Search Results for "Sychev"
Found 98 document(s)
Search documents
Use number, keyword, author, or MPEG number. Filter by meeting when needed.
Not signed in yet.
Abstract
This contribution describes the simulation software submitted for AGH18 on ultra-low latency and packet loss resilience based on contribution JVET-AK0193. The purpose of this document is to give an introduction and help to be familiar with software. The document describes main modifications of encoder and decoder applications of VTM, entry point for network transmission emulation, and basic logic of main modules. Content of this document was mainly presented during AHG18 teleconference call on 2025-02-19. The software is available in https://vcgit.hhi.fraunhofer.de/jvet-ahg-ull/VVCSoftware_VTM/
Loading full meeting notes...
JVET-AL0200 AHG18: Simulation software description [S. Ikonin, V. Khamidullin, R. Shabaev, I. Gribushin, M. Sychev, E. Alshina (Huawei)]
This contribution describes the simulation software submitted for AGH18 on ultra-low latency and packet loss resilience based on contribution JVET-AK0193. The purpose of this document is to give an introduction and help to be familiar with software. The document describes main modifications of encoder and decoder applications of VTM, entry point for network transmission emulation, and basic logic of main modules. Content of this document was mainly presented during AHG18 teleconference call on 2025-02-19. The software is available in https://vcgit.hhi.fraunhofer.de/jvet-ahg-ull/VVCSoftware_VTM/
It was commented that the software provided in the link requires an update.
It was explained that the software is running encoder and decoder separately. The encoder app has a “jitter buffer” which gets input from the network simulator and simulates the feedback to be sent from the decoder if a packet loss would occur. The decoder gets the bitstream which comes with the packet losses and subsequent repair packets.
It was commented that trace files should not be used in optimizing concealment and packetization strategies.
Decisions
It was commented that trace files should not be used in optimizing concealment and packetization strategies.
Citation
Abstract
This contribution illustrates video transmission performance simulated using wireless Wi-Fi network simulator under low latency constraint. It is observed that average and worst case reconstructed quality, and frame freeze ratio strongly depend on the latency requirements specified in the test. With the transmission latency restricted to 16 ms, corresponding to one frame interval at 60 fps, the video freeze ratio is more than 20%, and average reconstruction quality is degraded by 5.0 dB (from 39.1 to 34.1 dB). Due to low latency restriction and wireless channel fluctuation decoder cannot be always guaranteed to receive all the packets necessary for frame reconstruction in time. It is suggested to consider packet loss resilient reconstruction as a necessary functionality to support low latency requirement.
Loading full meeting notes...
JVET-AK0193 Low latency Wi-Fi transmission simulation and suggestions on codec requirements [S. Ikonin, X. Ma, I. Gribushin, M. Sychev, V. Khamidullin, R. Shabaev, E. Alshina (Huawei)]
This contribution illustrates video transmission performance simulated using wireless Wi-Fi network simulator under low latency constraint. It is observed that average and worst case reconstructed quality, and frame freeze ratio strongly depend on the latency requirements specified in the test. With the transmission latency restricted to 16 ms, corresponding to one frame interval at 60 fps, the video freeze ratio is more than 20%, and average reconstruction quality is degraded by 5.0 dB (from 39.1 to 34.1 dB). Due to low latency restriction and wireless channel fluctuation decoder cannot be always guaranteed to receive all the packets necessary for frame reconstruction in time. It is suggested to consider packet loss resilient reconstruction as a necessary functionality to support low latency requirement.
Simulator uses a pattern for losses (using NS-3, similar concept as JVET-AK0057), but additionally allows generating losses based on latency. Encoding/decoding delay is currently not considered.
Error patterns of various network types could be used.
A video demo is included in the contribution.
At decoder, a lost slice is discarded. Encoder-side error concealment is used, changing reference picture of subsequent pictures based on feedback that a slice was lost. This would require the feedback channel to be reliable. Loss of I picture is not considered currently. Other approaches of error...
Decisions
Decision (SW): Create an AHG branch in VTM for the software of JVET-AK0057 and JVET-AK0193.
