天津职业大学 文本 Performance and Comfort on Near-Eye Computer Displays_眼屈光学

WT0T5saW红 C®2Amad时wOe时 ORIGINAL ARTICLE Performance and Comfort on Near-Eye Computer Displays JAMES SHEEDY,OD,PhD,FAAO and NEIL BERGSTROM,PhD The Olis State Lniveri.College ofOproweir Colba OH (S)InVis Carponion.Sayalt.CA (NB) ABSTRACT:Rackground.Very small high-resolution displays (SVGA,800 x 600 pixels)worn near the eye and imaged to create a virtual image have potential as alternatives to traditional computer displays.Methods.Twenty-two subjects performed text-based tasks on five displays:monocular virtual,binocular head-mounted virtual,hard copy,flat panel,and a small format portable display.Outcome measures included performance speed,symptoms,visual acuity,and heterophoria. In a second experiment,subjects performed a proscribed routine of head and body movements designed to elicit motion. related symploms.Resus.Performance speed on monocular virtual was generally cumparable with performances on flat panel and hard copy.Overall,performance speeds on the binocular virtual display were about 5%slower than normalized performances,6.75%slower compared with the traditional flat panel and hard copy displays.Symptoms of eyestrain and blurry vision were significantly higher on monoculr virtual than on other displays.No significant changes in visual acuity or heterophoria occurred with amy of the displays.Motion-related symptoms with the head mounted near-eye display were not significantly difterent than with other displays tested.Conclusions.Pertormance and comtort on the near-eye displays in this study was more similar to traditional displays than in many previous studies with head mounted displays.This is likely due to lack of task movement,partial instead of full immenion,better display resulution,and concordance of the accommodative and vergence stimuli.(Optom Vis Sci 2002;79:306-312) Key Words:head mounted displays,vision perfarmance,sympeoms,cumputer displays,reading,visian fery small high-nolin diprysa now availahle that cicstreading and iry invld rypogahi: are too small for direct viewing but are viewed through factoes in the printing industry and the effocts oighting. optical systems that crcare a lrpe viewable virtual image of Reading performano on compuner displrys has al boen the digply.Such a virmml display can he prexrnted m ane cye in a sivdly meaemd Farlier snadieextahlishal that nading and proce. hand-held device,or rwo sch displays can be mounted in a spec- reading from acthod my tube(CR'T display was 20tu30%slower tack-typ:carrier.a form of head moantod display (HMD).Such than perfoming the smme task on primadl puper.The CKI had grcin digplys can enabl ter to view common full-sized eleerronie files lumincs characters cn a daek bckgrnund and the hom copy wns such as weh pages and word peocessing documens if the pid peinted on poper and hand hed Au count is adeqaee.A ner-eye virtl diplry has poeentiall m he a tempted sn isbte the variahe ropombl fur the performanced reble funcional alterthe to reading test from hard copy or ence and ccecluded it was the reult of a coembination of vrhle cradicional computer displays that are viewed in real space.This with img qualiry being paticuarly imporant potential mocivated the aarmnt stikly. Imprening the quliry of an dectmne display can alsn improwe Theprpose ofthe cument srudy is to measure human performnce performance.Increased pisel densiry on a CRT screen resalted in and comfort on high-cesdlurion (SVGA 800 X GUU pies)near eye significantly faster reading speed and also fewer measured eye virual displys.Performance and comfoon paragraph reading.kter reated sympomsAnother way to improwe character appearance count,and woud search tasks is coenpared with performance on hard is to use gray pixels instead of all black and white pixds.the bas copy.at pand,and a reduced format handheld computer dsplay. for this is the perceived displacement of a border by a thin gray Viewing conditions arequlid o the exent possible but alkw for strip.The use of a gray ale has been shown to significanthy the fundmenral diffirenits betwoen the dispby types improve realling pcrlurmano and visual cumfor,* Reading text is one of the most imporcant human activicies and is Near-eye displays,the sabject of the current imestigaion,have eential foreducation,wurk,and rocreatiun.The overall visal nliy seweral diffrenes compared with traditional computer displyx. of the tear can aff human nading performance and comfor.The Traditional dlisplays are constraimed by their physical size and arr (iyrewelry awl Vidon Seinsre,Vul.75.No.5.Mar 2002 CopyrightAmerican Academy of Optometry.Unauthorized reproduction of this artcle is prohibited

