Avia PROVideo Test DVDGuy Kuo, Ovation Multimedia, Inc.

IntroductionThis chapter discusses the video test signalsfound on the Avia PRO Video Test DVD and itssupplemental 4:3 Ratio DVD. Avia PRO containsthe most extensive set of video setup and testsignals ever released. Over 1,000 new digitallygenerated and mastered test signals (5,000counting variants) are included. Professionalcalibrators, engineers, reviewers, and designerswill recognize familiar signals in Avia PRO, butmany patterns include extra test features or arecompletely new. We recommend reading thischapter to learn about the new video tests andfeatures in Avia PRO.

Avia PRO’s extensive test set enables calibratorsto choose a pattern tailored to the task. Examplesinclude: geometry grids selectable in intensity,tinting, and spacing; grayscale windows availablein 2.5, 5 and 10 IRE increments; and APLintensities adjustable during black level adjust-ment. These and other signal options allow AviaPRO users to enjoy a near custom fit betweenpattern and task.

Avia PRO Video Test DVD's were designed forprofessional users rather than consumers. Theybehave as a responsive test signal source ratherthan an instructional video with incidentalcalibration signals. Professional users demandimmediate access to test signals with a minimumof material "in the way." Avia PRO Video TestDVD's feature quick access of all test patterns vialogically arranged menus or chapter/title numbernavigation.

As a professional level suite, Avia PRO was freedof limitations that protect lay users but interferewith experts. For instance, automatic patternadvance is the default on the consumer AviaGuide to Home Theater DVD. This appropriatelyreduced screen burn-in risk for consumers, butcalibrators, reviewers, and engineers benefit fromexplicit control of signal changes. Patterns inAvia PRO do not advance unless explicitlydirected by the user.

Avia PRO on-disc signal arrangement took intoconsideration the tasks commonly performed bycalibrators. Separate suites of patterns wereculled for CRT and fixed panel (digital) display

setup. Each suite features patterns selected tomeet the special needs of that class display. TheCRT suite places greater emphasis on geometryand convergence patterns while the fixed panelsuite includes more scaling, gamma tracking, andsignal limit tests.

Small sets of patterns were linked together tospeed commonly performed tasks. For instance,navigating between window patterns duringgrayscale calibration was cumbersome beforeAvia PRO. Calibrators either had to accept thehigh and low IRE window patterns that happenedto be adjacent on a disc or inconveniently stepbetween their desired window patterns. Avia PROusers can choose the high and low IRE windowsets appropriate for the situation. Once windowvalues are selected, Avia PRO speeds navigationbetween the desired windows during grayscalecalibration.

About Pattern GenerationTest patterns on the Avia PRO Video Test DVD'swere digitally synthesized in native 480P resolu-tion using signal synthesis software customwritten by the author. Direct synthesis in native480P resolution allowed creation of finelydetailed test material that would have beenmathematically impossible to create in a non-native resolution. However, detailed and accuratesignal generation is not enough. Test signals mustalso be accurately encoded for delivery on theDVD medium.

When the Avia PRO project was started, nosuitable MPEG2 encoding solution existed whichwas able to achieve the low noise, signal levelaccuracy, and full bandwidth desired. Highaccuracy encoding solutions were developedthrough close collaboration with Innobits, themakers of BitVice encoder. With special com-pression software supplied by Innobits, we wereable to preserve original signal levels, reducenoise, and maintain flat response for both lumaand chroma out to Nyquist limits. The result isthat Avia PRO delivers the most accurate, highlydetailed, wide bandwidth test signals on NTSCDVD. Users of BitVice now benefit from ad-vances made through the collaboration betweenOvation Multimedia and Innobits.

Resolution, FrequencyResponse, Group Delayand Y/C TimingAvia PRO provides tests for evaluation andadjustment of image detail and resolution. Thissection first discusses patterns designed for visualinspection. After that, instrument based tests offrequency response, group delay and Y/C delayusing oscilloscopes or waveform monitors areconsidered.

Resolution Chart

Resolution Chart: Image sharpness, resolution, edgetransition fidelity, geometry, scaling match, as well asoverscan can all be inspected with this single pattern.

Coupled with the right test signal, even visualinspection of a system's image output can detectloss of fine image detail, smearing, false outlinesor ringing at edge transitions. The ResolutionChart in Avia PRO contains multiple elements fortesting a display or entire video playback chain.In the center you find a zone plate pattern (con-centric circles of varying spacing) which testsfrequency response over a range of signal fre-quencies and angles. Also present are cross hairsindicating frame center. Two resolution wedgesindicate vertical resolution in TVL and tworesolution wedges indicate horizontal resolutionin MHz and TVL. T and T2 pulse bars at eachside of the pattern check positive and negativeedge transitions for ringing and smear. Threesmall, corner zone plates test frequency responseat fixed frequencies including 3.58 MHz, thecolor carrier frequency. A fourth circle at lowerright corner contains a 6.75 MHz burst that is atthe maximum limit of NTSC DVD resolution.Along the lower and right edges of the pattern,every other pixel stripes check scaler function.Triangular frame edge indicators and overscanlines every 1% show how much of each frameedge is lost to overscan. A large circle in thebackground serves as a geometry check. Finally,small text in each corner helps test corner focus.

Despite this chart being primarily a pattern fortesting resolution and scaler performance, manycalibrators will take advantage of its geometryfeatures to adjust display centering, geometry,and overscan. It has uses beyond resolutiontesting. When using this pattern for testingresolution, pay particular attention to the resolu-tion wedges, zone plates, T pulses and T2 pulses.

Resolution Wedges: The vertically long wedges(sinusoidal lines which begin widely spaced andgradually squeeze nearer together) measurehorizontal image resolution and bandwidth. Thehorizontally long wedges measure vertical imageresolution. On a perfect display system, the everyother pixel detail at the fine ends of the wedges isdistinctly visible. On most real life displaysystems, the higher frequencies of the wedges areattenuated causing the lines to blur together ordim. Some roll-off is normal, but ideally thefinest details will still be present. To estimateresolution, find the point on each wedge at whichthe lines are no longer distinct and read either theTVL or MHz number of the scales along eachwedge. MHz values are to the left of each verti-cally long wedge. TVL values are to the right ofthe vertically long wedges. The horizontally longwedges are only marked in TVL resolution.

Circular Zone Plates: ZP's are useful becausethey allow examination of performance at allangles. They can be fixed in spacing as in thecorner zone plates or vary in frequency as in thecentral zone plate which runs from 2.0 to 4.0MHz. Just as with resolution wedges, highfrequency losses appear as a blurring or dimmingof finer spaced detail. If video scaling is poorlydone, zone plates may exhibit moiré.

6.75 MHz Patch and Scaling Stripes: NTSCDVD has a maximal or Nyquist frequency of6.75 MHz. This corresponds to every other pixelin a pattern. The every other pixel spacing ofvertical lines in the lower right circular patch andin the stripes along the right and bottom edges ofthe pattern should be visible if the display re-solves the finest details recordable on NTSCDVD. If the display or scaler combination cannotexactly match or represent the fine detail in thispatch, you may see beat patterns (periodiclightening and darkening or bunching of lines)

Home Theater PC or fixed panel users strivingfor 1:1 pixel alignment can use the scaling stripesto detect when 1:1 pixel mapping is attained. Thenumber of visible beats within the scaling stripesis roughly equal to how many counts the currentscaled size is away from perfect 1:1 pixel scaling.

T and T2 Pulses: The black and white verticallines at each side of pattern are sine squaredpulses that increase in intensity (white) or de-crease in intensity (black). The shapes of thepulses are special. They are designed to keeptheir frequency components within limits thatshould be cleanly handled by normal playbackequipment. The slower (thicker) T2 pulses shouldshow neither overshoot nor undershoot (ringing)on equipment with flat frequency and good phasepreservation. The T pulses are more difficult toreproduce cleanly. You may see ringing of the Tpulses on consumer equipment.

Usage with Sharpness Controls: As you turnsharpness controls up or down, you will seevarious frequencies on the zone plates or wedgeslighten or darken. Brightening indicates that thecontrol is emphasizing a frequency. Ideally, thewedges will be even in intensity without anyportions overemphasized. Pay attention to the Tand T2 pulses and try to keep ringing minimizedat least around the thicker T2 pulses.

The Avia PRO Resolution chart provides multiplevisual indicators for frequency response testingas well as other display parameters. Other visu-ally oriented test patterns for resolution, band-width and testing are supplied by Avia PRO. TheSharpness pattern is a useful alternative forsetting both vertical and horizontal sharpness aswell as assessing chroma bandwidth.

Sharpness & Peaking

Sharpness & Peaking: Like the Resolution pattern, theSharpness pattern provides multiple frequency re-sponse and edge transition tests, but it is optimized foradjusting sharpness controls rather than determiningdisplay resolution.

The Avia PRO Sharpness & Peaking patternshares some features with the Resolution pattern.Overscan percentage marks, centering crosshair,edge indicators, scaling stripes and corner platesare common to both. Central content of thesharpness pattern is better attuned for adjustingsharpness controls. A luma frequency sweep

from 2.0 to 5.0 MHz, as well as expanded zoneplate range, and both horizontal and vertical T &T2 pulses allow easier observation of whichfrequencies are being affected and the presenceof ringing. Just as in the resolution pattern, a graybackground allows both undershoot and over-shoot around the T pulses to be visible. A blackbackground would not allow undershoot to bevisible. A white background (particularly on adigital display) could hide overshoot. Speakingof gray backgrounds, the intensity is not 50%gray but rather a special shade of gray.

Gray Background: The gray backgrounds of theResolution, Sharpness & Peaking, Zone Plate andResolution Plates patterns are not 50% gray.They are instead an intensity that matches theaverage integrated brightness of a full amplitudesine wave on a 2.2 gamma monitor. Thus, thespecial gray background in these patterns pro-vides a visual indication of signal attenuation oremphasis. Signal frequencies for which a displayhas flat response appear the same brightness asthe special gray background. Attenuated frequen-cies appear darker than the gray background.Emphasized frequencies appear brighter. A 50%gray would not have that special relationship.However, be cautious when reading results.Scaling artifacts can upset the relationship, soverification with a waveform monitor or oscillo-scope is recommended.

Chroma Sweeps: Horizontal Cr and Cb sweepsfrom 0.8 to 2.4 MHz are at the bottom of theSharpness pattern. These are for inspectingchroma frequency response. Unlike conventionalCr and Cb sweeps, the chroma sweeps in AviaPRO are RGB legal. In this pattern, their polari-ties are also restricted for better visual contrast.RGB legal chroma sweeps mean that even if thissignal is converted from component color spaceinto the RGB domain, the signal remains valid.Ordinary Cr and Cb sweeps distort and clip ifconverted between RGB and component colorspace. Avia PRO's sweeps allow testing despitethe use of RGB processing in a system.