Citation
Abstract
This contribution identifies key Low Latency Controlled Complexity (LLCC) scenarios, namely cloud gaming, game casting, video surveillance, remote control of systems, video conferencing and summarize their characteristics in terms of video formats, requirements, constraints on coding tools. This contribution proposes baseline LLCC CTCs as follow:
Loading full meeting notes...
JVET-Y0043 AHG7: LLCC Scenarios and baseline configurations [G. Martin-Cocher, S. Puri, T. Poirier, K. Naser (InterDigital), J. Xu (Bytedance), D. Nicholson (Ektacom), M. Sychev (Huawei), L. Wang (Nokia), S. Liu, W. Yang (Tencent), J. M. Tiesse (VITEC), M. Karczewicz (Qualcomm)]
This contribution identifies what are reported to be key Low Latency Controlled Complexity (LLCC) scenarios, namely cloud gaming, game casting, video surveillance, remote control of systems, video conferencing and summarize their characteristics in terms of video formats, requirements, constraints on coding tools. This contribution proposes baseline LLCC CTCs as follow:
The VTM LLCC-1 configuration is defined as a modification of LD-P, with flat QP, GDR enabled with the GdrInterval set to 1second and a GdrPeriod = GdrInterval+ 4P frames. LMCS chroma scaling is turned OFF. QP42 is added and BD-rate calculations are proposed to be performed on QP22-37 and on QP 27-42.
For ECM, the LLCC-1 configuration is similar to the VTM LLCC-1 configuration, but instead of GDR, an IDR is set every second, until GDR will be implemented in ECM.
The VTM and ECM LLCC-2 configuration is defined as a modification of LD-B, with a maximum of 2 reference frames. QP42 is added and BD-rate calculations are proposed to be performed on QP22-37 and on QP 27-42. RPR is proposed to be enabled once per clip for class E.
The contribution further proposes baseline principals to refine these CTCs.
A reference software encoding run time to real time ratio (ET/RT ratio) principle for a hypothetical single threaded non optimised...
References:
Decisions
This was agreed to be further refined and discussed according to the aspects above. It was suggested to start from a baseline without GDR and RPR, LB with 2 reference frames, and investigate a set of enabled tools that provide a reasonable tradeoff regarding encoder complexity vs. compression.
Citation
Abstract
This contribution presents signalling of mapping sub-layers to picture rates in video coding. Two solutions are proposed: one is a signaling parameters in Supplemental Enhancement Information (SEI) another one using and extension of the HRD parameters for signaling. More specifically, the signaling of the table is proposed that allows mapping of Temporal IDs (TID) of sub-layers with corresponded picture rates.
Loading full meeting notes...
JVET-P0610 AHG8/AHG17: Sub-layer picture rates [M. Sychev (Huawei)]
References:
Cited:
PATENTS:
4022929
1.00
20220217389
1.00
WO/2021/061024
0.34
WO/2021/061025
0.31
3348058
0.30
20220295073
0.28
20250267290
0.27
WO/2020/185128
0.26
Decisions
JVET-P0610 AHG8/AHG17: Sub-layer picture rates [M. Sychev (Huawei)]
Citation
Abstract
Performance and algorithm proposed in JVET-P0170 is verified in this report. The proponent provided a simplied method for deriving the variable ZeroPos[ n ] that has negligible impact on BD-rate. This fact had been investigated and then confirmed.
Loading full meeting notes...
JVET-P0966 Crosscheck of JVET-P0170: CE7-related: on Simplification of coding transform coefficient level [M. Sosulnikov, M. Sychev (Huawei)]
References:
Decisions
JVET-P0966 Crosscheck of JVET-P0170: CE7-related: on Simplification of coding transform coefficient level [M. Sosulnikov, M. Sychev (Huawei)]
Citation
Abstract
This contribution presents derivation a portion of least significant bits of Picture Order Counter (POC) based on Temporal Layer ID in video coding in the Versatile Video Coding (VVC) and other video coding standards. The usage of Temporal ID value for derivation of the least significant bits of POC LSB when dyadic hierarchical picture order structure is used in temporal sub-streams. Two methods are proposed the first simple method allows to derive one LSB bit of POC, the second method provide more reduction of signaling of POC LSB value. One SPS flag “dyadic_temporal_nesting_flag” is proposed to support this derivation of POC LSB value basing on Temporal ID value. Additional conditional check in signalling of HRD parameter signaling is proposed based on the value of dyadic_temporal_nesting_flag flag.
