Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Technology Update Better Pixels

3,578 views

Published on

Presentatie van: Maryline Clare (Orange Labs)
Jérôme Vieron (ATEME)
Event: Technology Update Beter Pixels
Datum: 3 februari 2016

Voor meer informatie: http://www.immovator.nl/agenda/technology-update-better-pixels-effect-image-quality-immersiveness

Published in: Technology
  • Be the first to comment

Technology Update Better Pixels

  1. 1. Why and how to go to Ultra HD phase 2 Maryline CLARE, Orange Labs Jérôme VIERON, ATEME For Enhanced Video ExpeRience
  2. 2. Summary • 4EVER-2: description and methods • UHD-TV phase 2: the real gap in quality of experience • Focus on HDR – what is HDR – several technical solutions and standardization committees – subjective tests – 4EVER-2 trials www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 2 • HFR: – what it is – subjective tests • compressor challenges – HDR – HFR • The first UHD-TV phase 2 + NextGen audio production = 4EVER short • Conclusions, perspectives
  3. 3. What is 4EVER-2? What do we work on? How? www.4ever-2.com 34EVER-2 : go for UHD-TV phase 2!
  4. 4. 4EVER: All about visual quality of TV content www.4ever-2.com 4 spatial definition (4K) Quality of Experience? UHD-TV phase 2 UHD-TV phase 1 Extended colors (Wide Color Gamut, WCG) Temporal definition (High Frame Rate, HFR) Contrast definition (High Dynamic Range, HDR) 4EVER-2 : go for UHD-TV phase 2!
  5. 5. 4EVER-2 structural keypoints • Consortium and its 9 partners known (4EVER): active contribution to standardization bodies, TV professional services, … • Complementary skills over the whole end-to-end audiovisual chain • Lab research experimented in real premium events (sports, culture) • Integration of the audio dimension for a real audiovisual quality content evaluation • Collaboration with worldwide major actors – Groups and forums : EBU, BBC, IRT, HD Forum – Consumer and professional equipment manufacturers www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 5
  6. 6. Nine French partners 4EVER-2 : go for UHD-TV phase 2! 6www.4ever-2.com Shooting Production Encoding Contribution Distribution Decoding Playing Perceived quality on end-to-end audiovisual chain
  7. 7. 4EVER-2: the end-to-end AV chain www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 7 • Orange Labs: coordination, audiovisual assessment tests • francetélévisions: content production, joined audio visual subjective tests • AMPVisualTV: live production workflows , UHD-TV equipment evaluation, • ATEME: HEVC encoding of UHD-TV phase 2 (HDR and/or HFR and/or WCG) • INSA-IETR: HEVC decoding of UHD-TV phase 2 (OpenHEVC open source) • HTS: colorimetry science • Télécom ParisTech: HEVC player (GPAC open source), HDR quality assessement, rendering HDR depending on viewing room conditions • TeamCast: terrestrial and satellite TV transmission (DVB-T2, DVB-S2X) • GlobeCast: technical infrastructure for end-to-end content distribution (satellite, IPTV)
  8. 8. 1/Experimental shooting sessions • Evaluate production equipments and complete workflow • Generate ultra high quality sequences with controled parameters www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 8
  9. 9. 1/Sequences shared worldwide www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 9
  10. 10. 2/Evaluation of video perceived quality • Display parameters: brightness, contrast • In the room: light control around the display • Distance to the screen • SAMVIQ method (ITU-R BT 1788): adapted to evaluation of video quality on short sequences • 25 to 20 non expert observers • SEOVQ software Subjective tests in standardized environment (REC ITU-R BT.500) 4EVER-2 : go for UHD-TV phase 2! 10www.4ever-2.com
  11. 11. 3/Research and software development • Colorimetry: science of color • Research and development – HEVC encoding • Professional prototypes for encoder: ATEME – HEVC decoder • Academic research and open source for decoder : INSA-IETR (open HEVC) – Video player: Télécom ParisTech (GPAC) • Academic research and open source to integrate HEVC decoding and UHD-TV specificities www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 11
  12. 12. 4/Experiment on visible events www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 12 Go on actual premium events to experiment prototype audiovisual end-to-end chain Stressful, but extremely useful!
  13. 13. 5/Disseminate results and conclusions 4EVER-2 : go for UHD-TV phase 2! 13 September 2015 On EBU booth, demonstrating HDR in collaboration with BBC HFR in collaboration with EBU, BBC, IRT, RAI www.4ever-2.com
  14. 14. 5/Disseminate results and conclusions www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 14 April 2014 UHD-TV in OTT internet HEVC + 4K + DASH + HTML5 for a smooth adaptation of bitrate (hence quality) to bandwidth variations UHD-TV for DTT HEVC + 4K + DVB-T2
  15. 15. 5/Disseminate results and conclusions 4EVER-2 : go for UHD-TV phase 2! 15 September 2014 back then, the first and only booth to demonstrate all 4 technologies of UHD-TV www.4ever-2.com
  16. 16. 5/Disseminate results and conclusions • Major communication events in 2016 – MIPTV 2016, Cannes France, April – NAB 2016, selected for Future Labs Park • Demonstrations : HFR, HDR, ATSC, 4K, HEVC, live, … • Please visit  – IBC 2016? • If interesting demonstrations to propose • If selected for Future Zone… www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 16
  17. 17. Ultra HD-TV phase 2 What for? www.4ever-2.com 174EVER-2 : go for UHD-TV phase 2!
  18. 18. Why a need for WCG, HDR and HFR? 4K not « better enough » in normal TV viewing conditions www.4ever-2.com 18 but only when sitting at 1,5*height of the display SD HD UHD Very good Good Fair Bad Very Bad JVC Pro Vérité 84" - Toshiba 55" (1.5H) 55" 84" SD HD UHD Very good Good Fair Bad Very Bad JVC Pro Vérité 84" - Toshiba 55" (3H) 55" 84" 1,5m 1m UHD 1m 3m HD 10 points better on a scale of 100 4EVER-2 : go for UHD-TV phase 2!
  19. 19. HFR: always better even at same bitrate www.4ever-2.com 19 100Hz native contents provide better quality than 50Hz native contents (about 15 points) 4EVER-2 : go for UHD-TV phase 2! work done in collaboration with • 100Hz native: much better perceived video quality than 50Hz native ones • The performance of Interpolation (MCFI) depends on the content’s motion features but in average, 50Hz interpolated at 100Hz cannot reach the perceived video quality of native 100Hz At the same bitrate, the perceived quality of a 100Hz source remains better than a 50Hz source (interpolated or not)
  20. 20. HDR: a real benefit expected Actual improvement potential if conditions are respected 4ever subjective tests results - 2014 www.4ever-2.com 20 Maximum of comfort obtained between 750 and 1500 cd/m² Perceived video quality increases with peak luminance level (20 points between 300 and 1500 cd/m2) 4EVER-2 : go for UHD-TV phase 2!
  21. 21. HDR: a good improvement… soon? High Dynamic Range: sequences look more natural, sensation of immersion – Better contrasts: better dark areas, better bright areas www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 21
  22. 22. Focus on HDR www.4ever-2.com 224EVER-2 : go for UHD-TV phase 2!
  23. 23. 23 www.4ever-2.com High Dynamic Range (HDR) • Increased contrast • Bright areas (e.g. > 1000 nits) , Dark areas (e.g. < 0.001 nits) • Luminance range rendering closer to the real world luminance range • HDR vs SDR higher gain than UHD vs HD • Perceived video quality increases with peak luminance level High Dynamic RangeStandard Dynamic Range Simulation 1 nit = 1 candela/m2 4EVER-2 : go for UHD-TV phase 2!
  24. 24. 24 www.4ever-2.com Today’s Standard Dynamic Range (SDR) 24 ~ 0,2 – 1 Lux > 30 000 OETF: Opto-Electrical Transfer Function EOTF: Electro-Optical Transfer Function Lum in cd/m2 or nits 8 bits signal 4EVER-2 : go for UHD-TV phase 2!