Zone Plate

Zone Plate: Resolution at multiple angles and frequen-cies are tested with zone plate.

Another pattern for visual testing of resolution,bandwidth, and scaling is the Zone Plate Pattern.The center of this pattern is filled by a large zoneplate with frequencies from 1.0 to 4.0 MHz.Amplitude loss or emphasis at various frequen-cies can be seen as brightening, darkening, orblurring of the circular lines. Scaler preservationof detail can also be examined using this pattern.Scaling artifacts can appear as moiré patterns.Zone plate frequency range was chosen to mini-mize moiré in the pattern itself. This meansvisible moiré when displaying this pattern isprimarily due to the playback chain and displayrather than the pattern. T and T2 pulses areprovide for those wishing to adjust sharpnessusing this pattern.

Upper Video Frequency RangeTestingConventional frequency sweeps are of limitedvalue above approximately 4.5 MHz due to thesampling limitations of the NTSC DVD 720 x480 pixel map. Phase alignment effects createmoiré patterns and render sweeps difficult tointerpret. The problem worsens as frequencyapproaches 6.75 MHz. Indeed, a 6.75 MHz signalthat is out of phase relative to the samplinginterval would appear as gray without any finedetail. Despite difficulty of measurement, infor-mation regarding performance in the upperfrequency ranges is desirable. Avia PRO providestwo means of examining phase and frequencyresponse between 4.5 and 6.75 MHz. One is theResolution Plates pattern series discussed here.The other, Polyphasic Sweeps, is discussed later.

Resolution PlatesThe Resolution Plates test pattern sequenceallows examination of several frequencies in-cluding some in the difficult, upper range. Zoneplates, with their multiple angles shift phasealignment, allow some information to be gained.Moiré will still be present, but one can at leastsee if detail is being lost at a frequency. Resolu-tion Plates are marked in both MHz and TVL. Aseach successive pattern in the sequence appears,the set of frequencies changes. Press pause onyour DVD remote to examine a plate in extendeddetail.

Resolution Plates: A sequence of patterns with zoneplates for examining higher frequencies. Pause the DVDplayer when the plates of interest are visible.

Scaling Test Stripes: The Resolution Platepatterns also have every other line and everyother two line scaling stripes. The stripes alongthe top and left of the patterns are on/off everytwo pixels. Scaling stripes along the right andbottom of the patterns are every other line inspacing. Every other line scaling stripes are oftennot rendered well on consumer displays, but alldisplays should have sufficient resolution andimage sampling to render the every two linestripes well.

Instrumented Frequency and PhaseResponse Testing

MultiburstFaithful reproduction of video requires thatsignals be preserved not only in terms of overallamplitude, but also bandwidth and group delay.Distortions that affect frequency and phaseresponse appear in images as blurring and/oredge transition ringing. Examination of wave-forms with an oscilloscope or waveform monitor(an oscilloscope optimized for video signals) canidentify problems in frequency or phase response.Although a video signal generator can also beused as the test signal source, only test signals ona DVD test the entire playback chain.

Multiburst Traditional: Packets of sine waves for testingresponse at frequencies of interest.

The most easily understood test patterns fortesting frequency response are multiburst pat-terns. Multiburst patterns contain equal ampli-tude, sinusoidal signal bursts at various frequen-cies. By examining the amplitude of each burston a scope, one can determine if a particularfrequency is attenuated or emphasized. Care wastaken to keep Avia PRO's on-disc burst ampli-tudes equal. Otherwise, observed attenuation oremphasis might be due to the test signal ratherthan the equipment being tested. Flags at theleftmost portion of the pattern indicate theexpected excursion of bursts. Intrinsic to severalAvia PRO patterns are 3 dB and 6 dB attenuationindicators along the top of the pattern. Theseappear on scope as lines making dB attenuationmeasurement simple even if the scope lacks dBindicators.

Avia PRO supplies luma multiburst patterns inboth labeled and unlabeled versions allowingeither easy identification of frequency on screenor uncluttered waveform scope display. 1/2amplitude patterns are also supplied for situa-tions, such as VCR alignment, in which fullexcursion signals are inappropriate.

Waveform of Multiburst Traditional: Some frequencypackets cannot be accurately represented due to NTSCDVD video sampling rate limitations. This appears asminor unevenness of affected packet waveforms even ifequipment response if flat.

The traditional version of multiburst has bursts at0.5, 1.0, 2.0, 3.0, 3.58, and 4.2 MHz. Thesematch the frequencies available from a conven-tional signal generator. However, the 13.5 MHzsampling rate of video destined for NTSC DVDdoes not allow precise representation of a sinewave at all these frequencies. This is seen onscope as uneven waveforms in the packets.Despite this limitation, the traditional versionmultiburst is useful because it contains importantfrequencies such as the 3.58 NTSC color carrierfrequency. However, uneven on-disc representa-tion of the traditional bursts can make interpreta-tion of frequency response difficult. Even ifequipment frequency response is completely flat,the traditional multiburst waveform looks slightlyuneven. To avoid this problem, Avia PRO alsoincludes multiburst DVD Nyquist patterns. Thesecontain frequency packets cleanly representedwithin the 13.5 MHz sampling rate of NTSCDVD. The frequency bursts are all related to theNTSC DVD Nyquist frequency, hence the nameMultiburst DVD Nyquist.

Multiburst DVD Nyquist: Packets of sine waves phasealigned and at frequencies suitable for undistortedrepresentation on NTSC DVD.

Waveform of Multiburst DVD Nyquist: Frequencypackets at Nyquist related frequencies better maintainconstant amplitude than traditional frequencies.

Sweep PatternsMultiburst patterns are easy to understand, buttest frequency response at only the frequenciesfor which they provide bursts. Problems withresponse at frequencies between the suppliedfrequencies are not detected with multiburstpatterns. To test frequency response more com-pletely, sweep patterns are useful. These providea continuously varying frequency sinusoidalsweep which test a frequency range rather thandiscrete frequencies.

Horizontal Sweep: Sinusoidal waveform varies infrequency across pattern. Luma and color componentversions are supplied.

Avia PRO sweep patterns are supplied in severalfrequency ranges, in full or half intensity, inmultiple component color versions, in vertical orhorizontal, in labeled or unlabeled form, and inAvia PRO's exclusive polyphasic form. This widevariety of sweep patterns gives great flexibility intesting.

The Avia PRO sweep patterns share some fea-tures with the multiburst patterns. You will againfind 3 and 6 dB attenuation indicators at the topof the patterns. The left side of the patterns hasindicators for black and white level. New at theextreme right side of the patterns is a 6.75 MHzNyquist burst to check top end frequency re-sponse. For chroma sweeps, the frequency limits

and burst frequencies are of course different fromthose in luma sweeps.

Attenuation Stripes: Stripes at the top of sweepsappear on scope as lines indicating 3 dB and 6 dBsignal loss. The scope must be in multi-linedisplay (line select off) for markers to appear.

Vertical Sweep: TVL resolution is tested using thispattern.

White and Black Flags: At each sweep pattern'sleft (or top if vertical sweep), flags indicate thetop and bottom waveform excursion expected forflat response. The white flag is at 100 IRE andthe black is at 7.5 IRE. If North American 7.5setup is not present, black is at 0 IRE.

Waveform of Horizontal Sweep: Avia PRO maintainssweep amplitude as flat as possible to facilitate inter-pretation of scope results. Note burst at 6.75 MHz isencoded on disc at full amplitude.

Frequency Labels and Markers are at thebottom of the pattern. If a scope is displaying allscan lines, the markers appear on screen andallow easy identification of test frequencies. Non-labeled sweeps are supplied to permit unencum-bered visualization of markers.

Nyquist Burst: At the extreme right end of thesweeps is a burst at Nyquist (6.75 MHz for lumasweeps/ 3.38 MHz for chroma sweeps). The burstis encoded at the full amplitude as indicated bythe flags at the left of the pattern. Attenuation ofthe burst is often seen in analog playback chains.Fully digital, wide bandwidth playback systemssuch as home theater PC's typically preserve theburst amplitude.

Polyphasic Sweeps

Mentioned earlier was the problem with sweepsand the upper portion of the video frequencyrange. Phase misalignment between sine wavesand the sampling pixel structure overwhelms thedesired content. Severe phasing and bandingdistortions appear in conventional sweeps thatrange too high in frequency. Avia PRO providesnew tests to help elucidate response at thosedifficult, upper frequencies. Resolution Plateswere discussed earlier. In this section, we revealAvia PRO's polyphasic sweeps. Polyphasicsweeps gradually shift the phase relationshipbetween their waveform and the sampling pixelstructure through 2 pi radians (advances through360 degrees of phase shift). You can now observefrequency response at all phase alignments usingpolyphasic sweeps. On an oscilloscope, theamplitude envelope of the sweep grows andshrinks as phase varies. By observing both peakand null response over multiple phases, one cangather information about high frequency responseunavailable from conventional sweeps.

Polyphasic sweeps are provided for luma, colorcomponents, vertical and horizontal versions.You will notice that the screen area covered bypolyphasic sweeps is smaller than fixed sweeps.This was intentionally done to safeguard signalquality. The constrained screen coverage reducesbit rate requirements so the highly detailedpolyphasic sweeps are better preserved on disc.

Sweep Patterns on Disc

Sweep patterns on the Avia PRO Video TestDVD's are supplied in both horizontal and verti-cal orientations. Vertical sweeps in both 1.33 and1.78 ratio versions are needed to take into ac-count the effect aspect ratio has on TVL resolu-tion. Chroma sweeps are supplied in eight varia-tions of chroma modulation. Chroma sweeps areRGB legal to facilitate testing even if conversionback and forth between component and RGBcolor space is performed by equipment.