Loading full meeting notes...
JVET-P0420 AHG8/AHG17: Reduction of overlapping between POC LSB and Temporal ID [M. Sychev (Huawei)]
References:
Cited:
Decisions
JVET-P0420 AHG8/AHG17: Reduction of overlapping between POC LSB and Temporal ID [M. Sychev (Huawei)]
Citation
Abstract
This contribution provides tool off simulations results for low delay configurations for Affine, ALF and DeBlocking Filter.
Loading full meeting notes...
JVET-P0622 AHG13: Low Delay results for Affine, ALF and DBF (Class A included) [M. Sychev (Huawei)] [late]
Coding studies on specific use cases (0)
Test material (0)
Conformance (6)
References:
Cited:
Decisions
Conformance (6)
Citation
Abstract not available in document
Loading full meeting notes...
JVET-O1134 Crosscheck of JVET-O0928 (CE7-related: Context reduction for entropy coding sig_coeff_flag) [M. Sychev (Huawei)]
References:
Decisions
JVET-O1134 Crosscheck of JVET-O0928 (CE7-related: Context reduction for entropy coding sig_coeff_flag) [M. Sychev (Huawei)]
Citation
Abstract
Luma mapping with chroma scaling (LMCS) was adopted into the VVC draft 4. The average value of predicted samples of the luma block is used to derive the scaling factor for chroma. Simplification of this process was encouraged at the 14th JVET meeting. Reference samples proposed to uses instead of predicted samples in JVET-N0477. Modifications proposed in this contribution consist of several aspects for Intra and Inter processing latency reduction jointly with simplification of the process. The first aspect of Intra simplification proposes to use luma DC prediction value instead of using an average value of luma samples for chroma scaling factor derivation. The second aspect about using shifted reference line with offset depended on directional intra prediction angle. Third aspect related to pipeline latency reduction in Inter processing by using reference samples corresponded to displacement with integer motion vector of current block.
Loading full meeting notes...
JVET-O0550 CE2-related: Further simplification of chroma residual scaling factors derivation [M. Sychev, E. Alshina (Huawei)]
Cited:
Decisions
JVET-O0550 CE2-related: Further simplification of chroma residual scaling factors derivation [M. Sychev, E. Alshina (Huawei)]
Citation
Abstract
This contribution proposes a method for reducing bit rate fluctuations for achieving low end-to-end delay with random access capabilities. The method is referred to as distributed decoding refresh (DDR) initially proposed in contribution JVET-N0116 [1]. The key idea is to distribute IDR pictures over multiple inter-coded pictures, and such an IDR picture therefore cannot be used by these inter-coded pictures, except for the last one, for inter prediction reference. Such an IDR picture is referred to as a distributed IDR (DIDR) picture. A DIDR picture is not for output. Rather, the inter-codded associated DDR picture, which is a different representation of the same source picture, is for output. It is asserted that the DDR-based approach can achieve rate smoothness while higher coding efficiency than GDR-based approaches, because no motion restrictions as in GDR are needed. Compared to usual IRAP picture based coding, the DDR-based approach needs to encode one additional inter-coded picture for each random access period, and the last of the multiple inter-coded pictures over which the VCL NAL units of the DIDR picture are distributed is less efficient than usual due to that it cannot use the DIDR picture for reference, while the DIDR would often be a number of pictures away in POC distance. The proposed detailed spec text changes are provided in an attachment. Simulation results for CE11-4 was obtained according CE11 specific test conditions [2]. Results reportedly show that the average luma BD-rate overhead is 3.5% in LDB configuration. The influence on encoding/decoding time is insignificant.
Loading full meeting notes...