ORIGINAL ARTICLE Performance and Comfort on Near-Eye Computer Displays JAMES SHEEDY, OD, PhD, FAAO and NEIL BERGSTROM, PhD The Ohio State University, College of Optometry, Columbus, OH (JS), InViso Corporation, Sunnyvale, CA (NB) ABSTRACT: Background. Very small high-resolution displays (SVGA, 800
600 pixels) worn near the eye and imaged to create a virtual image have potential as alternatives to traditional computer displays. Methods. Twenty-two subjects performed text-based tasks on five displays: monocular virtual, binocular head-mounted virtual, hard copy, flat panel, and a small format portable display. Outcome measures included performance speed, symptoms, visual acuity, and heterophoria. In a second experiment, subjects performed a proscribed routine of head and body movements designed to elicit motion￾related symptoms. Results. Performance speed on monocular virtual was generally comparable with performances on flat panel and hard copy. Overall, performance speeds on the binocular virtual display were about 5% slower than normalized performances, 6.75% slower compared with the traditional flat panel and hard copy displays. Symptoms of eyestrain and blurry vision were significantly higher on monocular virtual than on other displays. No significant changes in visual acuity or heterophoria occurred with any of the displays. Motion-related symptoms with the head mounted near-eye display were not significantly different than with other displays tested. Conclusions. Performance and comfort on the near-eye displays in this study was more similar to traditional displays than in many previous studies with head mounted displays. This is likely due to lack of task movement, partial instead of full immersion, better display resolution, and concordance of the accommodative and vergence stimuli. (Optom Vis Sci 2002;79:306–312) Key Words: head mounted displays, vision performance, symptoms, computer displays, reading, vision Very small high-resolution displays are now available that are too small for direct viewing but are viewed through optical systems that create a large viewable virtual image of the display. Such a virtual display can be presented to one eye in a hand-held device, or two such displays can be mounted in a spec￾tacle-type carrier, a form of head mounted display (HMD). Such displays can enable users to view common full-sized electronic files such as web pages and word processing documents if the pixel count is adequate. A near-eye virtual display has potential to be a reasonable functional alternative to reading text from hard copy or traditional computer displays that are viewed in real space. This potential motivated the current study. The purpose of the current study is to measure human performance and comfort on high-resolution (SVGA, 800
600 pixels) near-eye virtual displays. Performance and comfort on paragraph reading, letter count, and word search tasks is compared with performance on hard copy, flat panel, and a reduced format handheld computer display. Viewing conditions are equalized to the extent possible but allow for the fundamental differences between the display types. Reading text is one of the most important human activities and is essential for education, work, and recreation. The overall visual quality of the text can affect human reading performance and comfort. The earliest studies on reading and text quality involved typographical factors in the printing industry1 and the effects of lighting.2 Reading performance on computer displays has also been exten￾sively measured. Earlier studies3, 4 established that reading and proof￾reading from a cathode ray tube (CRT) display was 20 to 30% slower than performing the same task on printed paper. The CRT had green luminous characters on a dark background and the hard copy was printed on paper and hand held. A subsequent investigation5 at￾tempted to isolate the variable responsible for the performance differ￾ence and concluded it was the result of a combination of variables, with image quality being particularly important. Improving the quality of an electronic display can also improve performance. Increased pixel density on a CRT screen resulted in significantly faster reading speed and also fewer measured eye￾related symptoms.6 Another way to improve character appearance is to use gray pixels instead of all black and white pixels, the basis for this is the perceived displacement of a border by a thin gray strip.7 The use of a gray scale has been shown to significantly improve reading performance and visual comfort.8 Near-eye displays, the subject of the current investigation, have several differences compared with traditional computer displays. Traditional displays are constrained by their physical size and are 1040-5488/02/7905-0306/0 VOL. 79, NO. 5, PP. 306–312 OPTOMETRY AND VISION SCIENCE Copyright © 2002 American Academy of Optometry Optometry and Vision Science, Vol. 79, No. 5, May 2002