  25. 25. 25 www.4ever-2.com OETF, EOTF & OOTF Scene Luminance (min. 10-6 nits / max. 108 nits) Display Luminance (min. 10-3 nits / max. > 103 nits) Electro-Optical Transfer Function (EOTF) Opto-Electronical Transfer Function (OETF) Opto-Optical Transfer Function (OOTF) 4EVER-2 : go for UHD-TV phase 2!
  26. 26. 26 www.4ever-2.com HDR: “New” Opto Electronic Transfer Function (OETF) • “Old” gamma correction => Not well-adapted to wider dynamic range • PQ (Perceptual Quantizer - SMPTE ST 2084) => Luminance vs Code Value (display referred) 4EVER-2 : go for UHD-TV phase 2!
  27. 27. 27 www.4ever-2.com HDR: Grading and Settings • Color grading (tuning RGB levels, Hue, Brightness, Saturation, and Contrast) => Artistic intent • HDR content color grading relies on a set of parameters • Which OETF? ST-2084, HLG or proprietary Transfer Function • Color gamut or primaries? BT 709, P3, BT 2020, XYZ,… • White primaries? C, D65,… • Min brightness value? • Max brightness value? • Grading monitor characteristics? To make the viewing experience match the original “Artistic intent”, HDR color grading information must be known at the display side: metadata Sources: BlackMagic Davinci Resolve 4EVER-2 : go for UHD-TV phase 2!
  28. 28. 28 www.4ever-2.com HDR: what about backward compatibility? • HDR backward compatibility with SDR is an hot topic! • Several meanings! • Same workflows? cameras, encoders, post-production, etc… • Same spatial resolution? Same bitdepth? • What is the real need? • One HDR signal directly watchable on any SDR display • One signal from which we can extract both HDR and SDR signal • One HDR signal and one SDR signal simulcasted • 2 answers: • Compatibility with UHD phase 1 • Compatibility with “today” HDTV (HD,709) 4EVER-2 : go for UHD-TV phase 2!
  29. 29. 29 www.4ever-2.com HDR in a nutshell • New Transfer Function • To better match human visual perception at high luminance • HDR video signal grading • Color Gamut/primaries and grading based on OETF/EOTF • Backward compatibility • Metadata? • Artistic intent: to properly render the HDR signal whatever the technical characteristics of your HDR display • Backward compatibility: to reconstruct HDR and/or SDR signal from the encoded content 4EVER-2 : go for UHD-TV phase 2!
  30. 30. 30 www.4ever-2.com Alternative HDR Solutions • In addition to HDR10/PQ: Philips, Technicolor, Dolby and, NHK/BBC alternative solutions • OETF: New transfer functions • Single or Dual layer • Metadata (static, dynamic) • SDR (Standard Dynamic Range) backward compatibility Provider OETF Method Metadata Backward Compatible Layers 1 BDA/HDR 10 PQ HDR (SL) Static No Single 2 BBC/NHK HLG OETF (SL) No Yes Single 3 Philips/Technicolor -1 PQ HDR (BL) Static + Dynamic Yes (with IP) Single 4 Philips/Technicolor-2 PQ SDR (BL) Static + Dynamic Yes Single 5 Dolby Vision-1 PQ HDR (BL) +EL Static + Dynamic Yes (with IP) Dual 6 Dolby Vision-2 PQ SDR (BL) + EL Dynamic Yes (with IP) Dual 4EVER-2 : go for UHD-TV phase 2!
  31. 31. 31 www.4ever-2.com On HLG - Hybrid Log-gamma • Transfer function proposed by BBC/NHK (Scene referred) EOTF: PQ vs BT 1886 (SDR) vs HLG 4EVER-2 : go for UHD-TV phase 2!