JVET-O0116 CE11-4: Distributed decoding refresh DDR [M. Sychev (Huawei)]
Decisions
JVET-O0116 CE11-4: Distributed decoding refresh DDR [M. Sychev (Huawei)]
Citation
Abstract
Since this is an early draft, topics outside of the specific aspects that have been established by recorded meeting agreements are not included in the specification. Such aspects are to be determined by further development of the VVC project in JVET. The high-level syntax for the standard is yet to be developed. The aspects of high-level syntax in this early draft are provided only to show how certain features are likely to be controlled by some high-level syntax that may have a sequence level, a picture level, and a brick group level (a picture spatial region level that includes a subset of the CTUs of the picture).
Loading full meeting notes...
JVET-N0857 AHG12: Signalling for tile and brick partitioning [Y.-K. Wang, Hendry, M. Sychev (Huawei), R. Skupin, Y. Sanchez, K. Sühring, T. Schierl (HHI)] [late]
Discussed in coded picture regions BoG
Independently coded picture regions (19)
References:
JVET-K0052
JVET-K0063
JVET-K0072
JVET-K0103
JVET-K0122
JVET-K0184
JVET-K0190
JVET-K0230
JVET-K0232
JVET-K0237
JVET-K0251
JVET-K0307
JVET-K0310
JVET-K0315
JVET-K0325
JVET-K0337
JVET-K0346
JVET-K0351
JVET-K0357
JVET-K0367
JVET-K0369
JVET-K0371
JVET-K0500
JVET-K0529
JVET-K0554
JVET-K0556
JVET-K0564
JVET-K0565
JVET-L0045
JVET-L0047
JVET-L0053
JVET-L0054
JVET-L0055
JVET-L0059
JVET-L0064
JVET-L0074
JVET-L0081
JVET-L0082
JVET-L0083
JVET-L0090
JVET-L0100
JVET-L0104
JVET-L0118
JVET-L0124
JVET-L0136
JVET-L0142
JVET-L0147
JVET-L0158
JVET-L0165
JVET-L0191
JVET-L0194
JVET-L0198
JVET-L0209
JVET-L0217
JVET-L0231
JVET-L0248
JVET-L0249
JVET-L0256
JVET-L0265
JVET-L0266
JVET-L0271
JVET-L0272
JVET-L0274
JVET-L0279
JVET-L0283
JVET-L0285
JVET-L0293
JVET-L0340
JVET-L0361
JVET-L0362
JVET-L0366
JVET-L0369
JVET-L0410
JVET-L0414
JVET-L0428
JVET-L0449
JVET-L0468
JVET-L0553
JVET-L0628
JVET-L0632
JVET-L0646
JVET-L0664
JVET-L0678
JVET-L0686
JVET-L0694
JVET-L0696
JVET-M0063
JVET-M0064
JVET-M0092
JVET-M0101
JVET-M0102
JVET-M0111
JVET-M0113
JVET-M0118
JVET-M0119
JVET-M0126
JVET-M0128
JVET-M0132
JVET-M0140
JVET-M0142
JVET-M0145
JVET-M0147
JVET-M0160
JVET-M0166
JVET-M0170
JVET-M0171
JVET-M0173
JVET-M0192
JVET-M0193
JVET-M0201
JVET-M0238
JVET-M0246
JVET-M0251
JVET-M0255
JVET-M0264
JVET-M0273
JVET-M0281
JVET-M0297
JVET-M0300
JVET-M0303
JVET-M0328
JVET-M0361
JVET-M0381
JVET-M0407
JVET-M0415
JVET-M0421
JVET-M0427
JVET-M0436
JVET-M0444
JVET-M0446
JVET-M0451
JVET-M0453
JVET-M0464
JVET-M0470
JVET-M0479
JVET-M0483
JVET-M0487
JVET-M0497
JVET-M0502
JVET-M0512
JVET-M0685
JVET-M0853
JVET-M0883
JVET-M1001
Decisions
Independently coded picture regions (19)
Citation
Abstract
Since this is an early draft, topics outside of the specific aspects that have been established by recorded meeting agreements are not included in the specification. Such aspects are to be determined by further development of the VVC project in JVET. The high-level syntax for the standard is yet to be developed. The aspects of high-level syntax in this early draft are provided only to show how certain features are likely to be controlled by some high-level syntax that may have a sequence level, a picture level, and a tile group level (a picture spatial region level that includes a subset of the CTUs of the picture).