Near-Fye Compuner Displany's-Sheedly and Berpstmom 307 typically viewed at 50 to 70 cm from theeyes.The viewing distance 22±5.明particip四ted in the study..Subjects signed a comsent form of the virtual image in aneaye disply depends upon the opical appruved by a human subjects review board. design and can be selecced by the designer.Most commonly,the Subjecrs were askod to perform a set of trials.cach set comprised viewing distance of the near-eye virtl image is significantly four rf praph readling tak,four kletercin longer thain for traditional computer display&langer viewing dis- tuk.and four triak ofa woedl search task.The entiee sct of trials wap tancs have been shown tobe fivurable to visal comfort,suppos perfommed for cach of fre display conditioe.The five display cndi- odly bceausc of dencased acommodative and/or oalr comer- tions were mancear virtual,binooubr viral,flat pnd,hand copy. gonce demand.However,the mechanical alignment of the two and quartr VGA (QVGA).The tesring ordar of the five displny on- optical systems in a biooculr near-eye disply is a critical con- dcions w rndomiaed to nepate order and practice effecs.All ex straint that is not a factor for traditional real-world displays. was presemed in blackc-on-whice Times Romn 12-point foor.Test- A large difference berween virtual (near-eye)and traditional dis ng was perceded by oeientarion trils of the tasks lays is that the virtual display is not fixed in the environmenst.The The monoculr virtual disply wiss a hand-beld virtual SVGA binocular HMD)moves with head mowvement.as doexa monocalar (800 X 60 pisels)display (e-ce by InViso,Sumnyvale.CA)with hand beld display if beld firmly before the ere.This causes discor an occhuder for the nonviewing eye.The image dsance was 90 cm. dance between the visual motion signals and the propriocepcive Subjects oould select either eye to view the screen were seated at and vestibular motion signals.This has been shown to cause symp desk,and able to rest their elbow on the desk for support.The exit toms of motion sickness in several previous studies ofIIMDs- pupil of the display system was ellipeical,horixontal and vertical In most of those srudies,several subjects were unable to finish the dimensions were 11.25 X 6.5 mm at 25 mm behind the back plane trials because of symptoms.Eye and vision-rcted symptoms have of the disply.The binocular virtual disphy comprised SVGA abo been reported in sveral of the studies,as were displays with one before each eye in a universal fit head-borne changes in vision measures after HMD trials.Although holder similar sacks (e-shades by InViso).The image dis these studies have raised considerable oonoern about the safety and tanxe of each display and the comvergence angle between them was efficacy of HMDs,other studies have not shown significant per 175 cm.The exit pupil sice of each display in binocular virual was formanoe.sympeam,or vision changes with HMD.supposedly 7.5 X 7.75 mm (Hx V)at 23 mm.The flt pand displary was an beeas of better doign. active matrix ICD-XGA 15"display (VDP 150,Viewsonic Corp. The current study invesrigated performance and comfort on a Walmt.CA)driven in an SVGA window to cnabk a nonscalod pre- hand-beld monocular ncar-cyc dispby and a hcad mounted bin- sencion of an SVGA img.The hard copy digly sumpriscd oculr ncar-cye display.The near-eye displays and the tasks ud in peinced teet (300 dots per inch (dpi)bser printer)on whine paper this srudy differ in several significant ways from those in the above- located on a dant board.Both the flat panel and hard copy displys mentioned studies.Most of those studies used full immer- were viewed onadesk againsta neutral wall of6i od/m2.Subjects had sion6 1a.21 (no real environment seen).and most also used free head movement when viewing the flat panel and hard copy dis tasks with moving ohjeeis and/or hody tacking of some soet playsand the seaing/desk cnnfiguration wsarngd so the viewing create a virtual environment with which the subject interact- dstanee was centrrr on 58and 56cm,rexpectively.to accomplish the d.No menement ar tracking was used in the same angnlar xize of the monncdar virl disply.The QVGA (320 current sudy.and the pear eye devices allowed some peripheral x 240 pixcels)device was an iPAQ Pocket PC (Compaq.Hous.on. view of the real worl,therefore,subjects were only partially im TX)Subjects beld the QVGA by hand while seared at a desk and mersed in the virtual environment.Also,because of advanced tech coud sect comforable viewing distance.generlly about 12 to 16 nology.the daplys used in the present srudy had significantly inches.Table I peovides compsof primay display feaes. berter optical images [800 X 600 podls.high modulirion transfer Foe the four paragraph reading tials,subjects were insructedto tunttion optics)than previous studies.They also had rdativehy large quicdly read short story segments of approximately 325 words, exit pupils For the hand-beld monocular display used in this study. each story followed by three to four multiple-choice questions.The this attribute ebled good flexibility in properly holding the erye questions served to normalize subject attention to the task:answers device in front of the eye.For the binocular near eye display,this were not used as an outcome measure. enhled the use ofa uivers fiing frame ind one fixed inerdisplay Four letter counting trials immediately followed the four par distance to he used for aang of interpupillry distames as in a pno vious study TABLE 1. Summary of display characteristics. METHODS Viewing Angular Sice Experiment 1 Display Pixels (od/m2) Distance (diagonal icmi degrcesl Subjeers were rreruited hy newspaper adverisements an college campuses.They were sremed to meet the follnwing criteri:t Morocular virtual 109 90 34 80X600 least 2125 visnl acuiry in each eye unoorrected or with contact Binocular virtual 70 175 32 00×00 lenses,interpupillary distance of 61 to6mm (calculated accept- Flat panel 70 0 34 800×600 able range for binocular near-eye device).no presbyopia,nohistory Hard copy 70 56 34 300i of significant eye pathology.and no current medications ather Quarter VGA 40 varisble variable 320 x 240 than birth control.Twenry-two subjects (ages 18-59 years.mean aViA.64×40plx. OpmE切ad Vraian Srimre,V化79Na5Ma球2002 Copyright American Academy of Optometry.Unauthorized reproduction of this article is prohibited

typically viewed at 50 to 70 cm from the eyes. The viewing distance of the virtual image in a near-eye display depends upon the optical design and can be selected by the designer. Most commonly, the viewing distance of the near-eye virtual image is significantly longer than for traditional computer displays. Longer viewing dis￾tances have been shown to be favorable to visual comfort,9 suppos￾edly because of decreased accommodative and/or ocular conver￾gence demand. However, the mechanical alignment of the two optical systems in a binocular near-eye display is a critical con￾straint that is not a factor for traditional real-world displays. A large difference between virtual (near-eye) and traditional dis￾plays is that the virtual display is not fixed in the environment. The binocular HMD moves with head movement, as does a monocular hand-held display if held firmly before the eye. This causes discor￾dance between the visual motion signals and the proprioceptive and vestibular motion signals. This has been shown to cause symp￾toms of motion sickness in several previous studies of HMDs.10–16 In most of those studies, several subjects were unable to finish the trials because of symptoms. Eye and vision-related symptoms have also been reported in several of the studies,10, 12, 13, 16 as were changes in vision measures after HMD trials.11, 13, 16–19 Although these studies have raised considerable concern about the safety and efficacy of HMDs, other studies have not shown significant per￾formance, symptom, or vision changes with HMD, supposedly because of better design.20, 21 The current study investigated performance and comfort on a hand-held monocular near-eye display and a head mounted bin￾ocular near-eye display. The near-eye displays and the tasks used in this study differ in several significant ways from those in the above￾mentioned studies.10–21 Most of those studies used full immer￾sion10–16, 18, 21 (no real environment seen), and most also used tasks with moving objects and/or body tracking of some sort to create a virtual environment with which the subject interact￾ed.10, 11, 13–17, 19–21 No movement or tracking was used in the current study, and the near-eye devices allowed some peripheral view of the real world, therefore, subjects were only partially im￾mersed in the virtual environment. Also, because of advanced tech￾nology, the displays used in the present study had significantly better optical images (800
600 pixels, high modulation transfer function optics) than previous studies. They also had relatively large exit pupils. For the hand-held monocular display used in this study, this attribute enabled good flexibility in properly holding the near-eye device in front of the eye. For the binocular near-eye display, this enabled the use of a universal fitting frame and one fixed interdisplay distance to be used for a range of interpupillary distances as in a pre￾vious study.21 METHODS Experiment 1 Subjects were recruited by newspaper advertisements on college campuses. They were screened to meet the following criteria: at least 20/25 visual acuity in each eye uncorrected or with contact lenses, interpupillary distance of 61 to 66 mm (calculated accept￾able range for binocular near-eye device), no presbyopia, no history of significant eye pathology, and no current medications other than birth control. Twenty-two subjects (ages 18–39 years, mean 22 5.9) participated in the study. Subjects signed a consent form approved by a human subjects review board. Subjects were asked to perform a set of trials; each set comprised four trials of a paragraph reading task, four trials of a letter counting task, and four trials of a word search task. The entire set of trials was performed for each of five display conditions. The five display condi￾tions were monocular virtual, binocular virtual, flat panel, hard copy, and quarter VGA (QVGA). The testing order of the five display con￾ditions was randomized to negate order and practice effects. All text was presented in black-on-white Times Roman 12-point font. Test￾ing was preceded by orientation trials of the tasks. The monocular virtual display was a hand-held virtual SVGA (800
600 pixels) display (e-case by InViso, Sunnyvale, CA) with an occluder for the nonviewing eye. The image distance was 90 cm. Subjects could select either eye to view the screen, were seated at a desk, and able to rest their elbow on the desk for support. The exit pupil of the display system was elliptical, horizontal and vertical dimensions were 11.25
6.5 mm at 23 mm behind the back plane of the display. The binocular virtual display comprised SVGA displays with one before each eye in a universal fit head-borne holder similar to spectacles (e-shades by InViso). The image dis￾tance of each display and the convergence angle between them was 175 cm. The exit pupil size of each display in binocular virtual was 7.5
7.25 mm (H x V) at 23 mm. The flat panel display was an active matrix LCD–XGA 15” display (VDP 150, Viewsonic Corp, Walnut, CA) driven in an SVGA window to enable a nonscaled pre￾sentation of an SVGA image. The hard copy display comprised printed text (300 dots per inch (dpi) laser printer) on white paper located on a slant board. Both the flat panel and hard copy displays were viewed on a desk against a neutral wall of 60 cd/m2 . Subjects had free head movement when viewing the flat panel and hard copy dis￾plays, and the seating/desk configuration was arranged so the viewing distance was centered on 58 and 56 cm, respectively, to accomplish the same angular size of the monocular virtual display. The QVGA (320
240 pixels) device was an iPAQ Pocket PC (Compaq, Houston, TX). Subjects held the QVGA by hand while seated at a desk and could select a comfortable viewing distance, generally about 12 to 16 inches. Table 1 provides comparison of primary display features. For the four paragraph reading trials, subjects were instructed to quickly read short story segments of approximately 325 words, each story followed by three to four multiple-choice questions. The questions served to normalize subject attention to the task; answers were not used as an outcome measure. Four letter counting trials immediately followed the four para￾TABLE 1. Summary of display characteristics. Display Luminance (cd/m2 ) Viewing Distance (cm) Angular Size (diagonal degrees) Pixels Monocular virtual 109 90 34 800
600 Binocular virtual 78 175 32 800
600 Flat panel 70 58 34 800
600 Hard copy 70 56 34 300 dpi Quarter VGAa 40 variable variable 320
240 a VGA, 640
480 pixels. Near-Eye Computer Displays—Sheedy and Bergstrom 307 Optometry and Vision Science, Vol. 79, No. 5, May 2002

30 B Near--feC军uter Displays一小eedy andl Bergstrom graph reading trials.For each ktter oounting trial,subjects were the subject saw a vertical red line(penlight through Maddox Rod) asked to count the occurrences of an assigned search leter in a with ooe eye and the card with markings with the otber eye.The pargraph of nonserse words (all capital letters)randomly gener subject reported the number on the scale through which the red atod and organizcd in a five-linc paragph.The scanch ler was line appead to be alignod.This was rccorded as the hctcrophori. aigned at the beginning of cach trial and sdected frum a s of letters (D.E.F,H.N,P,R.U,V,2 cach with similar visibility. The search letter occurod 20 to 30 times in cach random para- Experiment 2 graph.The outcome measure was performance time. The purpode of experiment 2 was to place subjeces at risk foe Four-word search trials immedzntely stacceeded the four-letter motion-related symptoms while engiged in specific body move- counting trials For each word search tial,subjeces were aed to ment and while wearing the binoenlar virtl HMD.The sime 22 find three o of fouroccurrences of an assigned three-letter search subjects performed experiment 2 immediaely following experi ward in a 20 X 20 grid ereated in an Exrel spreadshert.70%afthe ment 1. oells wrre oocupied hy a thrreeter wond.Four of the cells con- Subjects performed the same set of body movements with and cained the search word,the subject searched until three occurreces withour wearing,the binocular viru display in altemnating order. were properly identified.For each search word identified,the sub A paragraph of ter was displayed on the screen during testing. ject vebally reported the row (mumber)and column (letter)of the Subjects were instructed to indicate verbally when and if,at any cell. time during the teing they began to experience an ines in any Foe all three tasks performance time was the outcome measure symptoms and/or they could Do longer continue testing because of and was measured in secoods with a stopwatch.Hands were not symptoms. allowed as visual guikes during testing Testing order of the task Body movements comprised the following strps files (e.g each specific paragraph reading file)and the display on 1.While sitting rotate the head 45 left and right six times cach which each was used ws asigned differenty across subjects to with a 1 s cadencc called by the experimenter,then rotate the oqualize order.practice,and story difficulty effects.The QVGA head up and donn sis times similar to the above. disphy had more froquent line wrap hocause uf fower pixcls than 2.While standing:six identical kft/right head mowements,six the ther displys,therefore paragraph rcading and r oounting identical up/down head movements,6 left/right body rocations had more lincs for cach trial despite showing the same text.The af with a 3 s cadence. word search task was not performed on the QVGA diply because 3.While sitting:six clnckwisefcounter-dockwise head molls an the search grid cosumed more than on pape. shoulders with a 3 s cadence. Ar the end of testing sach display condition (four trals of each 4.While ading six identical dockwiselcoumter-clockwis hrad task),the subjecr rated the magitde ofeach ofthe following nine symptome beadache,cyestrain.sore or irritated eyes,bhrry vidon, drnessn disoricntatinn,neck ache,and hackache.Symp- Symptoms were measured wich the same sympcom question- naire from experiment I whenever increised symptoms were re tom magnitudes were registered on an anakg scale proented as a pored and also at the end of testing sequences with and without 100 mm boricntall linc.The scale was labded"None'and "Se wearing the binoculae virtual display. vere"at the ends and"Mild,""Moderate,"and"Objectionble"st qurtile locations on the line.The subjecr marked a locarion along he line1 he pociti始representing the symptom magnitude解 RESULTS that moment.The location of the drawn line was measured to translate the symptom magnitde ro an integer value from to Performance Speed 100. Pertormance time data (s/task)were converted to logarithmic Binocular visual acuity using Bailey Lovie acuity charts wis values becmuse performance differences between display cooditions abo measured after trial secs on each display.Two different charts are most likely to be in equal ratios between subjects rather than were used.and viewing distances of 16,20,25,and 32 feet were equal linear time differences between subjects.For each subject/ rndomized to impede chart memorizstion.The visual acuiry display condition,logarithmis of tbe perfonance times for the four charts had five kttens per rows cach measurement of acuiry in- trials were averaged to establish a mean performance time.Forcach duded rows that could be cntirchy identified and also rows in which subiect,a normalized pertormance time for cach of the three tasks o letters could be identified.Each properly identified leter wis was established by averaging the mean performance times acros recorded and added to an acuity score based upon 100 ketters the tive display conditions.For each subject/display condition,the representing visual acuity of 20720 (logarithm ot the minimum mean performance time was subtracted from the pormalized per angle of resolution [ogMAR]0.00). formanee time,resulting in a normalized performance dfference Measurements of hinoatlr alignment (hetrmphnrial at 40 cm with an imene scale,i.c poitive values mprcsentd shoner times. and 3 m wene also made at the cnd af testing cach display.Heten The anti-log of the normalized performancc