1 / 47

Hybrid Bit-Stream Models (January 24-26, 2009 San Jose)

Hybrid Bit-Stream Models (January 24-26, 2009 San Jose). Issues to be determined. Are we going to use numerical values attached to each category? Decision:. Issues to be determined. Are we going to mix 10s and 16-24s SRC/PVS? Decision:. Issues to be determined.

Download Presentation

Hybrid Bit-Stream Models (January 24-26, 2009 San Jose)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Hybrid Bit-Stream Models (January 24-26, 2009 San Jose)

  2. Issues to be determined • Are we going to use numerical values attached to each category? • Decision:

  3. Issues to be determined • Are we going to mix 10s and 16-24s SRC/PVS? • Decision:

  4. Issues to be determined • Evaluators should be seated facing the center of the video display at the specified viewing distance. That means that subject's eyes should be positioned opposite to the video display's center (i.e. centered both vertically and horizontally) • Decision: Will be copied from the HD test plan.

  5. Issues to be determined • Are we going to include audio? • Decision:

  6. Issues to be determined • Display Monitors • QVGA/VGA: LCD • SD: CRT • HD: CRT/PDP/LCD • Decision:

  7. Issues to be determined • Are we going to use the same LCD panel if LCD display is used? • List of admissible LCD monitors? • Decision:

  8. Issues to be determined • For the HD/SD testing, the video will use the full screen dimensions and no background panel or black border will be present.Upscale SD to whole screen solution? Does this mean the same screen will be used for HD and SD testing? • Decision:

  9. Issues to be determined • How long will be the session length? • How many PVSs are there per session? • Decision:

  10. Issues to be determined • Randomization: maximum number of viewers per order? • Decision:

  11. Issues to be determined • Test design? Full matrix? • Decision:

  12. Issues to be determined • SRC admissibility? (e.g., at least 4 in ACR scale or ILGs will decide?) • Decision:

  13. Issues to be determined • SRC length • Option #1: extra 2 seconds or 20 frames • Option #2 (see test plan) • Decision:

  14. Issues to be determined • Cropping & Rescaling (help from Margaret?) • Decision:

  15. Issues to be determined • Test materials • MM sources • RRNR sources • NTIA • SVC • Other materials? • Decision:

  16. Issues to be determined • Codec • H.264 • MPEG2 • MPEG4 • HRC (bitrate, frame rate) • QVGA/VGA: same as MM • SD: same as RRNR • HD: same as HD • Decision:

  17. Issues to be determined • Max. length of freezing or skipping? • 40% of the total length? • Decision:

  18. Issues to be determined • Are we going to use the 11 scale ACR (the same decision as the HD project)? • Decision:

  19. Issues to be determined • Are we going to allow proponents to conduct subjective tests? • Decision:

  20. Issues to be determined • How about model entry (e.g., 1. per codec, per resolution)? • Decision:

  21. Issues to be determined • Currently, there codecs are considered (H.264, MPEG2, MPEG4). • Should a model handle all the three codecs at all resolutions (QCIF/QVGA & SD/HD)? • Or a model can be developed to handle just one codec at one resolution (QCIF/QVGA or SD/HD)? • Decision:

  22. Issues to be determined • Are we going to use numerical values attached to each category? • Are we going to use the 11 scale ACR (the same decision as the HD project)? • Are we going to mix 10s and 16-24s SRC/PVS? • For HD, 1080i footage may be de-interlaced and then used as SRC in a 1080p experiment? – Will be copied from HDTV • SRC length? (e.g., extra 2 seconds or 20 frames) • How long will be the session length? • How many PVSs are there per session? • Test design? Full matrix? • Randomization: maximum number of viewers per order? • Are we going to allow proponents to conduct subjective tests? • How about model entry (e.g., 1. per codec, per resolution)? • SRC admissibility? (e.g., at least 4 in ACR scale or ILGs will decide?) • Viewer position? • Are we going to include audio? • Are we going to use the same LCD panel for QVGA/VGA? • Max. length of freezing? • Decision:

  23. Decision from the last meeting (Sep. 2008, Ghent)

  24. Test conditions • PVS length: SD/HD => 16 • Rebuffering is NOT allowed • PVS length: QCIF/QVGA • Variable for rebuffering: 16(SRC)/upto24(PVS) • Fixed: 10 sec • Rebuffering is allowed • Definition: Rebuffering is freezing longer than 0.5s without skipping • Maximum delay/advancein PVS? (+/- 0.25 sec): other conditions will be identical as in the multimedia project [TBD] • Decision: the maximum time limit for freezing or rebuffering is 8 seconds forQCIF/QVGA.