Loading full meeting notes...
JVET-N0116 AHG14/AHG17: DDR - distributed decoding refresh [Y.-K. Wang, M. Sychev, Hendry (Huawei)]
Track B 2040 Saturday (GJS)
This contribution proposes a method for reducing bit rate fluctuations for achieving low end-to-end delay with random access capabilities. The method is referred to as distributed decoding refresh (DDR). The key idea is to distribute IDR pictures over multiple inter-coded pictures, and such an IDR picture cannot be used by these inter-coded pictures, except for the last one, for inter prediction reference. Such an IDR picture is referred to as a distributed IDR (DIDR) picture. A DIDR picture is not for output. Rather, the inter-coded associated DDR picture, which is a different representation of the same source picture, is for output.
It is asserted that the DDR-based approach can achieve rate smoothness while higher coding efficiency than GDR-based approaches,...
References:
JVET-K0052
JVET-K0063
JVET-K0072
JVET-K0103
JVET-K0122
JVET-K0184
JVET-K0190
JVET-K0230
JVET-K0232
JVET-K0237
JVET-K0251
JVET-K0307
JVET-K0310
JVET-K0315
JVET-K0325
JVET-K0337
JVET-K0346
JVET-K0351
JVET-K0357
JVET-K0367
JVET-K0369
JVET-K0371
JVET-K0500
JVET-K0529
JVET-K0554
JVET-K0556
JVET-K0564
JVET-K0565
JVET-L0045
JVET-L0047
JVET-L0053
JVET-L0054
JVET-L0055
JVET-L0059
JVET-L0064
JVET-L0074
JVET-L0081
JVET-L0082
JVET-L0083
JVET-L0090
JVET-L0100
JVET-L0104
JVET-L0118
JVET-L0124
JVET-L0136
JVET-L0142
JVET-L0147
JVET-L0158
JVET-L0165
JVET-L0191
JVET-L0194
JVET-L0198
JVET-L0209
JVET-L0217
JVET-L0231
JVET-L0248
JVET-L0249
JVET-L0256
JVET-L0265
JVET-L0266
JVET-L0271
JVET-L0272
JVET-L0274
JVET-L0279
JVET-L0283
JVET-L0285
JVET-L0293
JVET-L0340
JVET-L0361
JVET-L0362
JVET-L0366
JVET-L0369
JVET-L0410
JVET-L0414
JVET-L0428
JVET-L0449
JVET-L0468
JVET-L0553
JVET-L0628
JVET-L0632
JVET-L0646
JVET-L0664
JVET-L0678
JVET-L0686
JVET-L0694
JVET-L0696
JVET-M0063
JVET-M0064
JVET-M0092
JVET-M0101
JVET-M0102
JVET-M0111
JVET-M0113
JVET-M0118
JVET-M0119
JVET-M0126
JVET-M0128
JVET-M0132
JVET-M0140
JVET-M0142
JVET-M0145
JVET-M0147
JVET-M0160
JVET-M0166
JVET-M0170
JVET-M0171
JVET-M0173
JVET-M0192
JVET-M0193
JVET-M0201
JVET-M0238
JVET-M0246
JVET-M0251
JVET-M0255
JVET-M0264
JVET-M0273
JVET-M0281
JVET-M0297
JVET-M0300
JVET-M0303
JVET-M0328
JVET-M0361
JVET-M0381
JVET-M0407
JVET-M0415
JVET-M0421
JVET-M0436
JVET-M0444
JVET-M0446
JVET-M0451
JVET-M0453
JVET-M0464
JVET-M0470
JVET-M0479
JVET-M0483
JVET-M0487
JVET-M0497
JVET-M0502
JVET-M0512
JVET-M0685
JVET-M0853
JVET-M0883
JVET-M1001
Cited:
Decisions
It was commented that if there is camera motion, the efficiency of referencing this distributed picture may not be very good.
