Visual and technical quality control for high definition television

Visual and technical quality control
for high definition television

16/1

Tom Beckers
Wim Ermens

Overview

Video quality control in a filebased world
Impact of HDTV emergence on quality control
Video quality evaluation
Production chain simulation
Broadcast chain simulation

VRT-medialab: onderzoek en innovatie


AudioVisual Quality Control
! Audio
! Video
! Interaction between audio and video : e.g. lipsync

File based AV Quality Control
! Part of an automated or manual workflow
– Diagnose
– Repair / Redo



Technical quality control
! Container
! Metadata
! Interaction between container, audio and video : e.g. duration of tracks

File based technical quality control
! Part of an automated or manual workflow
! Application specifications



Automated QC


Application Specs DV-25 (News) D-10 (APP) HD
Contribution
MXF analysis & application Production

specification validation Play-out

Archive

Application
Specification

IRT MXF
Analyzer
MXF MXF
file Validator
analysis
Ardendo
CheckMXF
MXF
analysis
report



Application specification
validation report



Safeguarding audiovisual quality
! Maintaining quality throughout the production chain
– Choose material as close to source as possible
– prevent unneeded multi-generation
– Try to produce with the shortest/’most apt’ chain
– Prevent unneeded multi-generation
– Prevent transcoding 720p
720p 720p ! 1080p !
1080i !
– Check quality
1080p Black Box 720p
1080i
! Carefully design the production chain (in production/broadcast
– Choose the right codecs 1080i 1080i ! chain) 1080p!
720p ! 1080i
– Choose the right equipment 1080p
720p 1080i



Proliferation of raster formats
SD TV HD TV

1080p25

PAL
576i50 720p50 1080i50

1080p50



Proliferation of codecs
HDV 50 Mbps
ProRes Dirac
JPEG2K
10 bit sampling
4:2:2
4:4:4
YUV HDCAM-SR XDCAM HD42
DVCProHD
180Mbps
440Mbps 100Mbps
DNxHD
HDCAM RGB
4:2:0 / 4:1:1
SVC
AVC-Intra 8 bit sampling XDCAM HD
AVCHD


Viewer quality expectation
! Consumer expects ROI from flat panel investment
! Depends on content type (drama, news, sports, soap)

What codec to choose?
! Picture quality ! quality evaluation needed
! Bit rate
! Performance
! Support
! Scalability


Video sequence used for testing
(source EBU) :
! High quality source material
! 10-bit YCBCR samples
! Chroma subsampling 4:2:2
! HD-SDI compatible
! 720p50


Note on viewing conditions

Expert evaluation :
! Class 1 reference monitor @ 1H, 3H
! High quality consumer device (FullHD, Plasma) @ 1H, 3H

Present demo setup
! Overhead projector (internal processing?)
! Not sufficiently dark room
! Sub ideal circumstances

Interested parties : ask for demo in better circumstances...


Video quality in production chain

Simulation chain

without
pixelshift with
pixelshift (+2H, +6V)

Camera Post production encoding
encoding (4 generations)



Codecs tested :
! Camera : AVC-I, DVCProHD, XDCAM HD422
! PostProduction : DNxHD, AVC-I, DVCProHD, XDCAM HD422

Codec properties
Codec Bitrate Chroma Sample GOP Spatial subsampling
(Mbps) subsampling size
(bits)
AVC-I 100 4:2:2 10 I frames only None
DVCProHD 100 4:2:2 8 I frames only 3:4 horizontal :
960 luma samples/line
DNxHD 115 4:2:2 8 I frames only None
XDCAM 50 4:2:2 8 LongGOP None
HD422 (n=12 @ 720p)


Test setup

Sony PDW-HD1500* Panasonic AJ-HPM110** Avid Airspeed + DNXchange
XDCAM HD422 DVCProHD DNxHD
AVC-I
* test model provided by Sony Belgium
** test model provided by Panasonic Belgium


Test setup

Encode Decode

HD-SDI HD-SDI

Uncompr. Gen 0 Gen 1 Gen 2 Gen 3 Gen 3 Gen 4
source (Cam) (PP1) (PP2) (PP3) shifted (PP4)
Uncompressed YCbCr storage
HD-SDI ingest & playout


Evaluation methods :

! Objective tests

! Subjective tests

! Automated subjective tests

! Expert viewing



Objective tests : PSNR, SSIM, ...