  32. 32. 32 www.4ever-2.com HDR Eco-System Standardization Status (1/2) • OETF • PQ - SMPTE ST 2084 • HLG- ARIB STD-B67 • Metadata • SMPTE ST 2086: define standard color grading display metadata • SMPTE ST 2094: dynamic metadata describing transformation HDR <->SDR (On-going) • HDMI • 2.0a: Support standard ST 2084 + metadata ST 2086 • 2.1: Hybrid Log Gamma (HLG), ST 2094 ? • IMF • metadata ST 2086 support soon 4EVER-2 : go for UHD-TV phase 2!
  33. 33. 33 www.4ever-2.com HDR Eco-System Standardization Status (2/2) • Blu-Ray Disc Association (BDA) • HDR 10: HEVC main 10-bits + ST-2084 + ST-2086 Metadata • Optional: Proprietary metadata: Dolby, Philips, Technicolor • Compression and application standardization • MPEG/ITU HEVC: • VUI values for PQ and HLG • SEI messages: • ARIB: standardized the usage of HLG: STD B67 • DVB: PQ, HLG, Dolby under consideration  Mastering Display Colour Volume (ST 2086),  Chroma resampling filter hint,  Knee function information,  Colour Remapping Information  Content Light Level Information  HDR Compatibility Information 4EVER-2 : go for UHD-TV phase 2!
  34. 34. Recent 4EVER-2 HDR live trials set up real live TV production and distribution www.4ever-2.com 344EVER-2 : go for UHD-TV phase 2!
  35. 35. Backward compatibility in TV production • Main drawback of PQ is that it is not SDR backward compatible – SDR and HDR shall be simulcasted • The goal is to produce HD 709 + Ultra HD Phase 1 709 or 2020 and add HDR in the same workflow • With few extra costs ! – no additional cameras – using current technology (SDI 10 bits …) – with few extra charge for production teams • Because TV distribution will remain mostly HD 709 at least for a number of years! www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 35
  36. 36. Candidates for TV production in HDR 1. Each camera outputs two (independent) signals – one HD/UHD phase1 – one PQ 10 – vision engineering can optimize SDR and HDR separately 2. Hybrid Log Gamma – HDR and SDR are controled using one video signal – vision engineering shall optimize SDR and HDR jointly 3. HDR or SDR + Metadata ? – in a static or dynamic way? www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 36
  37. 37. 1/ Live sport in HDR and HFR Figure ice skating: HDR for brightness on ice and costumes, HFR for figure motion www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 37 • But, 13/11/2015 in France  … experimentation shortened, only HDR tested • 2 different cameras, one at a time - GrassValley LDX - Sony HDC4300 • 4K zooms: Fujinon 80x (UA80X9BE) 13-14 November 2015 Bordeaux, France
  38. 38. 1/ Live sport in HDR www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 38 • 1st test (LDX) => a single camera to generate two streams: one HDR and one SDR • 2nd test => an HDR-PQ camera (Sony 4300) and a SDR camera on a rig (Sony 2400) • Remote control from production truck • HDR and SDR comparisons
  39. 39. 1/ Live sport in HDR www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 39 Production truck: simulation of what will be next, when… …here!  An extremely valuable experimentation, very own production space for HDR camera remote control
  40. 40. 1/ Live sport in HDR www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 40 Many lessons learned on where difficulties are: how to control SDR and HDR independently, yet in a single production?
  41. 41. 2/ Bastille Opera live HDR production A single production to serve both HDR and non HDR: use of HLG technology www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 41 15 December 2015 Bastille Opera, Paris France
  42. 42. 2/ Bastille Opera live HDR production www.4ever-2.com 42 HDR monitoring 709 monitoring 709 color gamut Quad 3G SDI 4K HLG Quad 3G SDI 4K HLG downsizing HD 709 vision engineer focus in SDR, then control HDR result potential 709 graphics mapped to HLG with LUT for instance towards HEVC encoder with HLG signaling for instance towards H264 encoder 4EVER-2 : go for UHD-TV phase 2! 709 monitoring OBvan HD signal Sony 4300 (HLG prototype version) zoom Canon 90x (J90x9B) Our HLG workflow, generating both 4K-HDR and HD-SDR for comparison to broadcasted commercial HDTV signal
  43. 43. 2/ Bastille Opera live HDR production www.4ever-2.com 434EVER-2 : go for UHD-TV phase 2! Comparison between HD and UHD + HDR
  44. 44. www.4ever-2.com 444EVER-2 : go for UHD-TV phase 2! A big success for invited producers and directors of photography: loved both the SDR quality and the HDR potential 2/ Bastille Opera live HDR production
  45. 45. www.4ever-2.com 454EVER-2 : go for UHD-TV phase 2! Despite tough content parameters… 2/ Bastille Opera live HDR production
  46. 46. Ultra HD phase 2 focus on HFR www.4ever-2.com 464EVER-2 : go for UHD-TV phase 2!