Luma Sweeps

Horizontal Sweeps1.0 to 4.0 MHz0.5 to 2.5 MHz1.5 to 3.5 MHz2.5 to 4.5 MHzPolyphasic 3.5 to 6.75 MHz

1/2 Amplitude Horizontal Sweep1.0 to 4.0 MHz0.5 to 2.5 MHz1.5 to 3.5 MHz2.5 to 4.5 MHzPolyphasic 3.5 to 6.75 MHz

Vertical Sweeps100 to 250 TVLPolyphasic 100 to 405 TVL

1/2 Amplitude Vertical Sweep100 to 250 TVLPolyphasic 100 to 405 TVL

Vertical Luma Sweeps in 1.33 Ratio100 to 300 TVLPolyphasic 100 to 540 TVL

1/2 Amplitude 100 to 300 TVL1/2 Amplitude Polyphasic 100 to 540TVL

Chroma Sweeps

Each chroma sweep is supplied in the fol-lowing modulations:

Cb Only Cr OnlyCb Negative Cr NegativeCb Positive Cr PositiveCb & Cr Inverse Cb & Cr

Sweeps are as follows:

Chroma Sweeps in 1.78 RatioHorizontal 0.5 to 2.0 MHzHoriz. Polyphasic 0.5 to 3.38 MHz

Vertical 60 to 120 TVLVertical Polyphasic 60 to 200 MHz

Chroma Vertical Sweeps in 1.33 RatioChroma Sweep 40 to 160 TVLChroma Polyphasic 40 to 160 TVL

Group Delay & Phase ResponseSignal reproduction requires not only flat fre-quency response but also correct time alignmentof the frequency components. Multiburst andsweep tests check only frequency response.While that may be sufficient for some purposes,phase response or group delay problems cancreate significant distortions even if frequencyresponse in a system is flat. Unfortunately, phaseresponse is more difficult to explain and too oftenignored in equipment reviews.

Group delay errors are easier to understand if onethinks of the video signal as the sum of manyfrequency components that must be reassembledto form the original signal. The amplitude of allthose components must be preserved (flat fre-quency response) and the timing of the compo-nents must be kept aligned (zero group delayerror). If a system has group delay errors, somefrequency components arrive sooner than others.The reconstructed output is distorted becausecomponents from different instants in time areadded together instead of those that belongtogether. Hence, distortion can result even thoughfrequency response is perfect. Group delay errorscan blur edge transitions and create ringing-likeartifacts.

Multipulse

Multipulse: Group delay and frequency response areboth tested with this pattern.

The multipulse signal is useful for checkingphase response or group delay as well as fre-quency response. Like the multiburst signal,multipulse signals include packets of sinusoidalwaves at various frequencies. However, unlikemultiburst patterns, the waves are within sinesquared (T 25 or T 12.5) pulse envelopes insteadof rectangular envelopes. The lower and upperfrequency components of the test signal are phasealigned so one can test whether the upper andlower frequency components are delayed relativeto each other in a display system. To make that

task easier, the amplitudes of the sine waves aregenerated such that their bottom excursion tipsperfectly align with the pulse baseline. Bylooking at how well the waveform tips maintainposition on the pulse baseline, one can discernboth group delay errors and frequency responseproblems.

Because waveform shape is absolutely critical forinterpreting multipulse patterns, the encodedwaveform must be free of intrinsic distortions.Otherwise, one cannot tell if observed distortionsare due to the pattern or the equipment beingtested. Recall that the traditional frequencies usedin multiburst patterns cannot be completelyrepresented on NTSC DVD. That same problemwould affect multipulse patterns if traditionalfrequencies were naively used or an existingmultipulse pattern was captured from a standardNTSC signal generator. Frequency and samplinginterval distortions would render such a mul-tipulse pattern much less useful. Avia PROmultipulse patterns were generated using fre-quency pulses that are cleanly encodable onNTSC DVD.

Waveform of Multipulse: Alignment of lower waveformtips exactly on baseline of the pulse baseline indicatesflat frequency response and absence of group delayerror.

Examination of waveform excursions within eachpulse reveals problems in frequency responseand/or group delay. For instance, if high fre-quency components are attenuated relative tolower frequency components, the wave tips nolonger reach baseline. If high frequency peakingis present, the waveform tips may exceed the

Multipulse showing high frequency loss.Waveform tips fall short of pulse base-line. Note that distortion is symmetric andcentered indicating no group delay error.

pulse baseline. Group delay errors appear assinusoidal distortion of the baseline because theamplitude changes of the high frequency sinewaves no longer coincide with pulse envelopeamplitude.

Multipulse patterns in Avia PRO include 3 dBand 6 dB attenuation markers as well as blackand white level flags. The first pulse is a T2 pulsethat is useful for scoped assessment of impulseresponse. Inverted multipulse is also provided toverify signal linearity.

Inverted Multipulse: Signal levels are inverted to helpverify response linearity.

Luma & Chroma Timing DelayLuma and color component signals travel throughseparate processing in many video systems. Thediffering paths may result in luma and colorinformation not arriving simultaneously. Indeed,the error can even originate at the DVD player.Y/C delay errors create artifacts at color transi-tions. If severe, visible displacement of colorinformation can be seen. If Y/C delay errors are

large, visual inspection of the display can detectthe delay. Smaller delay errors require an oscillo-scope for measurement. Avia PRO provides bothvisual, pixel based and instrument based tests ofluma/color component delay

The first feature built into Avia PRO for detectingluma and chroma delay errors is in the color barpatterns. Normally, color bars are limited in therate at which they transition from one color toanother. This is to remain within over the airtransmission limits that are more restrictive thanon-DVD capability. Avia PRO is a DVD basedtest set never intended for transmitter alignment.An FCC compliant signal generator with moreprecise output levels than a DVD player shouldbe used for that purpose, never a DVD player. Onthe other hand, Avia PRO is intended for evalu-ating playback chains from DVD to display. Itmade good sense to use the higher than broadcastsignal resolution possible on DVD to providemore rigorous test signals. Consequently, thecolor edge transitions in Avia PRO color bars areextra sharp and pixel aligned with MPEG2 colorsubsampling positions. Examine color bar edgetransitions and see if each color cleanly becomesthe next. If errors are large, you may see lumaand chroma transitions displaced relative to eachother.

Y/C Delay Chart

Y/C Delay Chart allows estimation and detection of Y/Cdelay errors without instrumentation. Cr and Cb compo-nent delay as well as RGB testing are tested.

The Y/C delay chart allows measurement if delayerrors are one pixel or greater. This is sometimesreferred to as a pixel based Y/C delay test. Usethe chart by examining the columns of colorstrips and finding the strip whose left edgehorizontally best aligns with the accompanyinggray strip. Nanoseconds of color delay areindicated next to the strip. Delays of Cr and Cb aswell as red, green, and blue can be read. The Crand Cb signals in this chart are RGB legal. Thechart can be used even if the signal has beenconverted into RGB color space.

Multipulse waveform showinggroup delay error in addition to highfrequency loss. Sine shapeddistortion of tip/baseline alignmentindicates that group delay error ispresent.

In the rightmost portion of the chart are redstripes in a yellow background. Red and yelloware two colors that differ greatly in both luma andchroma values. Transitions between them requirelarge changes in both luma and chroma. Y/Cdelay errors horizontally separate the two typesof transitions. That causes yellow to red transi-tions to appear different from red to yellowtransitions. Compare the left and right edges ofthe red stripes. If Y/C delay is present, the leftand right edge transitions of the red stripes willdiffer from each other.

Bowtie Patterns

Bowtie (500 KHz, RGB Legal): Luma/chroma delayerrors smaller than one pixel can be measured on anoscilloscope using bowtie patterns. Viewed on a normaldisplay this pattern yields little information.

Viewed on a scope, bowties reveal timing and amplitudepreservation. Note the null centered in middle of patternindicating zero delay. Markers indicate every 10, 50 and100 ns. of shift.

Pixel based Y/C delay charts can only measuredelays of one pixel (74 ns) or greater. Moreprecise testing of Y/C delay errors requires use ofa bowtie pattern and oscilloscope. Bowtie pat-terns work by providing two sine waves thatslightly differ in frequency. The waves areprecisely phase aligned at the center of the testpattern. The small frequency difference betweenthe two waves causes a phase difference betweenthe waves that increases as one moves away frompattern center. Subtracting the two sine wavesfrom each other and displaying the result on an

oscilloscope reveals a null point where the twowaves are exactly in phase. The resultant wave-form is shaped similar to a bowtie, hence thename of the pattern. If time alignment is perfect,the central null of the subtracted waveforms ispositioned at the center of the pattern. If onesignal is delayed relative to the other, the nullshifts left or right.

Chroma delays smaller than one pixel are easilymeasured using bowtie patterns. In this example usingan Avia PRO 1 MHz bowtie (double resolution), a 1/2pixel chroma delay creates a rightward, 37 ns. shift ofthe null.

Bowtie patterns test not only timing delay butalso relative signal amplitude and quality. If onesignal is attenuated, the null thickens verticallyon scope. If noise is present, the null becomesless sharp.

Bowtie patterns encode one sine wave in theluma channel and the other wave in the colordifference signals. To use the pattern, the lumasignal is fed into one input of an oscilloscope andthe color component signal to be tested is fed intoa second channel of the scope. The scope is set todisplay the two channels subtracted from eachother.

Bowtie (1 MHz, RGB Legal): Higher frequency basedbowtie magnifies delay errors.

1 MHz Bowtie doubles time resolution of bowtie test.Bowtie is also more distinctly visible.

Avia PRO Bowtie FeaturesAvia PRO Bowtie patterns contain featuresunavailable from previous test signal sources.

RGB Legal: The bowtie patterns in Avia PROare RGB legal. Conventional bowtie patternsseverely distort and become unusable if con-verted into RGB color space. The bowties in AviaPRO are RGB legal and remain valid even ifconverted from component color space into RGBspace and back to component space.

Delay Markers: Avia PRO makes reading oftime delays easy even if the oscilloscope lackselectronic calipers. The bowties include markersat every 10 nanosecond shift in addition to every50 and 100 ns. The three types of markers differin height and also flip polarity to indicate Cr vs.Cb testing. This compares favorably to olderpatterns with only monotonic marks every 20 ns.

Higher Resolution: 1 MHz based bowtiesdouble the measurable time resolution comparedto conventional 500 KHz based bowties. Finerreadout of delay errors is now possible.

Inverse Chroma Bowties: Bowties normallyrequire a dual input oscilloscope with channelsubtraction capability. Avia PRO allows bowtieanalysis even if only a single channel scope isavailable. Bowties with inverted color compo-nents allow luma and color difference signals tobe added with a simple resistor network anddisplayed on a single channel oscilloscope.

Chroma Delay Selectable: Some bowtie patternson DVD appear to have encoded their colorcomponents with an inherent 1/2 pixel delay(right shift) instead of encoding their colorcomponent signals time aligned with luma. AviaPRO provides both time aligned and 1/2 pixelchroma delayed versions of bowties in case theuser needs to emulate the behavior of bowtieswith an intrinsic chroma delay.

Levels, Grayscale, andGammaBlack level, white level, grayscale tracking, andgamma response are underpinnings of imagereproduction. If these are improperly adjusted,highlight and shadow details may be obscured,and midrange contrast can be too bright or dark.Avia PRO has families of signals to help thecalibrator achieve proper black level, white level,grayscale tracking, and gamma response.

Needle Pulses

A commonly used signal for setting white level isthe Needle Pulse pattern. Avia PRO expands thefunctionality of needle pulse patterns by addingfeatures to verify black level, test for white levelclipping, evaluate below black foot room, testabove white headroom, visually check grayscaletracking, and verify gamma response. A secondpulse at the right side of the patterns helps detectgeometry issues that may not be apparent withjust a single pulse. Hence, the patterns in AviaPRO are named Needle Pulses instead of NeedlePulse. A family of patterns provides a choice ofplain or more full featured versions of NeedlePulses.

Needle Pulses + Log Scale + Gamma pattern is part ofa family of patterns used to set white level, check blacklevel, verify gamma response, and test grayscaletracking. (Black falsely elevated for illustrative purposes)

Black Level Bars: Traditional PLUGE patternsrelied upon equipment passing below blacksignals. Today's digital processing equipment andDVD players may not necessarily pass blackerthan black signals and render older PLUGEfeatures less useful. Avia Guide to Home Theaterwas the first to pioneer animated black level barsallowing indication of black level even if belowblack signals are not passed by equipment. AviaPRO adds a third indicator stripe allowing preciseindication on both equipment with and withoutblacker than black signal handling.

Avia PRO features three animated stripes in itsblack level bars. These black level bars alsoappear in window patterns and some color bars.The leftmost stripe is 4 IRE below black andserves the same purpose as the below black stripein conventional PLUGE signals. This blackerthan black bar may not be visible during adjust-ment of equipment that does not pass blackerthan black signals. The middle black level bar isa scant 1 IRE above black. The rightmost blacklevel bar is 2 IRE above black. The 1 and 2 IREabove black bars are very close to black. Thisallows finer indication of proper black level thanpossible with stripes of higher amplitude. If adisplay's black level is set just 1 IRE too low, themiddle bar (1 IRE above black bar) becomesinvisible. If display is set 2 IRE too dark, therightmost (2 IRE above black) bar disappears.Avia PRO animates its black level bars makingthem easier to see during calibration.

Adjust black level to make the blacker than blackbar the same black as the black background whilestill allowing the 1 IRE and 2 IRE above blackbars to be visible. Equipment with poor shadowrendering or insufficient bit depth may havedifficulty achieving both a black backgroundwhile still displaying the 1 IRE above black bar.Other test patterns specifically for setting blacklevel are discussed later. Those are useful forsetting black level with respect to varying aver-age picture level.

Waveform of Needle Pulses shows crossed stepsrunning above white and below black as well as ani-mated black level and white level bars.

White Level Bars: Similar to black level bars,Avia PRO includes animated indicators for whitelevel. These are particularly useful with digitaldisplays and processors that may clip signallevels more severely than older analog systems.The leftmost white level bar is 2 IRE belowwhite. The middle bar is 1 IRE below white, andthe rightmost bar is 1 IRE above white. Use ofAvia PRO's white level bars helps avoid thecommon problem of near white detail clipping ondigital displays. Adjust white level to keep at

least the 1 and 2 IRE below white bars visible.The above white bar can clip without endanger-ing details of properly mastered material. If eitherof the 1 or 2 IRE below white bars disappears,the system is clipping image details that are nearwhite. Don't assume that lack of clipping issufficient to ensure white level is correct on adigital or plasma display. Also check grayscaletracking with the log steps as discussed later.

Needle Pulses: Just as with conventional NeedlePulse patterns, the pulses in Avia PRO can beused on CRT displays to check geometry stabilityacross APL changes. SVM presence can be seenas a difference in width of the white and blackportions of the pulses.

Crossed Steps: Digital video internally encodesluma black internally at level 16 and white andlevel 235. Digital signal levels from 1 to 15 arereserved for below black foot room. Levels 236to 254 are for above white headroom. Levels 0and 255 are reserved. Proper mastering of mate-rial for DVD places black in the picture at digital16 and white at digital 235. Unfortunately, not alldiscs are carefully mastered. Some include lumainformation outside the digital 16 to 235 range.Because of the occasional poorly masteredrecording, an ability to process signals in thebelow black and the above white regions can beuseful.

Shallow, crossed steps above white and belowblack signal test the ability of a system to handlesignals above and below normal range. The darksteps in the black background cross at black. Thebrighter steps in the white background cross atwhite. Depending on whether a display hasenough headroom, you may or may not see thesteps that go above white. A display with prop-erly adjusted black level shouldn't show theblacker than black portion of crossed steps.Viewed on a waveform monitor, the steps indi-cate whether signal processing clips or includesabove white and below black details.

Log Steps: The central log steps provide visualtesting of grayscale tracking. The steps exponen-tially increase in intensity, but should remain thesame color of gray if grayscale tracking is cor-rect. If grayscale tracking is poor, the color ofgray may differ when you compare steps ofdifferent intensity. This pattern feature is par-ticularly useful while setting white level ondigital or plasma displays. On some displays,grayscale tracking becomes poor even beforewhite level is raised to the point of white levelclipping. On displays with such limitations,contrast may need to be kept below the point at

which grayscale tracking deteriorates. Look forthis by observing the log steps while adjustingdisplay contrast control. If a color shift in thesteps or clipping of the white level bars occur,white level has been set too high.

Gamma Check Steps: At the left side of the LogSteps, some of the Needle Pulses include gammacheck steps for visual verification of gammaresponse. The average intensity of the lines ineach gamma step equals the brightness of logstep immediately to its right. The gamma stepsand log steps should change in intensity togetheras they run from light to dark if the displaymatches CRT gamma response.

Some caution should be taken while interpretinggamma steps. Scalers often alter the brightness offine lines. This is particularly true if a scaler doesnot gamma correct the video signal back into alinear representation of light intensity prior toscaling. Such a scaler would then need to re-gamma the scaled signal. Recall that the NTSCvideo signal is not linear to light intensity but isinstead one which includes the inverse gammafunction. Scaling the inverse gamma relatedsignals instead of a linear representation createsgamma induced errors. These appear as falsebrightening or darkening of small details and asedge artifacts. Unfortunately, scalers in consumergrade equipment usually do not convert the videosignal into a linear representation prior to scaling.If such is the case, the gamma steps will notmatch the brightness of the accompanying logsteps. You can still use the steps but look forthem to darken and lighten in ratio with theaccompanying log steps rather than identicallymatch the log steps in brightness.

Black Level PatternsImperfect DC restoration or light scatter in adisplay sometimes confounds black level adjust-ments. Avia PRO provides a family of BlackLevel patterns with selectable APL and content.

Black Level Bars + Log Steps + Gamma Steps: Simul-taneous black level, white level, as well as grayscaletracking can be tested with this low APL pattern.

Waveform showing animated black level bars and whitelevel bars analogous to those in Needle Pulses.

Animated black level bars throughout Avia PROare all encoded with the same trio of levels. Theleftmost black level bar is 4 IRE below black.The middle black level bar is 1 IRE above black,and the rightmost black level bar is 2 IRE aboveblack. Although other patterns in Avia PROcontain the same black level bars, the BlackLevel Bar family of patterns is optimized forsetting black level.

Black Level Bars + Variable Half: Variable APL loadingof display tests black level stability. (Black level elevatedand white decreased in figure to illustrate normallyinvisible blacker than black and whiter than white bars)

If variable intensity right half is 90 IRE or higher, whitelevel bars at 2 & 1 IRE below white and 1 IRE abovewhite are added to help detect clipping.

DC restoration is rarely perfect in analog proc-essing chains. For this reason, a compromiseblack level must often be set at intermediateaverage picture intensity. Actual APL selectedwill vary with display behavior. Avia PRO allowsthe user to adjust pattern APL to a level appropri-

ate for the display while setting black level. TheBlack Level Bars + Variable Half pattern's righthalf of the pattern can be varied from black towhite using the play, pause, and reverse controlsof the DVD player. If a 90 IRE or brighter righthalf is chosen, Avia PRO's standard white levelbars are added to the test pattern. These are at theusual 2 IRE below white, 1 IRE below white and1 IRE above white levels.

Digital displays typically retain black level betterthan CRT displays, but even if a display hasperfect DC restoration, variable APL black levelpatterns are still beneficial. Selecting a low APLcan make the black level bars easier to see byreducing light scatter.

Black Level Bars + Log Steps + Gamma Strip + VariableAPL: This pattern adds a central grayscale calibrationtarget.

Some calibrators may wish to verify grayscalecalibration simultaneously with gamma testingand both black level and white level checks. Thisunique pattern combines multiple test featuresalong with grayscale measurement. The grayscalemeter should have limited acceptance angle touse this pattern.

Chip ChartThis electronic equivalent of a camera view of alog reflectance chip chart allows visual checks ofgrayscale tracking as well as black and whitelevel. Such crossed step patterns were less usefulwith CRT projection because of that technology'spropensity to color shift between one side of the

screen to the other. Side to side color shiftobfuscated observations of grayscale tracking somuch that step patterns fell out of favor in homecinema setup. Digital projection systems, withtheir better side to side color uniformity, oncemore allow chip chart usage.

Chip Chart: Crossed steps test grayscale while indica-tors for shadow detail and white level clipping are incenter of pattern. The "chips" or steps should vary inintensity but not in color.

Waveform of Chip Chart: Note gradated stripes of nearblack and near white stripes.

Avia PRO's Chip Chart has an 18% brightnessbackground not 18% signal intensity. Crossedsteps run from 0% (black) to 100% (white) signalintensity. In the center are a black band and awhite band. Within these bands, stripes test nearblack and near white performance. Unlike con-ventional "zebra stripes" which are of only asingle intensity, the stripes in Avia PRO vary inintensity from left to right. This allows finertesting of shadow and highlight detail perform-ance. The left most stripe is dimmest in eachseries of five stripes. The near black stripes are at1, 2, 3, 4, and 5% intensity. The near whitestripes are at 95, 96, 97, 98, and 99% signalintensity. On a display with good shadow andhighlight rendering, all the near black and nearwhite stripes should be visible.

Step Patterns

Horizontal 5 IRE Step Pattern: Avia PRO includes steppatterns in 5 IRE and 10 IRE increments from black towhite as well as steps from below black to above white.

Scope tracing of horizontal 5 IRE steps should showlinear response as in this waveform. The first step from7.5 to 10 IRE is intentionally smaller than the remainingsteps.

Step patterns in both horizontal and verticalorientation are provided. Most test signal sourcesonly provide 10 IRE increment steps, but AviaPRO also provides 5 IRE steps. The finer grada-tions are particularly useful while visuallychecking digital system grayscale tracking.Crossed versions of steps are useful for visualtesting because they juxtapose differing intensitygrays making comparison of gray colors easier.

Avia PRO's windowboxed steps permit ends of stair-cases to be observed even on displays with moderateoverscan.

High Monotonicity StepsDigital video encodes luma values with black atlevel 16 and white at level 235. The values 0 and255 are reserved. In order to accommodateovershoot and imperfect mastering, below blackfoot room and above white headroom are allo-cated digital levels 1 to 15 and 236 to 255 re-spectively.

Ideally, a display allows differentiation of eachdigital value. If the display is digital, it must alsohave sufficient internal bit depth to represent allluma values 16 to 235 after applying its gammatable. Insufficient bit depth will cause somesignal levels to be indistinguishable from othersand create banding, contouring, and clippingartifacts.

Normal step patterns, even those in 5 IRE incre-ments are neither fine enough in level change normonotonic enough to test preservation of digitalvalues. Avia PRO includes tests specifically forchecking monotonicity and bit depth adequacy.Five High Monotonicity Step patterns:

Steps 16 to 235 by 4Steps 16 to 235 by 2Steps 16 to 235 by 1Steps 1 to 254 by 1Steps 0 to 255 by 1

The digital luma range for each pattern is givenin its name. The last number in the pattern nameis how much, in digital value, the luma valueincreases with each step. Patterns, which runfrom digital 16 to 235, test the range from trueblack to true white. The 1 to 254 pattern tests theentire legal digital range including foot and headroom. The final pattern includes steps at theillegal but still encodable values of 0 and 255.DVD players that do not pass blacker than blacksignals will clip all values below 16 and causeloss of the dark end of the latter two highmonotonicity step patterns.

Standards compliant material encodes all in-tended visible picture information within the 16to 235 (true black to white) digital range. How-ever, mastering is not always perfect. Some DVDreleases have white values well above the stan-dard 235. Such non-compliant discs look brighterbut also induce severe highlight clipping on manydisplays.

Waveform of High Monotonicity Steps 0 to 254 by 1:Digital luma values from darkest below black to brightestabove white are encoded in this pattern. This exceedsnormal range for picture information and includes footroom and headroom testing.

High Monotonicity steps are best used by firstverifying that the DVD player and processor arepreserving each step value. This can be done withan oscilloscope or waveform monitor. Withoutscope verification, observed problems cannot beproperly attributed to the player/processor vs.display. Once DAC monotonicity and all stepvalues are confirmed to be present, visual obser-vation of a display reveals whether it has suffi-cient bit depth to render all luma values. Insuffi-cient bit depth on digital displays causes someadjacent steps to become indistinguishable.Analog displays do not exhibit this problem.

Ramp Patterns

Ramp patterns check amplifier linearity and grayscale tracking more continuously than steppatterns. Both conventional and highly mono-tonic "deep" ramps are provided in Avia PRO.Conventional ramps run nearly the width of thevideo frame and range from true black to white(digital 16 to 235). This corresponds to 7.5 to 100IRE if North American setup is present.

Single ramps, crossed ramps, and double-crossedramps are supplied in both vertical and horizontalorientation. The ramps were intentionally win-dowboxed to allow their entire extents to bevisible on most displays. Older, non-windowboxed ramps can have their ends cut offby overscan or DVD player pixel cropping. Theramps are not dithered, and cleanly delineateeach luma level. However, because the conven-tional ramps have a pixel width that is not inte-grally related to the number of signal levels,some periodic banding will be seen on screen.Dithering the ramps could have reduced thisbanding, but that would reduce smear signallevels. Because of this normal banding, we do notrecommend the conventional ramps for visual

tests of display banding. Instead use Avia PRO's"deep" ramps, which are inherently not banded.

Horizontal Ramp: Conventional, undithered ramp fromblack to white is windowboxed to remain on screendespite display overscan. Slight, periodic, on screenbanding of this pattern is normal due to the non-monotonic width that each intensity level occupiesacross the ramp.

Waveform of Horizontal Ramp: Conventional, undith-ered ramp from black to white should be a straight lineon scope if video amplifier response is linear.

Horizontal XX-Ramp: Two crossed ramps meet eachother in the center of the pattern. The sharp centraldemarcations at white in the top half and black in thebottom half become blunted if clipping is present.

Waveform of Horizontal XX-Ramp: Inflections at whiteand black in center of pattern simplifies testing forclipping.

Deep RampsDeep ramps have several advantages over con-ventional ramps. First, the width of the rampswas specially chosen to make each digital signallevel occupy exactly the same width on the ramp.This means the ramps are intrinsically bandingfree without resorting to dithering or otherfiltering. Second, the range of the "deep" rampsruns from deepest below black to highest abovewhite. Hence, these ramps are called "deep"ramps.

Deep Horizontal Ramp: The left rectangular area is trueblack (digital 16). The right rectangular area is true white(digital 235). Ramp runs from digital 1 up to digital 254.On displays with adequate bit depth, Avia PRO DeepRamps should exhibit little or no banding. (Image levelsin this illustration were altered to display blacker thanblack and whiter than white details.)

Waveform of Deep Horizontal Ramp shows plateaus atblack and white at each end of the pattern. Ramp runsfrom below black to above white.

Deep ramps are supplied only in the horizontalorientation, but single, crossed and doublecrossed versions are all provided. Viewed on anoscilloscope these deep ramps test extent andlinearity of the entire signal range. The ramps arehighly monotonic and should show little or nobanding on screen. If banding is visible usingdeep ramps, the display or processing system isfailing to represent some luma values.

Deep Horizontal Double Cross Ramp: Beyond conven-tional ramps, this deep ramp visually tests for bandingand loss of headroom and foot room signal levels.

The signal peaks and valleys of deep ramps indicatehow the entire signal range is preserved. Here, asystem's ability to handle the full foot room and head-room range (digital 1 to 255) is evidenced by thewaveform's sharp upper and lower inflections.

Modulated Steps and Ramps

Modulated ramps and steps contain a constant ampli-tude and phase chroma signal with luma staircase orramp. Viewed with a waveform monitor/vectorscope,these patterns test how well chroma phase and ampli-tude are preserved as luma changes.

Although not categorized as grayscale or gammatests, modulated steps and ramps are discussedhere because they combine aspects of grayscaleramps and steps with a constant chroma signal.Displayed on a screen, these patterns look likeyellow-green tinted versions of ramps or steps.Little information can be gained by visuallyinspecting them. Their utility is in instrumentedchroma signal analysis.

Modulated ramps and steps are used for testingdifferential gain and phase. This is done onwaveform monitors and vectorscopes. Scopeusage is beyond the scope of this chapter, but onwell functioning systems, the chroma amplitudeand phase should not vary as luma varies. Ifdifferential gain or phase errors are present, colorfidelity suffers.

Avia PRO includes modulated steps and rampsencoded with 40 IRE amplitude chroma signalson the negative Cb axis. Luma values in modu-lated ramps and steps are selectable betweenstandard black to white range (digital 16 to 235)or "deep" range (digital 1 to 254). The deepversions are windowboxed rather than horizon-tally sized for high monotonicity. Hence, somevisual banding in the ramps may be seen. Thisbanding is of no consequence in the instrumentedanalysis for which these patterns are intended.

Full range modulated ramps and step patterns canonly exist in the component color space. Theyinclude values that are not RGB legal. Avia PROsupplies full range and the even wider range"deep" versions of these patterns. Unlike theRGB legal chroma sweeps and bowtie patternselsewhere in Avia PRO, the modulated steps andmodulate ramps must never be converted intoRGB color space. Doing so would severelydistort results.

Window PatternsGrayscale calibration is an important and oftenperformed task. Avia PRO provides new tools tomake the process faster and easier. Windowpatterns in Avia PRO enable black level verifica-tion, white level clipping avoidance, and auto-mated grayscale measurement. Windows areavailable in increments as fine as 2.5 IRE andrange from below black to above white. Profes-sional calibrators now gain the ability to selecttheir high and low intensity windows rather thanaccept the values supplied on other discs or suffercumbersome navigation between two valuewindows. Avia PRO includes windows in "sets"which provide easy selection and rapid naviga-tion between desired IRE window levels.

Gray Windows: Animated black level bars, edgetransition tests, and IRE level labels are integrated intoeach window pattern. (Black level falsely elevated forillustrative purposes)

Waveform of Window Pattern: Black level bars are at 4IRE below black, 1 IRE above black, and 2 IRE aboveblack.

Higher IRE level window patterns include Avia Pro'sthree white level bars to indicate clipping.

Waveform of High IRE Window Pattern: Windows at 90IRE or higher include white level bars at 2 IRE belowwhite, 1 IRE below white, and 1 IRE above white.

Previous calibration DVD’s supplied windowpatterns in much more limited configurations.Avia PRO supplies grayscale calibration windowsin three manual sequences, three automatedsequences, and user selectable quadruplet sets.

Manual Sequences: Windows are available inascending order in 2.5 IRE, 5 IRE, or 10 IREincrements. The 2.5 IRE increment sequencestarts below black and extends above white. Eachwindow displays indefinitely until the "nextchapter" button is pressed on the DVD remote.Remote control "return" or "previous chapter"buttons step backwards in the sequence. "Menu"returns to the grayscale sequence menu. Individ-ual windows are selectable from the 5 and 10 IREmenus, but that would require too many menuitems in the 2.5 IRE sequence. For the 2.5 IREsequence, every other window is directly acces-sible. The window intensities between those inthe menu can be accessed with a subsequentpress of the "next chapter" button.

Note: The DVD specification limits the totalnumber of items permitted on a DVD menu. Thislimit is higher for 4:3 aspect ratio menus than for16:9 menus. For this reason, the disc intentionallyuses 4:3 menus when a large number of choicesare listed.

Automated Sequences: Grayscale calibrationreport generation can be tedious. Avia PROincludes automated sequences of grayscalewindows for use with computer controlledmeasurement equipment. Each window of thesequence is displayed for 5 seconds beforeadvancing to the next window. Just before eachwindow appears, a synchronization tone triggersthe sensing equipment to read a sample. Auto-mated sequences are supplied in 2.5 IRE, 5 IRE,or 10 IRE increments.

Calibration Sets: The optimum high and lowwindow values for calibration are not the samefor every type of display and instrument. AVIA

PRO makes it easy to select and use the appropri-ate intensities instead of accepting limitedchoices on disc or working through slow naviga-tion. Calibration sets are quadruplets of fourpatterns that form self contained loops. Eachquadruplet contains a high and a low window aswell as a middle window and black level screen.The middle window is used to verify middlegrayscale tracking as the ends are adjusted. Theblack level screen serves as an additional checkof black level correctness. The window patternsthemselves contain black level bars and steps toallow continual monitoring of black level duringgray scale calibration.

Window quadruplets give professional calibratoreasy and rapid navigation between desired highand low windows as well as black and middlepatterns to check progress. Quadruplet combina-tions were designed to cover a wide range ofconditions. They have the following combina-tions of IRE windows + black:

40 / 70 / 100 / black40 / 65 / 90 / black40 / 60 / 80 / black

30 / 65 / 100 / black30 / 60 / 90 / black30 / 55 / 80 / black

20 / 65 / 100 / black20 / 60 / 90 / black20 / 50 / 80 / black

Grayscale calibration is faster, more convenient,and flexible with Avia PRO.

Edge Transitions: Three types of horizontaledge transitions are present in each windowpattern. These are useful for testing systemresponse for ringing, bandwidth, and group delay.Transitions in the upper third of each window areinstantaneous. Transitions in the middle third areT2-Steps with a critical frequency of 6.75 MHz(NTSC DVD limit). Transitions in the lower thirdare T2-Steps with a critical frequency of 4.0 MHz(NTSC broadcast limit).

The upper third, instantaneous transitions arevery severe tests of system frequency responseand timing. Very few display systems other thanhome theater PC's can reproduce transitions ofthe upper third without ringing. The middle thirdtransitions are less severe and should be cleanlyreproducible by higher performance homecinema systems with sharpness correctly set. Thebottom third transitions are significantly fre-quency limited and should be cleanly reproduci-

ble on most video systems when sharpness iscorrectly set.

Gamma

Simply making red, green, and blue primariestrack together at D65 was once sufficient toestablish a good foundation for imaging. Today’sdigital displays add another aspect to be consid-ered – gamma response. The NTSC video systememploys non-linear video signals. The overallsystem (camera through display) response isalmost linear, but CRT displays inherentlyrespond to input voltage with an exponential lightoutput. The relationship is roughly:

Light output = constant * input^ gamma

Gamma is approximately 2.3 to 2.4 for real CRTdisplays. There is actually a linear tail region nearblack and gamma that is taken as 2.2 in formaldefinitions, but the above is sufficiently accuratefor the purposes of this discussion.

To make overall system response linear, thecreators of the NTSC system had to choosebetween requiring prohibitively expensivegamma correction circuitry in every television vs.making the signals pre-compensate for thetelevisions. The choice made was to pre-compensate at the source. NTSC cameras andother signal sources apply an inverse gamma of0.45 (1/2.2) to the video signal. That inversegamma encoded signal then yields a near-linearoverall system when played back through dis-plays with proper gamma response. Becauseactual CRT gamma is slightly higher than 2.2,overall system gamma is slightly higher thanunity. The slight, overall positive gamma con-veniently compensates for dark surround effect.

Proper gamma response automatically happenson CRT displays by virtue of electron gunphysics. Newer, digital display technologiesdon’t inherently have this response, but must stilldeal with display video signals encoded with a1/2.2 inverse gamma. Proper gamma response isaccomplished in such displays by processing theincoming signal through gamma tables. Thesegamma tables convert input levels into internallevels that yield the desired gamma response.

Many digital displays have selectable or adjust-able gamma tables and curves. Factory defaultgamma tables do not always emulate standardCRT response. Non-standard tables can dramati-cally alter image appearance, but not always tobeneficial effect. Just like excessively blue

grayscale, gamma tables that boost middle andhigh range contrast at the expense of shadow andhighlight detail are common. If one wishes toachieve an accurate and pleasing image, gammaresponse must also be adjusted to better matchstandard gamma response – the response forwhich NTSC material is authored to look best.

Gamma response is most accurately and thor-oughly characterized by measuring output atmultiple IRE levels and plotting the curve. AviaPRO’s automated grayscale window sequencesmake that task easier, but a quick means ofestimating gamma response is also desirable.Gamma tables allow rough estimation of gammaresponse by comparing the light output of 50%area coverage white vs. the level of gray neededto match that intensity. White remains the sameintensity no matter the display gamma. Patternsin which white covers 50% of the area, such asevery other line patterns, have half the lightintensity of solid white. Gamma charts rely onthat relationship in order to work. By comparingthat non-varying 50% light against patches ofgray (which are affected by gamma), the intensityof signal need to produce a matching gray can berelated to screen gamma at the 1/2 intensity point.

Gamma Basic Chart: The gray patch whose brightnessbest matches that of the background indicates displaygamma.

Use basic gamma charts by defocusing your eyesand finding the gray patch which best matchesthe background. Avia PRO includes three basicgamma patterns, all of which have coarse 50%coverage white background patterns. The back-grounds are intentionally coarse to help mitigatefactors that could change the brightness of 50%coverage white. Frequency response problemsand scalers can alter the brightness of the back-grounds, particularly if the background includesfine lines or details. The coarser backgroundbasic gamma patterns should be used first be-cause they are the most immune to scaling andfrequency response effects.

Tri-Background Gamma

Tri-Background Gamma: Finer gamma estimation ispossible with this series of patterns.

The Tri-background Gamma pattern series isdesigned to measure gamma in finer incrementsthan the basic gamma patterns. Step through thepatterns until you find the best match between thegray patch and background. Three types ofbackgrounds are supplied. All are 50% coveragewhite lines, but they differ in coarseness. Thefinest is at the left. The coarsest is at the right.Gray patches across a row are all the sameintensity. If you see a difference in brightnessbetween the backgrounds, assume the coarserbackground is most accurate and compare thepatches against that background. The threebackgrounds not only provide partial immunity toscaling and frequency response effects, but alsoindicate when scaling and frequency response arealtering the appearance of fine image features.

Gamma Tracking Pattern

Gamma tracking pattern determines display gamma andverifies tracking against log steps.

If testing with the Tri-background gamma se-quence shows that at least the middle backgroundis the same intensity as the rightmost back-ground, then the Gamma Tracking pattern can beused. This pattern has relatively fine features andis the most susceptible to scaler and frequencyresponse effects. The gamma steps and log stepsof this series will track together when the patterngamma matches screen gamma.

Again caution is warranted if a scaler is presentwhich does not convert signals to linear spaceprior to scaling. If results are confusing, it is bestto avoid gamma charts and actually measureresponse using window pattern series. Measure-ment and plotting of window pattern intensitiesworks accurately even if scaling and frequencyresponse problems are present.

GeometryGeometry correction has always been an exactingpart of CRT projection setup. Avia PRO containsflexible, high resolution geometry and conver-gence tests to assist the process.

Checkerboards

3x3 Checkerboard 6.75MHz T2 step edged checker-board pattern is one of several variations available forcontrast, geometry, and edge response testing.

Avia PRO checkerboard patterns extend theusefulness of checkboards beyond simple ge-ometry and contrast measurements.

Selectable Size: Checkerboards range from 1 x 1to 8 x 8, rather than the usual limited choices.Odd numbers of rectangles aid screen uniformitytesting by providing a central measurement point.

Edge Transition Testing: Three selectable edgetransitions allow evaluation of sharpness controlsand their effect on edge transitions ringing. Theedge transitions are instantaneous, 6.75 MHzbased T2 step and 4.0 MHz based T2 step edges.All display should be able to display the 4.0 MHzT2 step edges cleanly, but the other two types oftransition are more demanding. Very few dis-plays can reproduce the instantaneous edgetransitions cleanly.

Triplets: Each family of checkboards is arrangedin a triplet of patterns. Advancing forward via theDVD remote enables rotation among the threepatterns of a triplet. Each triplet is composed of aregular, a luma inverted, and an aiming targetversion of the pattern. The triplets speed displaycontrast measurement. Inverting checkerboardswithin each triplet means ANSI contrast styledata can be obtained without moving the lightsensor head. Formal ANSI contrast can also bemeasured in the conventional manner with 4 x 4checkerboards.

Crosshatch PatternsAvia PRO includes the most complete andadvanced tools for geometry setup and conver-gence. Rather than the usual limited set of ge-ometry patterns, Avia PRO allows user selectionof hatch spacing, style, intensity, coloration, andaspect ratio markers.

Hatch Pattern 2.35 in 1.78 Screen: Hatch pattern withmarkers indicating overscan and area of aspect ratiocontent.

The above crosshatch example shows somefeatures included in Avia PRO’s crosshatchpatterns.

Highest Resolution Possible on NTSC DVD:Designed for today’s high resolution, progressivescan displays, fine lined hatch patterns are at thevery single pixel limit of 480P NTSC DVDresolution. Grid thickness of only one pixelallows keen observation of convergence. Fordisplays incapable of using high-resolutionpatterns, two pixel thickness patterns are alsosupplied.

Centering Ticks: Tick marks precisely indicatethe center of the DVD pixel frame. Some othercalibration products improperly center testpatterns. Avia PRO maintains accurate alignmentwithin the 720 x 480 NTSC DVD pixel map.

Overscan Markers: Markers show overscanpercentage in 1% increments. Major area markersalong edges and corners indicate 5% and 10%overscan and assist with linearity and centering.Markers can be displayed or excluded as desiredby the calibrator. Both 1.33 and 1.78 aspect ratioscreens are supported. Within each screen type,aspect ratio and overscan markers for the fol-lowing are selectable:

a. Plain (No Aspect Markers)b. 1.33 Aspect Ratioc. 1.78 Aspect Ratiod. 1.85 Aspect Ratioe. 2.00 Aspect Ratiof. 2.35 Aspect Ratio

Safe Action and Safe Title Areas: Coincidingwith 10% and 5% overscan, these importantframing areas are indicated.

Selectable Grid Intensity & Colors: Crosshatchintensities of 40, 60, 80 and 100 IRE as well astinted and multicolor hatches are available.Tinted grids make red, green, and blue approxi-mately the same visual intensity so convergencecan be performed without turning off or cappingcolor guns. Multi-color grids vary in color andreveal small convergence errors as grid motion.

Adjustable Grid Spacing: Variable sizedhatches place grid intersections exactly whereneeded. Use DVD player play, pause, and reversecontrols to select the grid spacing desired.

Circle Hatch: Circles verify size and linearity adjust-ments of overall and corner geometry.

Diagonal Hatch: A useful pattern for setting linearity. Bysighting along diagonal lines and adjusting for straight-ness, calibrators can quickly adjust linearity withoutextra measurements.

Avia PRO provides calibrators a tool to fit the jobrather than forcing calibrators to compromisewith limited geometry pattern choices.

Checker Steps

3x3 Checker Steps Deep: Overall grayscale trackingand display contrast is seen with this pattern.

3x3 Checker Steps Shallow: Shadow and highlightdetail preservation are seen with shallow checker steps.

Checker Steps add grayscale tracking verificationas well as highlight and shadow detail preserva-tion to checkerboards. They are provided inmultiple rectangle counts and with either shallowor deep cross steps. All steps should be visible.The darkest near black step is above black. Thebrightest near white step is below white.

Color CalibrationOnce geometry, black and white level, grayscale,and gamma have been calibrated to form afoundation for the image, colorization is added tomake the final image. This section discusses thenew color adjustment tests available in AviaPRO.

Color Checkerboard

Color Checkerboard contains the same 75% colorvalues as in standard color bars, but is designed tosimplify visual matching of color patches. (Image alteredto show normally invisible blacker than black bar)

Visually guided adjustment of saturation, hue,and color decoder axes requires precise determi-nation of when color patches match each other inintensity. This pattern improves accuracy byplacing test colors in a checkerboard layout.When match is achieved the checkerboardbecomes a nearly uniform color when viewedwith the other two color channels shut off. Theupper half of the pattern indicates saturationcorrectness. The lower half indicates hue correct-ness.

Color filters can be used to selectively observelight of a single primary, but will almost alwaysallow some light of other channels to leakthrough. Stacking layers of filter material canboost filter selectivity, but doing so reduces totallight and that in itself reduces observer accuracy.Shutting off color channels provides the mostaccurate observation condition for precision colordecoder adjustment. Optical filters can be used ifcare is taken to compensate for their intrinsicallyimperfect selectivity.

Color Bars in Avia PRO

A wide variety of new and familiar color bars areavailable in Avia PRO. Many have features fortesting white level, black level, and color decoderaccuracy in addition to the functions of conven-

tional color bars. Blue color bars in the calibra-tion suites even include overscan markers.

Blue Bars WB: Color bars arranged for easiest blue-onlysaturation and hue adjustment. Windowboxing helpskeep important elements on screen. Gray reference barallows color decoder evaluation. Black and white levelsaltered in this image to illustrate blacker than black andwhiter than white level bars.

Red Bars WB: As pioneered in Avia Guide to HomeTheater, color bars are also rearranged to aid red-onlyor green-only based color adjustment.

Flashing Patches: Ovation Multimedia’s exclu-sive feature of flashing patches improves ob-server precision. Flickering of the patches isminimized when color match is achieved. Theprofessional calibrator is strongly encouraged tofurther improve observer precision by shuttingoff color channels rather than using filters.

White Level Bars: Avia PRO’s standard whitelevel bars (2 IRE below, 1 IRE below, and 1 IREabove white) are within the lower left whitepatch. Animation improves their visibility.

Black Level Bars: Bars at 4 IRE below, 1 IREabove, and 2 IRE above black in the lower rightof indicate proper black level.

Gray Reference Stripe: A 75% gray referencestripe across top of pattern allows convenienttesting for color decoding problems such as redpush.

4.0 MHz T2 Edge Transition: Horizontal edgetransitions of white patch are bandwidth limited.

Over sharpening should be suspected if ringing isseen at the left or right edges of the white patch.

Y/C Timing Testing: Color transitions betweencolors are sharp and aligned at MPEG colorsampling boundaries to permit evaluation of Y/Cdelay errors.

Color Bars Gray Middle, Variable Intensity: Variableintensity color bars with gray reference in middle ofpattern.

Variable Intensity: The calibrator and systemtester can now select intensities of color barsother than 75% or 100%. The DVD player play,forward, reverse and pause controls can be usedto select a color bar of desired intensity.

Color Decoder Check: Comparison of color patchintensity against gray background tests color decoderaccuracy in 5% accuracy.

Color Decoder Checks: Available in 10% or 5%precision, these allow quantification of colordecoder problems such as red push and greenpull.

Chroma Upscale Error Tests

CUE Test Red: Red and Cr version of Color UpscaleError test.

Chroma information on DVD is under sampledboth vertically and horizontally. During recon-struction of the image, a reconstructed chromavalue must be given to each pixel. In other words,the recorded chroma information is upscaled bythe DVD player. Unfortunately, many playersproduce an artifact during the chroma upscalingprocess. This is known as Chroma Upscale Error(CUE) and the origin of this problem is describedbelow.

Chroma information is correctly downsampledduring encoding using two different methods.One method is used for progressive scan mate-rial. The other is used for interlaced material. Inorder to avoid CUE, A DVD player must selectand use the chroma upscaling method that undoesthe method that was used during down sampling.Unfortunately, many DVD players only imple-ment one method– the method used for interlacedmaterial. If the original down sampling methodwas progressive and the interlaced method isused for upscaling, the resultant upscaled chromais corrupted. Most films on DVD are progressivematerial so this problem can frequently appear.

Other causes of chroma upscale errors also exist,but the cause described above is probably the oneof greatest interest because it affects criticallyviewed film material. On smaller displays, CUEartifacts are minimally visible, but on a large,high-resolution home cinema display, CUE canbe seen as jagged edges along colored diagonaledges and every other horizontal line colorintensity changes in gradients.

The CUE tests in Avia PRO are for detecting useof interlaced chroma upscaling during display ofprogressive material. This simulates the chromaupscale error that may be seen on film-basedmaterial.

Diamond Stacks: Columns of diamond shapeswith internal gradients allow visualization ofCUE induced jaggedness of diagonal edgetransitions. If CUE is present, the diagonal edgesnearly halve in resolution.

Chroma Sweeps: Some video processors areable to hide CUE because they under samplechroma. This filters out finer chroma details andas a side effect also hides CUE. Filtering ofchroma resolution can be detected with thevertical and horizontal chroma sweeps in thispattern.

Progress has been made regarding CUE. SomeDVD players now upscale chroma using thecorrect algorithm for each type of material andare free of CUE.

Vector Grid

Vector Grid 75%: Color patches form a linear grid onvectorscope if color processing is correctly.

Vectorscope of Vector Grid 75%: Sample colors areuniformly and linearly aligned on vectorscope. The 75%version of Vector Grid has colors limited within 75%primary and secondary color.

Vector grids are used with a vectorscope to verifythat color decoding is uniform at multiple satura-tions and phases. If chroma decoding is correctand linear, the sample points and trace lines forma linear grid on vectorscope. Apices at maximalred, green, blue, yellow, cyan, and magentacorrespond to the colors in color bars. Both 75%

and 100% intensity version are available. Thecolor patches in Vector Grid 100% are the samecolors found in the Avia PRO Gamut test series.

Gamut Test

Gamut Test: Automated, detailed characterization ofdisplay gamut is possible using this test and computercontrolled color measurement equipment.

The range of colors reproducible by a display isknown as its gamut. A wide gamut is desirable.However color accuracy requires not only a wideenough range of colors but also accurate repro-duction of each color specified in a video signal.Avia PRO includes the ability to perform exten-sive color accuracy testing across the displaygamut. For the manufacturer, this is a tool forensuring designs produce accurate coloration(flesh tones, greens, blues etc are all the correctshade of color). For the equipment reviewer, thegamut test provides an objective means of quanti-fying and comparing coloration errors in dis-plays. For the equipment installer, this is an aidfor recommending displays or compensating for agiven display’s color gamut.

Phosphor characteristics of CRT displays pro-duced reasonably acceptable color rendition ifgrayscale and color signal decoding were cor-rectly set. Today’s digital display devices canhave markedly non-standard primaries, gammacurves, and signal processing quirks. Despitebeing set up correctly using gray windows andcolor bars, displays with non-standard primariescan still produce an incorrectly colored picture.

The Gamut test is a series of colored test win-dows at the same colors as the Vector Grid 100%test pattern. Its 91 color samples test colorperformance through a wide color range ratherthan just primaries and secondaries. Each gamutwindow is displayed for 5 seconds before auto-matically advancing to the next sample. An audiosync tone cues measuring equipment to measureeach sample. After all samples are acquired, thedata can be analyzed to show display gamuterrors. Armed with this information, the calibra-

tor can adjust color decoding to more accuratelyproduce expected colors.

Pixel Cropping

Pixel Cropping: DVD Players often crop off severalpixels from video frame edges.

This pattern measures how much image a DVDplayer crops. This differs from overscan losses ona display. The maximum active line duration forNTSC video signals is not long enough to ac-commodate the entire 720-pixel frame width. Asa result, a DVD player's NTSC video outputtypically crops about 9 pixels from the encodedimage’s 720-pixel frame width. The degree ofcropping may vary from 6 to 20 pixels or more.This cropping may be on just one edge or takenfrom both left and right edges. Sometimes topand/or bottom pixels are also lost.

It is possible for a DVD system to display theentire 720 pixel wide image, but doing so re-quires circumvention of either NTSC timing orMPEG-2 bandwidth limits. For example, com-puter DVD playback systems can usually displaythe full 720 x 480 pixel frame on a computermonitor because the RGB video signals to themonitor need not follow NTSC timing limits.However, the NTSC video output from a normalDVD player must still crop pixels if it is bothNTSC and MPEG-2 compliant.

This Pixel Cropping pattern measures how manypixels are missing from each edge of the full 720x 480 pixel image. First, set your display tounderscan mode (reduce vertical and horizontalsize enough to make raster edges visible.) Alongeach edge of this pattern you will see a dashedline. Each segment of the dashed line is 1 pixelfurther from the frame edge. Markers indicatehow many pixels have been cropped from anedge. Find the outermost visible dash segmentand read the adjacent markers to measure pixelcropping.

Layer Change TestDVD players vary in the amount of time neededto complete a layer change. Delays can rangefrom momentary to considerably long breaks in afilm. This test allows simple measurement oflayer change time with just a stopwatch.

Start the test and watch as the frame count downapproaches zero. You will also see horizontallines approaching each other. At the moment thelines meet (and the counter reaches zero), startyour stopwatch. The animation will continue andthe second half of the test will count towards themoment you need to stop your timer. Somestuttering or pauses in the animation indicatelayer change activity. Wait until the lines meetagain and then stop your timer. Subtract fiveseconds from the elapsed time and you have theplayer’s layer change time.

Layer Change Test measures how long a DVD playerneeds to change layers.