  25. Subjective Test • ACR HR (tentatively) for both QVGA/VGA and SD/HD • Graphical, continuous ACR version may be also used. • Numbers will be used along with semantic terms unless P.910rev is approved. • Proposal. • (1) Vittorio: Some concerns about ACR-HR, new subjective testing method. => prefer DSCQS, DSIS, at least 10 steps • (2) Silvio: Propose to use a scale from 1 to 5 with steps of 1/10, Propose to use MOS (or MOS_adj) as a target value • Decision: 11 scale ACR-HR(P.910)=10, DSCQS=1

  26. Model Inputs • Model A (P.NAMS): TS, RTP, UDP, IP headers & (optional) longer freezing Model B (P.NBAMS): TS, RTP, UDP, IP headers, ES information & (optional) longer freezing • Hybrid models (FR/RR/NR): bit stream data (TS, RTP, UDP, IP headers, ES information) & PVS (SRC if required)

  27. Proponents & Preferred input types (1) Hybrid models

  28. Proponents & Preferred input types (2)P.NAMS: Use only IP, UDP, RTP and TS headers Note: Potential proponents should declare their intensions to SG12 by Sep. 30.

  29. Proponents & Preferred input types (3)P.NBAMS

  30. ILGs • CRC • INTEL • Acreo • IRCCyN • Nortel • FUB • NTIA

  31. Reference Decoder [TBD] • Reference decoder to ensure error-free bit stream compliance to the standard (open source) • Open-source H.264 codec (JM) • Alternative: x264 • Open-source MPEG2 (FFMPEG) • Open-source MPEG4 - • momusys, Xvid?? NOTE: Proponents should provide a working system of entire chains (encoder, container, server, capture program, decoder, etc)

  32. Capturing bit stream data [TBD] • File format: PCAP, RTP dump, MS Network Analyzer format (.cap). • Capturing programs: Wireshark, Tcpdump, MS Network Analyzer format • Protocol: TCP/IP, UDP??

  33. Streaming Server [TBD] • Representative streaming server • Live555: working for MPEG2 • VLC: working for H.264 and MPEG4 • Dvbstream: working for ??? • Helix: working for ??? • Ffserver: MPEG2 and MPEG4, H.264(?)

  34. Inputs to the model • Text file (Hybrid) • Bit stream data (Bitstream/Hybrid)

  35. Inputs to the model

  36. Issues with Inputs to the model [TBD] • If bit-stream data which contains transmission errors can’t be interpreted by some models, what will we do about the models?

  37. IP analyzer: converter & decoder • We need to convert the captured bit stream data so that the reference decoder can decode it. • programs??? (TCP replay, off-line simulator) • Free or open-source SW?? • Hire professional developer??

  38. IP analyzer • Any bit stream data which can be understood by the reference IP analyzer will be admissible. Models are required to handle the bit stream data. • Bottom Line: We need working programs for reference encoder, container, server, packet capturer, extractor and reference decoder so that the reference decoder can decode the captured bit stream data. Also the transmission error information should be outputted.

  39. Data Analysis • Same as the multimedia project –tentatively agreed.

  40. Target usage (PLR): TBD • SD/HD: (information from Broadband forum may be useful) • QCIF/VGA - internet: - mobile: 0-15%

  41. Drafting test plan • A draft of the hybrid test plan will be done by the co-chairs (JB, CL) and editors (David Hands, Nicolas Staelens, Vittorio Baroncini, Yves Dhondt) by borrowing heavily from the multimedia and HD test plans. • The draft will be presented to VQEG for approval.

  42. Schedule • Proposal for a working system (encoder, server, capturing program, decoder): next VQEG meeting (Jan, 2009) • Finalization of test plan: VQEG meeting after the reference IP analyzer is available: VQEG meeting (July, 2009) • Model submission: Six months after the test plan is finalized.

  43. Remaining issues [TBD] • Subjective test • Multimedia project approach: proven, but delays may happen (about 6 month delay after model submission). Easy to correct errors • HDTV project approach: Fast, but there can be impasse if serious PVS problems occur. Difficult to correct errors.

  44. Number of subjective tests (Proponents) BT 1 Ericson1 DT 1 Ghent Univ. 1 KDDI 2 Lancaster Univ 1 NEC 1 NTT 1Opticom 1 Psytechnics 1 Symmetricom 1 SQ 1 Tektronix Yonsei 3 Total: 16

  45. Number of subjective tests (ILG) • CRC 0 • INTEL 0 • Acreo 0 or possibly 1 • IRCCyN 1 • Nortel • FUB 1 • NTIA 1? • NOTE: ATIS option may be considered.

  46. Remaining issues [TBD] • Calibration limit: currently same as HDTV & Multimedia (will be decided over the reflector)

  47. Remaining issues [TBD] • Bit-stream models • P.NAMS (audio/video, video only) • P.NBAMS (audio/video, video only) • P.NBAMS with some additional information from decoder such as freeze frames (audio/video, video only) • audio/video(P.NAMS) vs. Hybrid (video only) • Decision: Initially video only. If enough audio(with video) subjective data is available, models for audio and audio/video will be also validated.

More Related