1 mquot;1 n quot;1 MAX R
MSE frame = !! R(i, j ) quot; I (i, j )
2
PSNR frame = 20 ! log10 (
mn i =0 j =0 MSE frame

Reference Impaired Squared difference


Objective tests : PSNR, SSIM, ...

Advantages Disadvantages
objective 3 resultsets per sequence (Y, Cb,
Cr)
reproducible
unsensitive to grouping of pixel
automated test
differences


Subjective tests : DSCQS, TSCES, ...


Upper anchor
Source material

Material to be
evaluated

Lower anchor
Worst case
(D)MOS scores



Subjective tests :
DSCQS, TSCES, ...

closest to real life experience difficult to test many sequences
timeconsuming to setup
difficult to get sufficient statistics
worst case reference definition


Automated subjective tests : Tektronix PQA-500, Opticom PEVQ Analyzer, ...
! Aims to provide (D)MOS score
! Simulate :
– Monitor setup and characteristics
– Viewing setup and conditions
– Viewer setup and characterisics
– Human visual system (= tricky)

automated test good correlation to subjective test
results to be proven
closer correlation to subjective test
sults than PSNR typically slower than objective tests
no worst case reference mandatory



Expert viewing
! Aim :
– Visually evaluate video quality
– Determine typical artefacts
– Compare source with test material
– Compare test material with other test material

! Method :
– Create splitscreen images
– Evaluate on (reference) monitor
– Create ranking from intercomparison



Expert viewing (“The proof of the pudding is in the eating”)



Uncompressed vs Compressed video material



Compressed vs Compressed video material


Video quality due to de-interlacing
source de-interlaced

De-interlacing: difficult to perform
Good quality -converters exist (price!)
720p DNxHD (Gen0) 1080i - DNxHD (Gen0) – de-interlace – 720p


Video quality in distribution chain


Distribution encoder

Distribution encoder = device that compresses the broadcaster’s signal into
a format suitable for transmission/distribution
Codecs: MPEG-2 or MPEG-4 H.264 (tested)
Interfaces: HD-SDI quot; ASI
Typical bitrate in/out:
! From 50-200 Mbit/s quot; 10 Mbit/s (or less)

Parameters:
! Encoding delay
! Encoder bit-rate
! Profile/Noise reduction/GOP size/…


Distribution encoder: bit-rate

AVC/DNxHD – 16Mbit/s H.264
Which bit-rate to choose?

Uncompressed
! Distribution channel ‘defines’ available bit-rate
! MPEG-4 does a fine job
! Motion in video is important



Distribution encoder and production codec

Uncompressed HD ~ 1 500 Mbit/s
Production bitrate ~ 100 Mbit/s
Distribution encoder ~ 10 Mbit/s
Do we actually see the influence of the production codec?
AVC/AVC + 8 Mbit/s H.264 DVC/DVC + 8 Mbit/s H.264


Distribution: the end of the chain

Receiver (set-top-box)
! Interlacing – de-interlacing
! Scaling
Television
! A lot of choice
! De-interlace
! Scale
! Process (sharpen – noise reduction)
Make sure there’s no unnecessary signal conversion


Visual and technical quality control for high definition television

Recommandé

Recommandé

Contenu connexe

Tendances

Tendances (18)

Similaire à Visual and technical quality control for high definition television

Similaire à Visual and technical quality control for high definition television (20)

Plus de vrt-medialab

Plus de vrt-medialab (20)

Dernier

Dernier (20)

Visual and technical quality control for high definition television