  47. 47. HFR vs. TV set « sport mode » www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 47 High Frame Rate 50 frames / s transmission 100 frames / s transmission Display interpolation ("sport mode")
  48. 48. Why doesn’t interpolation work well? • Frame interpolation (refreshment) does not always work • Refreshment invents content => artefacts will happen: object is doubled, blurry, disappears, … www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 48 Refreshment as displayed on good consumer TV sets (HD 1080p50, HEVC 7Mbps)
  49. 49. - Collaborative work with EBU, BBC and IRT (BTF HFR) - Evaluate the added value of 100Hz in comparison with 50Hz in terms of perceived video quality - Uncompressed contents - Compressed contents - Using HEVC at 3 Mbps (BBC contribution) - MCFI professional solution - Using Alchemist OD (IRT contribution) CONTEXT AND TEST OBJECTIVE
  50. 50. - SAMVIQ (ITU-R BT. 1788) - Global notation for short duration clips - Continuous quality scale (0-100) using quality items - Bad, Poor, Fair, Good and Excellent - Use of explicit and/or hidden references - Observers - Non-expert viewers - Normal screening - Rejection criteria used for results analysis SUBJECTIVE TEST METHOD
  51. 51. VIEWING ENVIRONMENT - Based on Recommendation ITU-R BT. 500 - Background luminance vs display peak luminance - Viewing distance - Room colors
  52. 52. TEST ORGANIZATION - Viewing environment - Viewing distance: 3H (2m) - Background: 10% of the display peak luminance - Test setup - Subjective quality assessment software (Orange Labs tool) - PC-based player: HP Z820, Matrox DSX LE4 - 55-inch HFR display (Mikrom): 4 x 3G-SDI (HD, 200/240p) - 31 observers - Non-experts - Normal screening
  53. 53. TEST MATERIAL (1) - 5 contents (sources) - HD format (1920x1080p200, 16 seconds) - Different motion features (camera motion, etc.) BBC4 EBU BBC9 BBC3 4EVER
  54. 54. TEST MATERIAL (2) - 6 test conditions: 3 frame rates, 2 shutter values - Test 1: uncompressed contents - Test 2: compressed contents at 3 Mbps (HEVC) 200Hz 100% Averaging 100Hz 100% Averaging 50Hz 100% Down sampling 50Hz 50% MCFI 50 to 100Hz From 50Hz(50%) MCFI 50 to 100Hz From 50Hz(100%) Frame rate Shutter
  55. 55. RESULTS: UNCOMPRESSED CONTENTS 100Hz native contents provide better quality than 50Hz native contents (about 20 points) Frame rate : Shutter :
  56. 56. RESULTS: UNCOMPRESSED CONTENTS Frame rate : Shutter : Frame rate : Shutter : Frame rate : Shutter : Frame rate : Shutter : Frame rate : Shutter :
  57. 57. RESULTS: COMPRESSED CONTENTS (3 MBPS) 100Hz native contents provide better quality than 50Hz native contents (about 15 points) Frame rate : Shutter :
  58. 58. RESULTS: COMPRESSED CONTENTS (3 MBPS)
  59. 59. MAIN CONCLUSIONS - 100Hz native contents provide much better perceived video quality than 50Hz native ones - The performance of Interpolation (MCFI) depends on the content’s motion features - In the worst cases, the perceived video quality is lower than 50Hz native ones (visual artefacts) - In the best cases, the perceived video quality can reach the 100Hz native ones - Best performances at 50% shutter vs 100% - In average, 50Hz interpolated at 100Hz cannot reach the perceived video quality of native 100Hz - At the same bitrate, the perceived quality of a 100Hz source remains better than a 50Hz source (interpolated or not).