The test works by having precisely five secondsof animation between the two times the linesmeet each other. The DVD was then authored toplace a layer change during the middle of theanimation. Subtracting five seconds from theactual playback time and what is left is the timeneeded by the player to change layers. Thebackground of this pattern intentionally containsa great deal of noise. This forces a high bit rateand minimizes the effect of buffered data onlayer change time.

Uniformity

Uniformity: Orbiting dots against black, gray, white, red,green, and blue backgrounds test screen uniformity, andpresence of bad pixels.

The Uniformity pattern is a visual test of howwell a display maintains screen uniformity.Screen uniformity problems such as side-to-sidecolor shift on CRT projectors, imperfect opticalintegrators in digital projectors, uneven panelbias voltages, and hot spotting are all easilydetected using this test pattern. The test alsochecks for stuck or dead pixels on fixed paneldisplays.

Two dots revolve in opposing directions in frontof a succession of black, gray, white, red, green,and blue backgrounds. Use the pattern by allow-ing your eyes to follow the motion of one orbit-ing dot. Pay attention to the background as youreyes shift around the screen. The pattern back-ground is uniform and devoid of features. Uni-formity problems appear as screen areas differingfrom the rest of the screen. You may notice thatone corner of the screen is brighter, or one side isa different color of gray, or the center of thescreen is much brighter. These are all due toscreen and/or projector uniformity problems.

Picture elements stuck in the on or off positionare readily detected as anomalous spots if onetests using all six color backgrounds. Look forany pixels that fail to change color in step withthe rest of the pattern background. While somebad pixels are acceptable in computer and pres-entation applications, the more critical viewingneeds of a home cinema practically demand theprojector have zero dead or stuck pixels. Other-wise, viewer attention is distracted during films.

If a display has slow recovery time, a faint trailmay be seen behind the orbiting dots. This ismost likely to be seen on LCD panels with slowrecovery time. Counter rotation of the two dotsalso aids observation of color separation artifact.As your eyes track the motion of one dot, theymove relative to the other dot. The other dot may

break up into separate red, green, and blueelements on systems that sequentially display thethree primary colors.

Rainbow Dither

Rainbow Dither pattern reveals temporal dithering andcolor separation artifacts.

Some digital displays (and video processors)have insufficient bit depth to represent all lumavalues within a single video frame. A combina-tion of spatial and temporal dithering can in-crease the effective bit depth and renders gradi-ents smoothly during still images, but imagemotion dramatically reduces the effectiveness ofspatial and temporal dithering. During motion,bit depth may plummet and create visible con-touring or banding. On live material, one seessuch problems during camera pans. Avia PRO'sexclusive Rainbow Dither test pattern allowsobservation for temporal dithering motion arti-facts at various image motion velocities. Thepattern has a second function. It also helps detectcolor separation or rainbow artifacts on sequen-tial color displays.

The Avia PRO Rainbow Dither patterns aresupplied in both video motion and 2/3 pull downfilm motion versions. Moving placards in thepattern indicate velocity of motion in pixels/filmframe or pixels/video field. The patterns startwithout any motion so still image bit depth maybe compared against performance during motion.

Gray, dark gray, and flesh tone spheres in thepattern possess smoothly contoured gradients.These glide across the screen and test bit depthperformance at varying velocities. Better per-forming displays avoid contouring up to highervelocities. The author suggests familiarizingyourself with the appearance of the pattern on aCRT monitor to see how a display without ditherrenders the moving pattern. Once familiar withthe artifact free pattern appearance, you’ll be ableto recognize banding and contouring of thespheres more readily.

Rainbow artifact or color separation artifact canbe induced if a display sequentially presents thecolor components of an image. During eyemotion, image details may visibly break intoseparate red, green, and blue images. Colorseparation artifact is less visible as color se-quence rate increases. Watch for color separationartifacts by following the motion of the spheresacross the screen while paying attention to thethree vertical white bars in the pattern. As motionvelocity increases you may see the bars separateinto different colors. Color wheel speed has adramatic effect on how high pattern motionvelocity reaches before one sees color separation.

This unique Avia PRO pattern facilitates system-atic comparison of dither and color artifactperformance between displays with differentspeed color wheels and displays without sequen-tial color presentation.

Moving Zone Plate

Moving Zone Plate tests resolution and deinterlacing atmultiple motion velocities and directions.

Deinterlacing and scaling performance are centralto the success of today’s progressive scan dis-plays. Not only must a video processor deinter-lace appropriately, but it must also scale theimage to target display resolution. Both processespotentially degrade image resolution. The AviaPRO Moving Zone plate patterns test scaler anddeinterlacer performance over a range of veloci-ties and directions. Moving Zone Plate tests areprovided in 2/3 pull down film motion, interlacedvideo motion, and 2/2 pull down motion versions.

Motion Indicators: The first line of text in thepattern indicates the current type of motion. ZonePlates in 2/3 pull down film motion are flaggedand test the ability of a DVD player to detect andproperly follow pull down and repeat flags.Deinterlacing within a DVD player should bewithout loss of lock during film motion. If theinterlaced output of a DVD player is used,cadence detection of video processors is tested

during film motion. Video and 2/2 motions arenot flagged on disc and test cadence detectionrather than flag detection.

Motion in horizontal, vertical, diagonal, andcircular directions are all tested. Velocity isindicated in x and y pixels/frame. As each motiondirection is tested, velocity accelerates. Motionwas animated in integer pixel increments to avoidmoiré effects that would have occurred withfractional pixel motions.

Zone Plate: A circular zone plate from 2.0 to 4.0MHz tests for resolution losses and moiré arti-facts.

Resolution Wedges: Horizontal and verticalresolution wedges serves as additional indicatorsof image resolution during motion.

Color Plates: Small color zone plates at 2.0 MHztest red, Cr, and Cb resolution. The top colorplate is modulated in red, the middle is modu-lated in Cr, and the bottom modulated in Cb.Chroma components are encoded RGB legal topermit testing of processor which operate in RGBcolor space. On systems that degrade chromaresolution, internal circular features of the colorplates are attenuated.

Motion Transitions

Motion Transitions: Zone plate currently six frames pastlast edit as indicated by white dot in pattern and alsotext counter.

The Motion Transitions pattern is a rigorous testof deinterlacer accuracy and recovery. Simplemoving zone plate patterns test resolution andscaling during motion, but deinterlacers must alsocope with scene edits that alter film cadence orswitch between film and video motion. AviaPRO’s Motion Transitions test contains all 25film-to-film edits (including bad edits), 5 film-to-video edit transitions, and 5 video-to-film edittransitions. Cadence detection, stability, andrecovery time to regain lock are all tested withthis pattern.

This test is particularly severe for progressivescan DVD players. It is easy to design a progres-sive scan DVD player to merely follow on discflags, but not all DVD material is properlyflagged. Some film-based material may not evenbe flagged. An in-player deinterlacer mustactually have a means of detecting cadenceindependent of flags if it is to properly deal withall material. Rather than test the more trivial taskof following flags, the Avia PRO Motion Transi-tions pattern is not flagged. So, it tests actualability to detect cadence from video content. Thatis a task which external video scalers mustnaturally perform (since they never receive flaginformation). Simple in-player deinterlacers thatrely solely on flags will fare poorly with this test.

Detection of cadence is not without a price. Theprocess requires several frames of video to beanalyzed in order to decide current cadence. Thismeans video processors will often delay the videosignal slightly. If the delay is large, audio andvideo lip sync errors become noticeable. Thispattern includes a beep tone that can be used tocheck the number of fields video delay that havebeen introduced by the video processor. As thezone plate in the pattern reaches the top of itsmotion (and usually bottom) a sync check tone isplayed. If video delay is present, the tone will beheard before the zone plate reaches top position.The number of frames error can be estimated onscreen.

Motion Transitions: Zone plate during a frame that hastop and bottom fields from same film frame. Oncecadence lock is regained, the zone plate appears moresolid.

Zone Plate: A zone plate revolves inside acircular series of tick marks. Observe how wellthe zone plate is preserved. When cadence lock isincorrect, the zone plate will moiré or double.Once cadence lock is reestablished the zone plateappears markedly more solid.

Timing Marks: A white dot revolves in syncwith the zone plate. Each tick mark representsone video frame of time. Every fifth mark isdarker. A total of 2 seconds or 60 frames equals

one revolution around the circular path. Watchingthe zone plate and noting the white dot position atwhich lock is regained measures the number offrames needed to regain lock.

Edit Type Indicator: At the upper left of thepattern, the upcoming or previous type of edit isindicated.

Current Frame Type: Halfway down thepattern, at the left side are two letters. The upperletter indicates which frame the upper field isfrom. The lower letter indicates the frame thelower field is from. Text is intentionally nearlythe same color as the background to avoidconfusing deinterlacers with the intrinsically 2/2motions of the lettering changes.

Frames Since Transition Counter: Lower leftof screen indicates how many frames have passedsince the last edit transition. Each cadence is keptstable for at least 150 frames before the next editoccurs.

View each edit transition and note how long thedeinterlacer takes to regain lock. If the zone plateappears not to change appearance immediatelyafter edit transitions or always looks like it doesduring video motion, the most likely explanationis that the deinterlacer does not detect cadenceand has defaulted to a fixed video cadence. On awell functioning deinterlacer, the zone plate willusually moiré or double for several frames afteran edit transition. The number of frames neededto regain lock varies with type of edit transition.Observe the entire sequence to test all types ofedits.

You may sometimes see a processor lose lockbriefly after cadence has been stable. Closeexamination will probably reveal this is the pointat which wording for the next transition typeappears. This is normal and indicates the proces-sor detected the wording change as noise in thevideo stream. Cadence is usually quickly re-gained after such a “glitch.”

Disc NavigationControl of Avia Pro Video Test DVD is conven-iently performed through on-screen menus. DVDremote control buttons have the following ac-tions:

Title: Jumps to main disc menuMenu: Returns to current menuNext: Goes to next test signalPrevious: Goes to preceding test signalReturn: Also goes to preceding signalUp, down, left, right: Selects an itemEnter: Displays highlighted item.

Title and chapter number navigation enablesrandom access of test patterns. However, onsome DVD players, this feature is available onlywhile an actual test signal is being displayed, butnot during display of a menu.__________________________

Notes:

Frequency values in this chapter and on discassume interlaced output. If progressive scanoutput is used, actual signal frequency is doublethat indicated.

IRE labels assume 7.5 IRE setup is added by theDVD player. If no setup is added, then computeactual IRE as…

Actual IRE = (labeled IRE - 7.5)/92.5 * 100

GK