonghoon Park, Tammy Lee

Output documents: