Jelly Effect on iPad Mini 6 (Blur Busters Technological ...

An easier test is the TestUFO Scan Skew / Jelly Effect test which intentionally shows a bouncing UFO, to force your eyes to move around, and ... BlurBustersForums Whoyougonnacall?TheBlurBusters!ForEverythingBetterThan60Hz™ Skiptocontent BlurBusters.comTestUFO.com Search Advancedsearch Quicklinks Unansweredtopics Activetopics Search FAQ Login Register BlurBustersForums Boardindex TheLaboratory Area51:DisplayScience,Research&Engineering WelcometotheBlurBustersForums!Come JOIN theDiscussions! Search JellyEffectoniPadMini6(BlurBustersTechnologicalExplanation) Advanceddisplaytalk,displayhackers,advancedgameprogrammers,scientists,displayresearchers,displaymanufacturers,visionresearchers&AdvancedDisplayArticlesonBlur Busters.ThemastersonBlurBusters. PostReply Printview Search Advancedsearch 9posts •Page1of1 ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube JellyEffectoniPadMini6(BlurBustersTechnologicalExplanation) Quote Post byChiefBlurBuster»30Sep2021,22:36 TheiPadMini6JellyEffectThread Toprepareforapossiblefutureresearchwhitepaper,IamcrosspostingabunchofArsTechnicacommentrepliesherefortheirarticleabouttheArsTechnicaiPadMini6JellyEffectarticle. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:37 Post#1ofSeries JellyEffectisCausedbyaPerfectStormofMultipleCauses BothAppleandifixitisright: ItisnormalrefreshingbehaviourBUTunusuallyamplifiedinvisibilityfromtheconvergenceofunrelatedfactors: IwantedtocrosspostalatereplyImadefromanotherthread,thatappearstocorroborateifixit: ___ CheshireCatwrote:mdrejhonwrote:AgoodwebtestforthisistheTestUFOJellyScrollingTest. AlthoughlessvisiblethantheMini6,evenaDell60Hzofficemonitorshowssomescanoutskewingtoo. Testinbothportraitandlandscape,somedesktop/mobilescreensonlydoitinoneorientation. TestUFOsaysSafarivsync@120+isnotsupported. UsingiPadPro2020. Thatsaid,Icancertainlynoticethenativescanlinerefreshbendingverticallinesduringstupidlyfastscrolling,butitissomethingIhavetolookforanddefinitelynotnoticeablewithnormalscrollingspeed. IamafraidwhatcausesthisissueonthenewiPadminiisadifferentthing. It’snotadifferentthing—it’ssimplybecausethe1/60secscanoutdirectionisalongthe“longervsshorter”axisvsofthescreen.Aleft-to-rightscanoutvstop-to-bottom.Portraitscanversuslandscapescan. Scanoutalongthelongaxisofthescreen(asiPadMini5does)ismuchfaster,sothejellyeffectishardertoseeandjellyeffectisonlyvisibleduringultrafastscrollingwhilelandscape. Scanoutalongtheshortaxisofthescreen(asiPadMini6does)isslower,sothejellyeffectismuchmorepronouncedwhenscrollingwhileinportraitmode.Thescanoutdistanceisshortersothescanoutisslower.Scanoutona4:3screenis4/3rdsfasteralongthelongaxisthantheshortaxis.Soscandirectionisamajordeterminatorofmakingthejellyeffecteasiertosee,whenscrollinginadirectionperpendiculartothescreenrefresh’sscanoutdirection. Thus: (A)OniPadMini5,Jellyeffectbarelyvisibleinverticalscrollingduringlandscapemode (hardertosee,needsfastscrollingwhiletrackingeyesontexttobarelynotice) (B)OniPadMini,6Jellyeffectmorevisibleinverticalscrollingduringportraitmode (muchmorevisiblewithslowerscrolling) TheiPadMini6isaperfectstormofmultipleconvergingfactors: -MostpeopleholdaniPadinportraitmodewhenbrowsingtheweb -ThisisthescreenorientationthatjellyeffectappearsonMini6 -Italsohappenstobethe“slowestpossible”60Hzscanoutvelocity(theshortaxisofscreen) Youalsohavetosimultaneouslyeyetrackthetextwhilescrolling(notfixedgaze)toseethejellyeffect,differentpeoplegazedifferently(stationaryvsmovinggaze)sonoteveryonenotices. Aspecifichuman’smaximumeyetrackingspeedisalsoavariableonwhethersomeonecanseethejellyeffect.Scrollfairlybriskbutatavelocitynofasterthanyoureye-trackingability.Thenstareatthetextasyouscroll(orbouncescrollthetextofawebpageupndowninabounceliketheTestUFOtest). Screensoftenlookdifferentinfixedgazeversusmovinggaze.SomeTestUFOtestsdemonstratethisperfectly,suchasTestUFOEyeTrackingaswellasTestUFOPersistenceofVision. Likewise,thescientificdifferenceofstationary-vs-movinggaze(interactingwithdisplaybehaviors)isalsotrueforthejellyeffect,itonlyreallybecomesvisibleifyoutrackyoureyesonthetextwhilescrolling. Thusexplainsaportionofthearguinggoingonhere,fromsomepeoplewhodonotseethejellyeffect—eitherbecausetheyweredoingafixedgazewhiletextscrollspast—orbecausethescrollingspeedsnecessaryformakingjellyeffectvisibleisbeyondthatspecificperson’smaximumeyetrackingspeed—orthescanoutisnotperpendiculartoscrolldirection.Evenjustamereoneofthesefactorsmeansthehumanlikelywon’tseethejellyeffect. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:39 Post#2ofSeries DisplayresearchPart2ofmyhighlyinsightfulreplyonpage1. sprockketswrote:It'snormal,buteveryoneelseputsthescreeninthe"right"waytoavoidthisissue.Sometimes,likeOnePlusdidwiththe5,theyputitupsidedownduetotheirengineeringdecisionsbecauseofwheretheconnectorcablehadtogo. Technically,Ifirstthoughtup-sidedowndidnotmakesense,becausejellyeffectonlybecomeseasilyvisiblewithscrollingdirectionsperpendiculartoscanoutdirection. However,upsidedownisanadditionalfactorforanunexpectedreason: Ilaterdiscovered(withBlurBusterscontractswithdisplaymanufacturers)thatitactuallyhadafactor.BeinginventorofTestUFO,Icreatedaspecialjellyeffecttestpatternaboutacoupleyearsagospecificallytodiscoverwhyscreenshaveajellyeffect. OntheiPadMini5,withdefaultcasestand,thehomebuttonisattherightedge.TheiPadscanoutstartsatedgeoppositeHomebuttononMini5.Sowithdefaultcases,iPadsonatablestartscanningattheleftedge(lowestlag)towardstherightedge. Ilaterdiscoveredmosthumansstareattheleftedgeofthescreenfirstbecausehumansreadlefttorightinmostoftheworld’slanguages.Soifthescreenscanoutstartsattherightedgefirstandscansleftwardstotheleftedge,theleftmostedgeisthemostlaggedscreenedge.Thatamplifiesthevisibilityofthejellyeffectifthemostlaggedscreenedgeaccidentallyhappenstobetheleftedge! However,ifIholdtheMini5upsidedown(landscape,homebuttonatleftedge),thejellyeffectbecamemorevisible(Eventhoughonly3/4thsasvisibleasiPadMini6,fromthe4:3aspectratioandlong-axisscanout,mentionedinmyearlierpost). Soitisanadditionalfactor:Thegeneraleyegazeareaisusuallytheleftedgeofthescreen,becausethatiswheremostpeoplebegintostartreadingthetext.Ifthatedgeofthescreenismorelagged,thejellyeffectismorevisible. Sotheupsidedown+perpendicularscanoutcomboindeedamplifiesjellyeffectvisibility.TryholdingyouriPadMini5upsidedown(homebuttononrightedge)andstartscrollingtheArsTechnicacoverpageup/downinamedium-speed(sinewave)bounceupanddownbackandfourth(scrollArscoverpageatthesamespeedashttp://www.testufo.com/scanskeworslightlyfaster)—andinstantly,thejellyeffectismuchmorevisibleonMini5. TheMini6addsextracakebecausethereisnofronthomebuttonandpeoplearemorelikelytoholditsimultaneouslyupsidedown+perpenticularscanout.ANDthescanoutisalongtheshortaxis(slowerscan,easiertoseeinperipheralvision) Noteveryonecanseeit,butalargerproportionofmyfamilystartstoseeitontheMini5too,notjustMini6.Sothat’syetanotheradditionalfactor,thatIhadnothithertodisclosedtopublicuntilnow,andyouArscommentaudienceisthefirstaudienceI’mdisclosingthisto— FortunatelytheiPadMini5casesmeansmostpeoplewon’tbeviewingtheiPadMini5withthehomebutonontheright-handedge,whenproppingtheiPaddowninasmart-casestylecase.TheMini5startsitsscanoutattheoppositeedgeofscreenthanhome-button,sothatedgeofthescreenhasthelowestlatency—theareawherepeoplestartreadingtext.Becauseofthecase-hintingandhome-buttonhinting,iPadMini5arealmostneverheldinthescreenorientationthatamplifiestheMini5existingjellyeffect. Mitigationfactorsinclude: (1)Go120Hzor240Hz(preferred)asitisauniversalsolve-allforalljellyeffects. Thispushesscreenrefreshscanoutvelocitybeyondmajorityofhumanperception,evenintheworst-cases. (2a)Alternatively,designscreentoscanalonglongaxis. Along-screen-axisscanoutisfasterandmakesjellyeffectlessvisible. (2b)And/or,designscreentoscanverticallyindefaultholdingposition Rotationbias/hintingsuchascasedesignandhomebuttonlocationencouragesholdingiPadinless-sensitiveorientiations (2c)And/or,designdevicetoscanbeginningattheleftedge—whenrotatedinawaythatmakesscrollingdirectionperpendiculartoscanoutdirection. Samerotationbiasasabove (3)Alltheabove(1)(2a)(2b)(2c). Thiscoversthesmallerhumanpopulationthatisstillabletoseethejellyeffecton120Hzscreens.* *120Hzjellyeffectisstillwithinafewhumans’perceptionsifthesameperfectstormisintentionallydone:Ipersonallystillseeitonthe120HziPadsbutonlyinonesuboptimalorientation(whereleftedgeisthemostlaggedscanout)andwhenscrollingtwiceasfastatfully-outstretched-armviewingdistancewhileusingslightlylargerfontsizes.It’sHARDtosee,butasuper-faintjellyeffectstilloccursat120Hzatspecificscrollingspeedsnearmyeye-trackingspeedlimits.Iuseahighspeedcamera(usingthistestmethodology)tofigureoutwhichisthelaggedscanoutedge,andintentionallyholdtheiPadthatwayfortheleftedgetobethemostlaggededge.Atthispoint,Isuddenlystarttofaintlyseethe120Hzjellyeffectinhorizontaladbanners.Thistime,Idon’tnoticeifIamnotpayingattention—Ireallyhavetolookforithardandscrollfastwhileatfulloutstretchedarmextension.Easiertoseeinhorizontaladbannerswhichstartstoverticallyparallelogramslightlyatultrafastscrollspeedswhileeyetracking.UnscientificallyI’dwagerlessthan1/100thofpeopleseeingjellyeffectat60Hzwillseeitat120Hz,becauseit’srightatthethresholdofhumanperception.Butthefactthatatleastafewofusseesitwhenintentionallydoingtheperfectstorm,meansit’snot100%solved,andweshouldsomedaygo240Hzwhen“free.”.Theperfectstormofjelly-effectamplifyingfactorsoccursinonlyonespecificscreenorientation(upsidedownANDperpendicularscanout),andeye-tracking(whilescrolling)certainjelly-effect-sensitivewebpageobjectssuchaswidethinhorizontaladvertisementbannersduringfastscrolling. Ipreferoption(1)going120Hzor240Hzespeciallyasitwilleventuallybecomealmost“free”battery-wiseandcost-wise(likeretinascreensnoware)…HighHzisveryergonomicforbrowserscrollingaslongasitcanbedonewithoutsignificantpowerincrease(120Hzonsample-and-holdLCD/OLEDshalvesLCD/OLEDscrollingmotionblur,and240Hzgives1/4thmotionblur—240Hzisnotjustforgamesanymore.If240Hzcaneventuallybemade“free”cost-wiseandpower-wise,itshouldeventuallybeusedinsteadof120Hzsomeday.Butfornow,costwise,thisisnotthecase,so120Hzisagoodstartingpoint.Itisbecominghighend-mainstreamstandardinconsolesandmanynewerTVs,whichwillbringcostdropsto120Hztobecomeasfree/lowcostaddas4Kscreensandretinascreens). IthinkIshouldbegintowriteawhitepaperaboutthis.Mycommenthereisthefirst-everdisclosureofthisdiscoveryof“perfectstormoffactors”combiningtomakethejellyeffectmorevisibleinnormalLCDscanouts.Ifanyresearchersreadingaboutthis,pleasecreditmeforthisdiscovery(I’malreadycreditedinover20peerreviewedpapers).VergeandIFixitindeedmentionedindividualfactors,buttheyaren’tenoughbythemselvesalone—Idiscovereditrequiredaperfectstormoffactors(possiblycoincidential/accidental).Anyresearchersonthis,contactmeifyouwanttocollaborate/co-authoronthis. (MarkRejhon—displayresearcher,inventorofTestUFO) HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:40 Post#3ofaSeries Statisticalwrote:Simplesolutionwouldbetwodisplaydriversthatdrivethepanelbasedonorientation. Okthatwastonqueincheekasitisaverynotsimplesolutionbecauseitmeansthepanelhastobewiredtosupportlinesfrombothedges,itaddsalotofextratraces,andyouneedtoputaseperatecontrollerinthedevice.Stillwhoknowswithminiturizationitcouldbeaneconomicalsolutionsomeday. Workinginthedisplayindustry,going120Hzor240Hzisactuallycheaperthanhavingtwodisplaydrivers. Afasterscanoutsolveshuman-visiblejellyeffectconsideration,whilealsoaddingergonomicbonuses(reducedscrollingblurevenfornon-gameapps). There’scurrentlyapushtoreducethepowerandcostpenaltyof120Hz.Therearenowsub-$5004KHDTVswith120Hzbuiltin,thankstoPlayStationandXBoxforcingthecommoditizationof120Hztech.Iexpect120Hztobecommoditizedmuchlikeretinascreens. Besides,onflickerfreesample-and-holdscreens—120HzhalvestheLCD/OLEDmotionblurof60Hz,and240HzquarterstheLCD/OLEDmotionblurof60Hz,forthesamescreenscrollingvelocity.Textscrollingbluratdifferentframeratescanbedemonstratedathttp://www.testufo.com/framerates-text—trythisona120Hzor240Hzdesktopmonitorandtheergonomicbenefitsbecomeobvious. Itisthecurrentinternalindustryexpectationsfor120Hztoeventuallybecomea“freebie”featuremuchlike4Kdid.Itusedtobea$10,000feature,butnowyoucanbuysub-$300RCA4KHDTVsatWalmartorCostco. Notetootherreaders;I’mnottalkingaboutyesteryear’s“fake”120Hzbyinterpolation.Themodern120Hzisnativeoriginaluniquelow-latencyframesnativelyrenderedbytheoriginalnativecontentsuchasgamesorwebpagescrolling/panning/etc.Itisalso,additionally,amyththat120Hzand240Hzisonlyusefulforgames.Althougheventhough60HzLCDsdon’tflickerlikeyesteryearCRTs,screenmotionblur(amandatorysideeffectofflickerfreesample-and-holddisplays)andjellyeffectsaremajorsymptomsofatoo-lowrefreshrateoriginallylinkedtoACfrequenciesfordirect-driveofCRTelectronbeamrefreshfrequencies. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:41 Post#4ofaSeries Lumianwrote:Resolutionisalsoanissue ThelessPixelRowstherearethequickeryougettothenextone. AndPixelperRowalsotakeuptimetorenderout 60Hzscreensrefreshatthesamephysicalvelocityregardlessofscreenresolution. Theystillphysicallytake1/60sectorefreshfromoneedgetoanother. Higherresolutionssimplymeanthecontrollerisdoingrowrefreshfasterinordertorefreshthescreenoneedgetoanotherinthesame1/60sectime.ForAppleproducts,Idiscoveredresolutionhasnomeaningfuleffectonthejellyeffectbecausetheyjustrow-refreshfastertokeepthesamerefreshintervalforfirstthroughlastpixel. Thejellyeffectiscausedbyotherreasons(seePost#1andPost#2) binaryspiralwrote:Bumpthemallto120Hzandproblemsolved. This,this.Also,120Hzwilleventuallybecomenearly“freebie”power-wiseandcost-wisemuchlike4Kandretinadid.Thiswillgraduallyincrementallyhappenasthisdecadeprogresses. otsowrote:WXWwrote:otsowrote:Can’tthisbefixedwithVsync,https://en.wikipedia.org/wiki/Analog_te...ronization?I.e.,redrawattheexactsamefrequencythewholescreenisrefreshed.Edit:wording No,theissueisnotthattheimagessentaren'tsynchronizedwiththeLCDrefresh,butthatLCDslikethesedon'tdrawalltheirpixelsatonce. I’llthinkmoreandharder,butIhonestlystilldontgetit.Simplified:Renderimage1,waituntilallpixelshavebeendrawnfromimage1,renderimage2,waituntilallpixelshavebeendrawnfromimage2,etc. IcanconfirmthatiPadMini6doesexactlywhatyoudescribe:Synchronizedrefresh. Therealproblemislawsofphysics,combinedwithaperfectstormoffactors. Notallpixelsonascreenrefreshcanatthesametime,fromfirstpixeltothelastpixelofascreen. ImagineascreenasisessentiallysortofalargeRAM—It’sthesamewayyoucan’twriteallbytesofaflashchiporDRAMchipsimultaneously. -In1stscreenrotation,screenrefreshestop-to-bottom -In2ndscreenrotation,screenrefreshesleft-to-right -In3rdscreenrotation,screenrefreshesbottom-to-top -In4thscreenrotation,screenrefreshesright-to-left Nonon-strobedflickerfreeLCDscreenscandoglobalrefresh(makeallpixelsofaframevisiblesimultaneouslyinstantly). Seehighspeedvideos.A60Hzscreengenerallytakes1/60sectorefreshthefirstpixeltothelastpixel.Thisisthesourceoftheskewing,especiallywhencombinedwithaperfectstormoffactorsdescribedinmyPost#1andPost#2. TL;DR:Theperfectstormofjelly-effectvisibilityoccurswith (1)Screenorientationanddefaultholdingposition(casedesignandhomebuttonlocationencouragesholdingiPadinless-jelly-effectsensitivescreenorientations), (2)scanoutalongshorteraxisofscreen(slower)insteadoflongeraxisofscreen(shorterscandistanceisaslower1/60secsweepthatiseasiertoseejellyeffectin),and (3)edgeofscreenwherethescreenscanoutsweepbegins(i.e.beginningoftextrowsbecomingmostlaggedwherehumansbeginreading). Whenallthesefactorscombinesimultaneously,thejellyeffectisthemostvisible. AlliPadsrefreshinthesamesynchronizedway.It’sjustthatthere’stheaboveperfectstormoffactorsoccuringonMini6asdescribedinmy2posts.iPadrefreshesalongtheshortaxis(sidewaysscaninportrait),andmanyareholdingitinawaywherescanisright-to-left(causingbeginningoftextrowstobemostlagged). JellyeffectoccursonalliPadsbutmaynotbevisibletohumansdependingoncircumstances(butthejellyeffectbehavioursarealwaysabletobecapturedfromalliPadsvia1000fpshighspeedcameras).Eventhe120HziPadssuper-faintlybyveryfewpeoplewhoareunusuallysensitive,butit'salwaysvisibleinsufficientlyhighspeedcameras.ItisjustmostamplifiedontheiPadMini6duetotheperfectstormofsimultaneousfactors. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:42 Post#5ofaSeries OrangeCreamwrote:Rightnowtheyupdatethescreenassoonasthedataisready: Updaterow1;updateimage Updaterow2;updateimage Updaterow3;updateimage Wrong.iPadsdon’trefreshthatway. Inthelinkedpagefrommyearlierpost,Ialreadyhavea1000fpshighspeedvideoofanolderiPadMini(noteasilypronetoJellyeffect,theyrefreshfasteralongthelongaxisofthescreen). AttheOS,theyrenderawholeframeTHENdoarefreshsweepofthewholeframefromthesamebuffer.Rinseandrepeat60timesasecond(or120timesasecond).TheGPUdoesn’tre-rendereachpixelrowinabeam-racedfashion(likeanAtari),AppleusesfullcompositedframebufferslikemostmodernOSes,theysimplyhaveexcellentOS-levelframerate=Hzsynchronization. I’malsoasoftwaredeveloperandcanconfirmthatthesmoothscrollingistheartofaperfectVSYNCthatperpetuallykeepsframerate=Hz(andApple’siOSimplementationofaverylow-latencyscreencompositor). SmoothscrollingfamousonAppleproductsisbecauseofApple’sfamouslyperfectframerate=Hzsync(andexcellentlow-latencyVSYNCON).Iagreewithyouthattheirsoftwarestackisprettymuchagoldstandardinsilky-smoothscrolling. NochangehasoccuredtotheexcellentsmoothscrollingontheMini6. Thejellyeffectisalsounrelatedtohowsmooththescrollingis. Thejellyeffectissimpyanadd-ontiltingtotheexistingsupersmoothscrolling.Thejellyeffectislikeparallelogramming(ofthewholescreen)duringfastscrolling,essentiallyascreenequivalentofaslowrollingcamerashutter.Kindoflikehowajellysculpturetiltsleft/rightwhenyoushake/rocktheplatehorizontally.I’mnotsureifyouhaveseenityet,buttryhttp://www.testufo.com/scanskewonalargerscreensuchasadesktop60HzLCDortelevisionset—thiswillhelpeducateontheappearanceofwhatthejellyeffectlookslike. Someoldcameraswitharollingshutter(aslitinarotatingopaquewheelthat“scans”alongtheunexposedfilm)showedthisparallelogrammingeffectduringfastmotionsuchasracing: OntheiPadMini6,thejellyeffectiswhenleft/rightedgelookshigherthantheoppositeedgeofthescreen,creatingtheimpressionofslightly-diagonallinesoftextordiagonaladbanners,whenyouscrollfastWHILEinportraitmodeWHILEeyetrackingthescrollingtextWHILEwithinthespeedofyoureye-trackingabilities. Thetext-rowtiltingangleisapproximately3to5degreesontheMini6atmoderatescrollspeedsthatareeasilyeye-tracked. Themotionisstillsupersmooth,justsimplynowhasatiltvisibleinit. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:43 Post#6ofaSeries OrangeCreamwrote:AmImisunderstanding? Yes. Thejellyeffectissimplyaperfectstormoffactors.Seemyearlierpost,aswellasPost#1andPost#2. Iworkinthedisplayindustrywithcredit/referencesinover20peerreviewedindustrypapers. Thetilteffectisfromeyemovementwhilethescreenismid-refresh.Frameisglobal,butrefreshisnotglobal.Tomakethescanskeweffectdisappear—needtime-relativesync’dglobalrefresh(instantframe+instantrefresh)orsync’dbeamracing(beamracedGPUrefreshofeachpixelrowinsyncwithscreenscanout). Forphotography,thephotographicplateisstationarybutthesceneryisnotstatic,soyou’vegotatime-distortionalongonephotoaxis,iswhatcreatestheskewembeddedinaphoto.Theskewactuallydisappearsforthecar(butthebackgroundbeginsskewing)iftherolling-shuttercameraoperatorrotatesthecameraonthetripodtofocusonthemovingobject(e.g.racecar)instead.Anythingthatmovesrelativetothecamerafilm,getstheskew. Forhumaneyesonascreen,theOSglobalframerenderversusthetime-sequentialrefreshofadisplayinteractingwitheyemovement,iswhatcreatestheskewseenbyeyes,asclearlydemo’dinTestUFOscanskew. ...Theskewdisappearsifyoureyesarestationary,buttheskewreappearsifyoutrackthebouncingUFOwhenscandirectionisperpendiculartoUFObouncedirection. ...Theskewalsodissappearsforglobal-refreshscreens(e.g.DLPprojectors--notquiteglobalbutessentiallyglobal--theDLPscanoutofthelatestDLPchipsisanultrafast1-bitrefreshcyclescanoutsweepin1/1440secor1/2880secforthenewestDLPchips. Note:Mostglobalrefreshdisplaysarestillsequential-refresh,justsuper-fastorrefreshinginthedark(beforeflashedlikeastrobebacklight).Anexampleofsuper-fastscanoutisconsumerDLPchips,mostofwhichrunat1-bit1440Hz.Forexample,24bitx60Hz=1440Hzneededfor1-bit(2-color)refreshcyclestogenerate24-bit60Hz60fps.DLPcreatescolorsviatemporalditheringsolelyon/offprimary-colors.TheDLP'sASIC/FPGAgeneratesmany1-bitmonochrome(aka2-color)framespersignalrefreshcycle.Sotherecanbe24DLPrefreshcyclesforevery1signalrefreshcycle,runinvariouscolors(viacolorwheelorRGBLEDflashing).EvenfortheslowestDLPchip,thescanoutsweepissofastinafractionofamillisecondforthewholechip.Thus,itisessentiallyglobalrefresh,byallhumanvisibilityperceptionstandards.Differentimageartifactsideeffectscanoccurforsomehumansfromthistemporalbehavior--rainboweffectsand/ortemporaldithernoise--butyouneverseeanyskewingonDLP.. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube Re:BlurBustersTechExplanationOfJellyEffectoniPadMini6 Quote Post byChiefBlurBuster»30Sep2021,22:43 Post#7ofaSeries aerogemswrote:Yes,becauseI'mnotsurprisedyouapparentlydidn'treadmycomment,orthecommentIwasrespondingto,whereitwasproposedthesolutionmightbetoaddasecondcontroller,oneforeachorientation,notsimplymovethecontrollerfromonesidetotheother.Which,ofcourse,theOPrecognizesislikelyanunworkablesolutionfromabusinessperspective,evenifit'sprobablythecleanestfromanengineeringperspective. Thatwon'twork.Adifferentscreenrotationorientationwilljusthavethejellyeffect. mikeschrwrote:IhavethepreviousminiandImostlyuseithoriontally.ItsoundslikeIshouldbeseeingthiseffect,butIdon't. ItbecomesmorevisibleonMini5IF; (A)youholditlandscapewithhomebuttonatleftedge;AND (B)youeye-trackthewebpagewhilescrollingthepage;AND (C)youkeepyoureyeshorizontalyroughlymiddle;AND (D)youscrollmediumspeedonhigh-contrastobjectsbutdon’tscrollfasterthanyourreliableeye-trackingspeed;AND (E)EasierifyouholdtheiPadalittlebitfurtherbackthanusual(easiertoseetiltinperipheralvision);AND (F)Keepyourfingerhelddown&youbouncescroll(slidefingerupanddownrepeatedlyatmoderatespeeds); ThenthetiltingeffectbecomesmorevisibleonMini5especiallyonhorizontaladvertisementbanners&high-contrastwidephotos,orhigh-contrastrowsoftext.It’svery,veryslight(muchfainterthanMini6). Mini5jellyeffectislesspronouncedasit’shardertoreproduceontheMini5becauseoftheneedtointentionallydomoreoftheabovesimultaneously,andalsothefactthatthescanoutsweepisfaster(thelongaxisofthescreen) AneasiertestistheTestUFOScanSkew/JellyEffecttestwhichintentionallyshowsabouncingUFO,toforceyoureyestomovearound,andintentionallybounce-scrollstoalloweasiercomparisonoftwotiltingextremes,formaximumamplificationofthejellyeffectofascreen. hasibwrote:Otherthanthat,I’mjusttryingtounderstandwhetheritisfairtobegthequestionofaddingasecondcontrollerinthefirstplace,regardlessof“businessperspective”. ItcanbedonebutLCD/OLEDscreensaregiantlithographically-createdintegratedcircuitswithtransistorsandcircuitlines. TFT=ThinFilmTransistor,sothinkofascreenasagiantRAMorflashchip.Eachpixelhasatleastthreetransistors(oneforeachsubpixel),sometimesmoreforimprovedpixelholdmemory(e.g.abilitytorefresh1Hzwithoutneedingrepeat-refreshforlowframeratecompensation).Buttherearealsooffscreentransistorsattheedgestocontrolrow-columnaddressing. Shiftregisters!Otherwise,weneedtwopairsofunobtainiumribboncableswith3840and2160wiresrespectivelyforthehorizontal/verticaladdressorsofa3840x2160screen.Fortunatelyshiftregistersintheintegratedcircuitlithographedontothescreenedgedirectlyontothescreensubstrate.Thiskeepsribboncablesundercontrol,reducingthenumberofcircuitwiresonthatscreen-edgeribboncable. So,yes,fullyoperatingelectronicscircuitsdirectlylithographedontothescreenglass,usingthesamelithoprocessaspixeltransistorsonTFTactivematrixscreens).Screensaresheercomplexity,andnotalllogicelectronicscanberemovedfromthescreenglassitself! Asyounowrealize,row-columnaddressorelectronicsaremandatorilypartiallyintegratedintheedgeofascreen,soaddingabilitytodohorizontalandverticalrefreshsweepsalsomakeascreenalittlemorebulky(largerbezels). Italsoaddslithographiclayerstothescreensurfacetoo,duetomoreover/undercircuitpathsonthescreenglass,fortheadditionalon-glassdisplayrefreshingaddressorsandtheaddressor-methodswitchinglogic(supplementalon-glasscircuitstochangescanoutdirectioninrealtime). Nottomentionmoreprogramming/designfortheFPGA/ASICofthescaler/TCON(themainchipattachedbyribboncabletoscreenedge).Thatincreasesdesigndeliverytime. It’scheapertoincreasethescreenrefreshrateanyway.Most60HzLCDsonlyrequireminorintegrated-circuit-on-glassmodificationstobeeasilyupclockedto120Hz,providedtheLCDpixelresponseissufficientlyfastandtherowrefreshspeedcanbedoubledwithoutscreenartifactswiththesameintegratedelectronicsdesigns.Most60HzLCDglasscanbeoverclockedto120Hzalbietwithartifacts(especiallyifthescaler/TCONissorudimentary,orisreplacedwithaDIYFPGAlikeZisworksdidfortheexperimental540p480Hzonacheapchinese4K120HzLCD.Acheap60HzLCDsuccessfullyoverclockedto180Hz—hadseverecompromisessuchasslowerpixelresponse(notenoughpowerin1/3rdtimetorefreshthepixelfully)andstreakingeffects. Asalludedearlier,thecompromisesof120Hzarenowbecomingmoreandmorecheaplysolvable.Thelimitingfactorusedtobepixelresponse(andpowertoalesserextent),butpixelresponseisnowadaysfastenoughtomake120Hzusable.And120Hzpenaltyisusingasmallerandsmaller%ofbattery,asthecontinuedefficiencyimprovementofretinascreenshavealsoshown.Oncethecommoditizationweaklinksdissapear,120Hzwillslowlybecomeaneventualfreebiefeature(muchlike4Kandretinascreens)asthisdecadeprogresseswithoutmuchcompromises.Theboomofnative120Hztech(newPlayStation,newXBox,newiPhone,newGalaxy)ispushingaslowcommoditizationof120Hzinclusioninscreens. HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top ChiefBlurBuster SiteAdmin Posts:9896Joined:05Dec2013,15:44 Location:Toronto/Hamilton,Ontario,Canada Contact: ContactChiefBlurBuster Website Facebook Twitter YouTube JellyEffectoniPadMini6(BlurBustersTechnologicalExplanation) Quote Post byChiefBlurBuster»30Sep2021,22:45 Post#8ofaSeries GrahamJwrote:mdrejhonwrote:OrangeCreamwrote:AmImisunderstanding? Yes. Thejellyeffectissimplyaperfectstormoffactors.Seemyearlierpost,aswellasPost#1andPost#2. Iworkinthedisplayindustrywithcredit/referencesinover20peerreviewedindustrypapers. Thetilteffectisfromeyemovementwhilethescreenismid-refresh.Frameisglobal,butrefreshisnotglobal.Tomakethescanskeweffectdisappear—needtime-relativesync’dglobalrefresh(instantframe+instantrefresh)orsync’dbeamracing(beamracedGPUrefreshofeachpixelrowinsyncwithscreenscanout). Makessense.Iwonder,whatwoulditlooklikeifyourenderedthefirsthalfoftheframescanoutfromadelayedbuffer?Sooneachrefreshyouscanouthalftheframefromthedelaybuffer,thentheotherhalffromthecurrentbuffer,thencopythecurrentbufferintothedelaybuffer. Wouldthathalvetheeffect? No,itwouldnotfixscanskewing. Concurrentscanoutscanworsenotherkindsofartifactsoraddsaw-toothing.There’sagoodlate90spaperbyMr.Poyton,fromtestsonLEDmarqueerefreshingbehaviours.See,heknewaboutscanoutskewinginthe1990s!Itshowssomesideeffectsofsubdividingthescanouts.There’satearingartifactbetweenthesubdividedscreens.Ithastobeonecontinuoussweepofthesameframe. ThisisanartifactonaLEDbillboardthatrefreshestop/bottomhalvesconcurrently. It’sliketwoscanskewedscreensflushedge-to-edge. Artifactsoccurfromanyformoftemporallysubdividingrefreshbehaviors: (Forbesteffect,dotheTestUFOanimationsonadesktopmonitor) Slowscan-outsweep=tiltingeffect(Demo:TestUFOScanSkew) Interlacing=combing/venetianblindseffect(Demo:TestUFOInterlaceSimulation) DLPcolorwheel=rainboweffect(Demo:TestUFORainbowEffect) (Epilepsywarning:FlickerforTestUFOon60Hzscreen.DLPcolorwheelsrunat240-360+colorflashespersecond) Temporaldithering=noiseandcontouringartifacts(DLP&plasma) Concurrentscanouts=sawtoothingwithstationarytearingartifactatscanoutboundaries(likemultiplemisalignedscan-skewedscreens) Thereisacomplexdelayed-buffermultiscanalgorithmthatwillfixthesawtooth+tearingartifactissueofsubdividedscanouts,butitwon’tfixscanskew,asthescanskewislinkedtothevelocityofthescanout.ItisalsodescribedinthisBlurBustersForumsthreadoftheoreticalconcurrentscanoutsforatheoreticalnativetrue960HzOLED.Toavoidtearingduringmultiscanning,thesubdividedzonewouldneedtoinheritthesubdividedscreen’sframebufferfromthezoneabove.Sothatthescanoutsweepisatthesamevelocityonthesameglobalfull-screenframebuffer,sweepingthesameframeinonecontiguousscanoutsweep(evenifthere’smultipleconcurrentsweepsgoingon).Essentiallyemulatingatheoreticalmultiple-electron-gunCRTwhereeachbeamisassigneditsownuniquecontiguousrefreshcycleframebufferinonecontinuoussweepperbeam,despitemultiplebeamsgoingonatthesametime.Foraserially-renderedandserially-deliveredseriesofframebuffersfromaGPU,itdefinitelywouldrequirebufferdelays(asyousuggest),butitdoesn’tsolvetheskew,justthetearing.Basicallythescanoutswould“cascade”tothezonebelow(subdividedscreenmodule/partitioning),inordertoeliminatethezigzag/tearingartifact.Itdoesn’tfixscanskew,butitwouldfixthetearing. Tearingartifactisbecauseofalargetemporaldifferencebetweenadjacentpixelrows.Foroneformofsoftwaretearing(VSYNCOFF),it’sanewframewithitsrespectivesuddenincreaseingameclock,essentiallysplicedmid-scanoutontheGPUtransmission(signal/cable)thatisnotsynchronizedbetweenrefreshcycles—seetearingdiagramcreatedby432framespersecondon144Hz.Foroneformofhardwaretearing,it’sasuddentemporaldifferenceinpixelrefreshingbetweenadjacentrows. Forallformsoftearing(softwaresideandhardware-designside)itiscausedbythelargetimeinterval(oftherenderoroftherefresh)inthepixelrowsabove/belowthetearline.Soitisabigprobleminconcurrentscanouts(i.e.treatingthescreenasmultiplesubdividedscreens)unlessmitigated. Regardless,eveniftearingisfixed—concurrentscanoutsneverhelpscanskewing,asconcurrentscanoutsdon’tspeedupthescanoutsweep.Infact,itcanmakescanoutskewingworse.Concurrentscanoutsofa60Hzscreenmeansthescanoutsweepwouldbe1/30sec(1/60secfortophalf,thencascadedtobottomhalfforanother1/60sec,foragrandtotalof1/30secinaslowerscanoutsweep).Inotherwords,multiscanningcanworsenscanskewbecauseyou’rereducingtheglobaltop-to-bottomscanoutsweepvelocityviathenarrower-heightstackedequivalentsofmultiplescreens. Tounderstandbetterhowscanoutsubdivisioncreatesslowerscanouts: Althougheachsubdividedscreenslice(1/8thheight)isscanned-outin1/960sec,the8slicesmeansatotaltop-to-bottomglobalsweepis1/120secinsteadof1/960sec.SoscanoutsubdivisionismainlyusefulforfutureultrahighHzscreensmadeofmodulartechnology(likeanarrayofmanytinyscreens).Jellyeffectwouldbesuper-nastyfor60Hzand120Hzscreensdoingconcurrentscanouts. Somescreensnecessarilyusesubdividedmini-screens(e.g.modularLEDscreenslikeJumbotrons),andmanyofthoseLEDJumboTron/Daiktronicsscreenmodulesconcurrentlyrefreshat600Hz-1920Hz(refreshinga60Hzrefreshcyclebetween10to32times)asaskew/tearing/sawtoothingsolution. Themodular“equivalentofmanystitched-togethertinyscreensrefreshingsimultaneously”natureofgiantLEDscreenscanalsointheorybecommandeeredtocreateanative600Hz-1920Hzframerate+refreshrateasamotion-blur-eliminationmethod(bruteHztocreatelow-persistencesample-and-holddisplay)withsomerelativelyminormodificationstotheelectronics—forultrahighrefreshrategiantscreensinthefuture.Sothemodularityactuallyhelpsthetechnologicalwork-subdivisionnecessarytomakeultrahighrefreshratespossibleinanon-unobtainiumway. Multiscanningisaknownshortcuttoachievingretinarefreshratesusingtoday’stechnology.Retinarefreshrates(wherethediminishingHzcurveofreturnsnolongerderivehumankindbenefits)arescientificallywellinexcessof1000fps1000Hzforawide-FOVhighresolutionsample-and-holddisplayduetotheViciousCycleEffectofhigherresolutionsandwiderFOVamplifyingrefreshratelimitations(fromeithermotionblureffects,stroboscopicsteppingeffects,orothermotionartifactsgeneratedfromafiniterefreshratedemonstratedbyTestUFOPersistencewhichonlystopslookinghorizontallypixellatedatultrahighrefreshrates).ThisiswellknowninVR,butalsoappliestoFOV-fillingwallsizedphysicalscreens,someofwhichalsohappenstobebuiltmodularlikemanytinyscreensrunningconcurrentscanouts. Globalrefreshdisplaysinthepaststillrequiredsequentialbehaviours(e.g.multiplefastlow-color-precisionscanoutslikeplasma/DLP,orthingslikerefreshing/primingpixelsinthedarkbeforeadelayedscreenilluminationpulse).Soallpastlow-Hzglobalrefreshscreensalwayshaveworselagand/orworseartifactsthangoodLCDscreens. Metaphoricaltechnologicalthoughtexercise:Ifascreenhastrulyglobalrefreshedinstantlywithoutlag,thenit’sfullyidlingbetweenrefreshcycles—thenaskoneself,whyisthescreen60Hzinsteadofahigherrefreshratesuchas120Hz?Inotherwords,askoneself,whynotputanidlingscreentomoreworkdoingmorerefreshcycles?Rheoreticalquestiontoaskoneself.ModerncolorprocessingonGPUscangothousandsofframebufferspersecondnowadaysevenonamidrangedesktopGPUfor2Dgraphicssuchasscrollingorpanning.EvenmodernApple/SamsungmobileGPUscannowbrowser-scrollatafewhundredofcompositedframebufferspersecond,andstillonlyhit50%GPUutilization(forscrollingoverhead)onaretinascreenforsimplescrollingofrender-lightcontentlikestaticwebpages.Thepointbeing,there’snozero-lagglobalrefreshdisplays.Otherwise,we’dbemilkingtheextraHzsoonermorecheaply.Theonlytruezero-lagdisplaysaredisplaysthatarestreamingthesignaldirectlyontothescreenlikeaCRTinafullyrastersynchronizedmanner,andthat’sneveraglobalrefresh.Globalrefreshdisplayshistoricallyalwayshadmorelagthanevena60HzCRTormodernlow-lagfast-GtG60HzLCD. HighHzsinglescanoutisnowultimatelytheholygrailmethodofapproachingglobalrefreshinacompromises-freeway(lowlag,andveryfastscanoutsthatbehaveasdefactoglobal-refresh).Combiningglobal-refreshandlow/zero-lag,requiresadisplaytodoallofitsprocessingduringthepracticallyglobalrefresh.Sothescreeniseffectivelyidlingbetweenrefreshcycles.Atthatstage,it’salmostnoextracosttoallowthedisplaytoatleastoptionallybeabletoacceptmorerefreshcycles.Naturally,theconverseisalsotrue:Ifyou’vealreadydesignedahigh-Hzsingle-scanoutscreen,it’salreadydefactoclosertoalow/zero-lagglobalrefresh. Currently,themostscanout-motion-artifact-freetechnologies(ignoringghostingdifferences)arehighrefreshrateLCD,OLED,LED,MiniLEDandMicroLEDdisplaysthatdoasingle-passscanoutperrefreshcycle: Pixelsrefreshdirectlytofinalpixelcolorwithnotemporaltricks. Fixesrainbows/noise/contouringartifacts Refreshinguseonlyonescanoutsweepforminimumtemporaldifferencebetweenadjacentpixels,whilealsosynchronizingframerate=Hz. Fixestearingartifacts Highrefreshratesmeansanultrafastscanoutsweep(fromthepanel’sshorterrefreshintervalcapability). Fixesscanskewing(jellyeffect)—evenforworstjellyeffectvisibilityconditions HeadofBlurBusters-BlurBusters.com|TestUFO.com|Follow@BlurBustersonTwitter        TosupportBlurBusters:        •OfficialListofBestGamingMonitors        •ListofG-SYNCMonitors        •ListofFreeSyncMonitors        •ListofUltrawideMonitors Top PostReply Printview Display:Allposts1day7days2weeks1month3months6months1year Sortby:AuthorPosttimeSubject Direction:AscendingDescending 9posts •Page1of1 Returnto“Area51:DisplayScience,Research&Engineering” Jumpto TheMainLobby ↳ General—Displays,Graphics&More ↳ News/Rumors/Conventions ↳ InputLag/DisplayLag/NetworkLag ↳ ElectricalIssuesAffectingLatency--Interference,EMI,EMF EverythingBetterThan60Hz™ ↳ EliminatingMotionBlur—LightBoost/ULMB/ELMB/DyAc ↳ G-SYNC ↳ FreeSync ↳ XBOXConsoles—ONE/SeriesS/SeriesX ↳ BENQZowieTweaking—StrobeUtility/BlurReduction/DyAc ↳ DisplayOverclocking TheLaboratory ↳ TestUFOMotionTests ↳ Area51:DisplayScience,Research&Engineering ↳ SoftwareDevelopers/Low-LagCode/GameProgramming ↳ PursuitCameraTestingofDisplayMotionBlur ↳ HighFrameRateVideo(HFR,UltraHFR,120fps,240fps,1000fps) TheCafeteria ↳ OfftopicLounge ↳ ForumSystem/Registration/Suggestions BlurBustersForumsBoardindex AlltimesareUTC-05:00 Deletecookies Contactus PoweredbyphpBB®ForumSoftware©phpBBLimited Privacy | Terms BlurBusters.com TestUFOMotionTests BlurReductionFAQ ListofGamingMonitors Twitter Facebook Privacy About