differenoe was calu- phoria was measundby placingan appropriatcly oriented Maddox lated and scrved as the final mcaxure of perfurmance.Final unit Rod befure one cyc whil the subjoct viewd a card with a scak are task/s and,bccause the tasks were of oqual kngth,the final calibeated in A of exophora (ourward deviation)and esopbori meure is directy relred to perfoemnce speed.Noemlized (inwand devition)appropriace to the testing disrances of 40 cm mean perfomunce sperds,with standand deviations,for each dis and 3m.The zero position (orthophor)in the card had a hole in playitask are shown in Fig.1. it hehind which a penlight painted toward the subjeer.As a resule, Performance speed on hinoculr virtl was skrwer from 4.7 to Optometry ad Viioe Srimnre,Vc79Na气Mar200】 Copyright American Academy of Optometry.Unauthorized reproduction of this article is prohibted

graph reading trials. For each letter counting trial, subjects were asked to count the occurrences of an assigned search letter in a paragraph of nonsense words (all capital letters) randomly gener￾ated and organized in a five-line paragraph. The search letter was assigned at the beginning of each trial and selected from a set of letters (D, E, F, H, N, P, R, U, V, Z) each with similar visibility. The search letter occurred 20 to 30 times in each random para￾graph. The outcome measure was performance time. Four-word search trials immediately succeeded the four-letter counting trials. For each word search trial, subjects were asked to find three out of four occurrences of an assigned three-letter search word in a 20
20 grid created in an Excel spreadsheet. 70% of the cells were occupied by a three-letter word. Four of the cells con￾tained the search word, the subject searched until three occurrences were properly identified. For each search word identified, the sub￾ject verbally reported the row (number) and column (letter) of the cell. For all three tasks performance time was the outcome measure and was measured in seconds with a stopwatch. Hands were not allowed as visual guides during testing. Testing order of the task files (e.g., each specific paragraph reading file) and the display on which each was used was assigned differently across subjects to equalize order, practice, and story difficulty effects. The QVGA display had more frequent line wrap because of fewer pixels than the other displays, therefore paragraph reading and letter counting had more lines for each trial despite showing the same text. The word search task was not performed on the QVGA display because the search grid consumed more than one page. At the end of testing each display condition (four trials of each task), the subject rated the magnitude of each of the following nine symptoms: headache, eyestrain, sore or irritated eyes, blurry vision, dizziness, nausea, disorientation, neck ache, and backache. Symp￾tom magnitudes were registered on an analog scale presented as a 100 mm horizontal line. The scale was labeled “None” and “Se￾vere” at the ends and “Mild,” “Moderate,” and “Objectionable” at quartile locations on the line. The subject marked a location along the line at the position representing the symptom magnitude at that moment. The location of the drawn line was measured to translate the symptom magnitude to an integer value from 0 to 100. Binocular visual acuity using Bailey-Lovie acuity charts22 was also measured after trial sets on each display. Two different charts were used, and viewing distances of 16, 20, 25, and 32 feet were randomized to impede chart memorization. The visual acuity charts had five letters per row; each measurement of acuity in￾cluded rows that could be entirely identified and also rows in which no letters could be identified. Each properly identified letter was recorded and added to an acuity score based upon 100 letters representing visual acuity of 20/20 (logarithm of the minimum angle of resolution [logMAR] 0.00). Measurements of binocular alignment (heterophoria) at 40 cm and 3 m were also made at the end of testing each display. Hetero￾phoria was measured by placing an appropriately oriented Maddox Rod before one eye while the subject viewed a card with a scale calibrated in  of exophoria (outward deviation) and esophoria (inward deviation) appropriate to the testing distances of 40 cm and 3 m. The zero position (orthophoria) in the card had a hole in it behind which a penlight pointed toward the subject. As a result, the subject saw a vertical red line (penlight through Maddox Rod) with one eye and the card with markings with the other eye. The subject reported the number on the scale through which the red line appeared to be aligned. This was recorded as the heterophoria. Experiment 