  60. 60. live HFR + HDR workflow www.4ever-2.com 61 HDR monitoring 709 monitoring 709 color gamut 8x 3G SDI 4K HLG HFR 8x 3G SDI 4K HLG HFR downsizing HD 709 HFR 709 graphics upsampled to 4K HFR HEVC encoder with HLG signaling HEVC encoder HD 709 frame averaging H264 encoder 4K 709 frame averaging HEVC encoder WCG can also be introduced with extra complexity ?how to sync successive frames 4EVER-2 : go for UHD-TV phase 2!
  61. 61. Compressor challenges www.4ever-2.com 624EVER-2 : go for UHD-TV phase 2!
  62. 62. 634EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Compressor challenges • Metadata feeding • Static/dynamic metadata • No standard methods: IMF, Side XML, VANC, others… • Metadata generation in the encoder? • Content preparation • Chroma Subsampling/downscaling • Pre-processing • Color space conversion • Compression • HDR Signal properties => new video defects may appear • Chroma and luma errors • 10-bit banding and contouring
  63. 63. 644EVER-2 : go for UHD-TV phase 2! www.4ever-2.com HDR adapted content processing OETF is highly non-linear => Non-Linear algorithms have to be implemented • Chroma sub-sampling • HEVC Main 10 encodes 4:2:0 chroma samples • Source is usually RGB 4:4:4 =>Sub-sampling has to be performed prior to encoding • Conventional approaches can not be used => need to take transfer function characteristics into account Y components U,V components R,G,B components
  64. 64. 654EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Color Processing Pipeline
  65. 65. 664EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Halos and chroma shifts • Problem caused by the steep EOTF curve at high luminance • For instance: • An error of 16 in code value (1.5% of the full scale) • => a luminance error of 123 nits (10% of the full scale)
  66. 66. 674EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Small errors cause large defects • Strong mosquito noise and chroma errors around bright edges • Caused by quantization error spreading in blocks Small area of a HDR source picture Compressed result
  67. 67. 684EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Avoid error spreading • Take advantage of HEVC coding tree • Small TU on bright edges for visual improvement PU TU 64x648x8 CU
  68. 68. 694EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Banding artifacts • Most annoying and frequent visual defect when compressing HDR content • Appears in unexpected areas even at high bit-rate Area exhibiting banding Luminance map
  69. 69. 704EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Barten ramp vs 10/11-bit PQ • Barten ramp gives an indication of minimum visible contrast step • One code value change in 10-bit may be visible, not in 11-bit 0.1% 1% 10% 0.001 0.01 0.1 1 10 100 1000 10000 MinimumContrastStep(%) Luminance (cd/m2) Barten Ramp 10-bit ST-2084 PQ 11-bit ST-2084 PQ
  70. 70. 714EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Contrast visibility • All common 10-bit EOTF perform about the same • One code value change in 10-bit may be visible with all EOTF 0.1% 1% 10% 0.01 0.1 1 10 100 1000 10000 MinimumContrastStep(%) Luminance (cd/m2) Barten Ramp 10-bitST-2084 PQ 10-bit ARIB-B27 1200nits 10-bitBT.709 200nits
  71. 71. 724EVER-2 : go for UHD-TV phase 2! www.4ever-2.com On noise encoding • Avoiding banding artifacts is performed by adding dithering noise on the source • But HEVC, H.264 and MPEG-2 are not efficient on noise • The video codec tends to remove dithering noise leading to banding
  72. 72. 734EVER-2 : go for UHD-TV phase 2! www.4ever-2.com How the encoder can help? • Detect areas prone to banding an improve PSNR as much as possible • Add strong in-loop dithering noise prior to encode in relevant areas • Use Film Grain technology (generate noise on the receiver side) when available • Or increase contrast and add a stronger dithering noise un post-production…but… ST-2084 BT.709
  73. 73. 744EVER-2 : go for UHD-TV phase 2! www.4ever-2.com Bitrate impacts • PQ + metadata solution • HDR only • HDR10 + metadata : very few bytes on GOP or scene basis • Backward compatible • SDR + metadata => SDR base layer is encoded with increased bitrate of around 5% • HLG solution • No bitrate overhead => further experiments Dolby Vision Solution • Backward compatible • Base Layer (either SDR or HDR) + Enhancement Layer => EL bitrate is ¼ of the base layer bitrate
  74. 74. Realisation of the first UHD-TV phase 2 TV clip Using all UHD-TV phase 2 technologies (4K, HFR, HDR, WCG) plus NextGen Audio www.4ever-2.com 754EVER-2 : go for UHD-TV phase 2!