Citation
Abstract
This contribution is a resubmission of JVET-M0129. It is asserted that, if the group adopts a sub-picture-based video coding design, e.g., JVET-N0107, that allows different sub-pictures to have different tile configurations, then there would be no need to additionally have a flexible tiling scheme that allows to partition a picture into tiles in a manner that is more flexible than the current tiling scheme, where tiles are always in tile rows and tile columns.
Loading full meeting notes...
JVET-N0111 AHG12: Flexible tiling [Y.-K. Wang, Hendry, M. Sychev (Huawei)]
See the notes for JVET-N0066.
References:
Cited:
Decisions
See the notes for JVET-N0066.
Citation
Abstract
This contribution provides analysis on encoder-decoder delays for access unit (AU) based and decoding unit (DU) based encoder-decoder operations for bitstream coded with intra random access point (IRAP) pictures at regular interval and for bitstream coded with gradual decoding refresh (GDR) picture at regular interval. The objective of this analysis is to understand whether GDR feature can help reduce encoder-decoder end-to-end delay for both AU and DU based operations when compared to using IRAP pictures.
Loading full meeting notes...
JVET-N0114 AHG14: A delay analysis for IRAP and GDR [Hendry, Y.-K. Wang, M. Sychev (Huawei)]
Track B 1815 Saturday (GJS)
This is an information contribution, not a proposal.
This contribution provides analysis on encoder-decoder delays for access unit (AU) based and decoding unit (DU) based encoder-decoder operations for bitstream coded with intra random access point (IRAP) pictures at regular interval and for bitstream coded with gradual decoding refresh (GDR) picture at regular interval. The objective of this analysis is to understand whether GDR feature can help reduce encoder-decoder end-to-end delay for both AU and DU based operations when compared to using IRAP pictures.
The finding from the analysis is that a GDR feature can help reduce encoder-decoder end-to-end delay for both AU and DU based operations.
Details of the analysis are provided in the attachment.
It was suggested tha...
Cited:
Decisions
1035.86
Citation
Abstract
In this contribution, new filter coefficients proposed to improve performance of Intra interpolation and Inter Chroma prediction. Usage of proposed Smooth 4-tap interpolation filter instead of Gaussian filter for Intra prediction and Sharp 4-tap interpolation filter instead DCT-IF filter for both Intra prediction and for Chroma motion compensation demonstrates the following bit-rate savings: -0.03% (Y), -0.07% (U), -0.04% (V) for AI configurations and -0.03% (Y), -0.07% (U), -0.18% (V) for RA configuration, for Low Delay configurations Chroma gain is even better: 0% (Y), -0.14% (U), -0.42% (V) for LDB and -0.04% (Y), 0.01% (U), -0.49% (V) for LDP with up to -3.4% in Chroma of gain for some sequences. Reducing number of phases for Intra prediction from 32 to 16 leads to following results: -0.01% (Y), -0.04% (U), -0.04% (V) for AI configurations and -0.02% (Y), -0.10% (U), -0.19% (V) for RA configuration. There are no influence on encoding decoding time for both cases. There are no SW modification required apart from filter coefficients replacement.
Loading full meeting notes...
JVET-M0424 CE10-related: On enhancement of 4-tap interpolation filters [M. Sychev, J. Chen (Huawei)] [late]
Initial upload rejected as a placeholder.
References:
Cited:
Decisions
rejected
Initial upload rejected as a placeholder.
Citation
Abstract
This contribution proposes a method for signalling of tile partitioning where tile partitioning can be signalled based on the numbers of rows and columns or based on the width and height of the tiles. It is asserted that such flexibility provides bit saving for signalling of tile partitioning information.
Loading full meeting notes...
JVET-M0137 AHG12: On tile configuration signalling [M. Sychev, Hendry, Y.-K. Wang (Huawei)]
References:
Cited:
Decisions
JVET-M0137 AHG12: On tile configuration signalling [M. Sychev, Hendry, Y.-K. Wang (Huawei)]
Citation