  75. 75. First fiction in UHD-TV phase 2 + 3D audio www.4ever-2.com 764EVER-2 : go for UHD-TV phase 2! 4-7 November 2015 Paris, France « 4EVER short », « Une danse pour la vie » Quote from the author « All the technical tools to define tomorrow’s TV have a common goal: researching viewer immersion. This short story narrative goes the same way, immerging the audience into emotion »
  76. 76. Fiction in 4K, HDR, HFR, WCG and 3D audio www.4ever-2.com 774EVER-2 : go for UHD-TV phase 2! Bright lights in motion at night… Motion inside a structure
  77. 77. www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 78 Ultra High Technology to serve artistic intention, taking into account the power of the story
  78. 78. www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 79 All about lights and contrast
  79. 79. www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 80 Director of photography and HDR experts team Watch the making-of http://www.dailymotion.com/video/x3iw04w_4evershort_tv
  80. 80. Conclusions and perspectives www.4ever-2.com 814EVER-2 : go for UHD-TV phase 2!
  81. 81. HDR: high expectations… still some issues • No bitrate defined yet – New nature of the video signal impacts encoder performance • Current HDR TV sets are first generation – Between 500 and 1000 cd/m² in 2015 (target = 1500 cd/m²) www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 82 • No standards yet, too many options – Transfer functions (EOTF) => Dolby PQ, BBC/NHK HLG – Format: 1 or 2 layers? Metadata or not? => Technicolor, Dolby, Philips – Encoding: HEVC Main 10 or future MPEG technology But live production can be operational pretty soon!
  82. 82. HFR: nearly ready, yet some issues • is 8xSDI manageable? • monitoring in 50p and mix in 100p? • sync issues – how to sync 2x 50 Hz SDI feeds? • 100Hz clock? • 50 Hz clock? If two feeds have a 50hz clock, which one is first? Can synch be based on odd/even frames? – how to apply DVE to SDI feeds with same timestamps • ie: how to fade two pictures at the same time but with different fading parameters? • is it possible to add a delay of 10 ms for one picture effect? • instant replay / slow motion in 4K 100 ? – capture in 4K 300fps??? – interpolation? • HD HFR production is probably a better alternative…. www.4ever-2.com 834EVER-2 : go for UHD-TV phase 2!
  83. 83. Technology impact on perceived quality • Video subjective tests target in 2016 – HDR • Comparison of solutions in terms of perceived video quality • Impact of compression on HDR perceived video quality – HFR • Impact of HFR on 4K resolution in terms of perceived video quality • but displays? – HFR + HDR • Impact of HFR + HDR combination in terms of perceived video quality • NextGen Audio with UHD-Phase 2: tests end 2016/2017 www.4ever-2.com 844EVER-2 : go for UHD-TV phase 2!
  84. 84. Towards live HDR - WCG - HFR - 4K • Continue independent evaluations of – Technology (visual assessments) – Equipment : characterization of displays or cameras – End-to-end workflow • Promote HFR: a major added value for the first to jump in, a great way to differentiate • Shoot sequences ourselves: complete parameter control and rights – Feb/Mar16: tbd – second HDR test – distribution trial • Visible events – Apr16: HD/4K/HDR live production and distribution – Jun/Jul16: Sports - HD/4K HDR/HD HFR live production and distribution test www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 85
  85. 85. Towards live HDR - HFR - 4K • Disseminate results in standardization committees and conferences – DVB, ITU, MPEG, SMPTE – For information to be where and when people make decisions, worldwide – Selected for NAB2016 Futures Parks; – IBC2016 FutureZone? • Collaborations – display/camera manufacturers and everyone welcome! www.4ever-2.com 4EVER-2 : go for UHD-TV phase 2! 86
  86. 86. Merci ! Thank you! www.4ever-2.com maryline.clare @orange.com j.vieron@ateme.com
  87. 87. Appendices www.4ever-2.com 884EVER-2 : go for UHD-TV phase 2!
  88. 88. HDR and WCG workflows for TV production and distribution www.4ever-2.com 894EVER-2 : go for UHD-TV phase 2!
  89. 89. Prepare for a live UHD-HDR production • Which camera and attached processors? • Which HDR-SDR workflow? • Which methodology for vision engineers? • Which additional cost compared to HD, compared to UHD phase 1? • How convince producers, and their photographers? www.4ever-2.com 904EVER-2 : go for UHD-TV phase 2!
  90. 90. Backward compatibility in TV production • The goal is to produce HD 709 + Ultra HD Phase 1 709 or 2020 and add HDR in the same workflow – Adding HFR is also a must have, especially for sports! • With few extra costs ! – no additional cameras – using current technology (SDI 10 bits …) – with few extra charge for production teams • because TV distribution will remain mostly HD 709 at least for a number of years! www.4ever-2.com 914EVER-2 : go for UHD-TV phase 2!
  91. 91. ? PQ/709 OETF + WCG/709 workflow www.4ever-2.com 92 2020/P3 gamut 4K HDR OETF 709 gamut HD or 4K 709 OETF 3DLut WCG 2020/P3 to 709 Quad 3G SDI 4K PQ WCG Quad 3G SDI 4K 709 WCG Quad 3G SDI 4K PQ WCG Quad 3G SDI 4K 709 WCG 3DLut downsizing HD 709 common settings ? 709 specific ? PQ specific ? 709 graphics mapped in WCG with LUT, 4K or upsampled Quad 3G SDI 4K 7093DLut HDR 10 encoder HEVC encoder H264 encoder HEVC encoder + « fast track » Backward Compatible encoder? mastering information 4EVER-2 : go for UHD-TV phase 2!
  92. 92. ? HLG OETF + 709 workflow www.4ever-2.com 93 HDR monitoring 709 monitoring 709 color gamut Quad 3G SDI 4K HLG Quad 3G SDI 4K HLG downsizing HD 709 common settings ? SDR vs HDR tradeoff? 709 graphics mapped to HLG with LUT HEVC encoder with HLG signaling H264 encoder 4EVER-2 : go for UHD-TV phase 2!
  93. 93. ? HLG OETF + WCG workflow www.4ever-2.com 94 HDR monitoring 709 monitoring WCG color gamut Quad 3G SDI 4K HLG WCG Quad 3G SDI 4K HLG downsizing HD 709 common settings ? SDR vs HDR tradeoff? 709 graphics mapped to HLG WCG using LUT HEVC encoder with HLG signaling H264 encoder 3DLut WCG to 709 3DLut WCG to 709 static color remapping metadata 4EVER-2 : go for UHD-TV phase 2!
  94. 94. live HDR workflow issues Camera Paint menu operations • PQ vs 709 – common parameters: iris, color temp, details, noise, ..? – separate parameters: gamma/knee, max lum, blacks ? others? • HLG – how to make a tradeoff between SDR efficiency and HDR efficiency? using knee point, max luminance? others? • in both cases, some colorimetry mismatch should also be addressed, mainly for skin tones • and of course camera sensor dynamic is essential! www.4ever-2.com 954EVER-2 : go for UHD-TV phase 2!

×