2 The purpose of experiment 2 was to place subjects at risk for motion-related symptoms while engaged in specific body move￾ment and while wearing the binocular virtual HMD. The same 22 subjects performed experiment 2 immediately following experi￾ment 1. Subjects performed the same set of body movements with and without wearing the binocular virtual display in alternating order. A paragraph of text was displayed on the screen during testing. Subjects were instructed to indicate verbally when and if, at any time during the testing, they began to experience an increase in any symptoms and/or they could no longer continue testing because of symptoms. Body movements comprised the following steps: 1. While sitting: rotate the head 45° left and right six times each with a 1 s cadence called by the experimenter, then rotate the head up and down six times similar to the above. 2. While standing: six identical left/right head movements, six identical up/down head movements, 6 left/right body rotations of 90° with a 3 s cadence. 3. While sitting: six clockwise/counter-clockwise head rolls on shoulders witha3s cadence. 4. While standing: six identical clockwise/counter-clockwise head rolls. Symptoms were measured with the same symptom question￾naire from experiment 1 whenever increased symptoms were re￾ported and also at the end of testing sequences with and without wearing the binocular virtual display. RESULTS Performance Speed Performance time data (s/task) were converted to logarithmic values because performance differences between display conditions are most likely to be in equal ratios between subjects rather than equal linear time differences between subjects. For each subject/ display condition, logarithms of the performance times for the four trials were averaged to establish a mean performance time. For each subject, a normalized performance time for each of the three tasks was established by averaging the mean performance times across the five display conditions. For each subject/display condition, the mean performance time was subtracted from the normalized per￾formance time, resulting in a normalized performance difference with an inverse scale, i.e., positive values represented shorter times. The anti-log of the normalized performance difference was calcu￾lated and served as the final measure of performance. Final units are task/s and, because the tasks were of equal length, the final measure is directly related to performance speed. Normalized mean performance speeds, with standard deviations, for each dis￾play/task are shown in Fig. 1. Performance speed on binocular virtual was slower from 4.7 to 308 Near-Eye Computer Displays—Sheedy and Bergstrom Optometry and Vision Science, Vol. 79, No. 5, May 2002

Ner-Eye Compuler Displays-Sheexly and Bergslruen 309 115 TABLE 3. 110 Results from Friedman's test comparing symptom:score 16 acrass the 5 displays. 1r0 Symptom p Value Poct Hoc Results 00 Headache .5976 Eyestrain 0.0131 MV"significantly different (greater) 00 han all o国her display5. Q.31390 06 Blurry vision 0.0182 MV significantly differert igreater) 070 2 than FP. 0.8039 FR 103 DGs 0四 1.08 634 Dizzy Nausea 0.6515 =L0 DP6 1面 1四 10 0.0629 WS 181 50m Neck ache 04779 口甲 Rack ache 0.D44 FP significantly different (greaterl FIGUKE 1. than MV.HC and QVGA Nommalized performance speeds,mean and standard devations for 22 "MV,monocular virtual:BV.binocular virtual:FP,flat pancl; sd时t for tiree Lks purig年he4 dirg,[PRL ketler counting ILC]and HC,hard copy:QVGA,small lurmal display. word search IwsD on 5 displays Imonocular virtual IMVI.hinocuar iud,Lat panel F甲l,unl copy [Hi▣and smll fumat IQVGAl deplayi comparisons of the displays were performed using the method described by Conover.Resualts are in 'Table 3. 9.1%compared with fat panel,and 0.5 to 11.7%compared with Sympioms of eyestrain and blurry vision were significantiy hard copy.Performance ed on binocular virtul was slower by higher on monocular virtual than other displays.Backache symp an average of 6.75%h compared with the traditional flat panel and toms were significantly higher on flat panel compured with mast hard copy displarys Paragraph reading performanc was slenvet on other displays. QVGA.whereas lette-ounting performince was fastest on QVGA Vision Measurements Statistical tosting,using a 5 display x 3 task repeated measurs ANOVA,was performed on the averaged (four trials)normalized Across all cooditions.the mean visual acuiry measurement wis logarthm values.The p vallue foe an interaction between display -0.13 lopMAR.or 20/16 1.7 lerters on the 20/12.5 row.This and task was 0.06 Post-hoc comparions using Takey's method2 very high scuity average results from binocular meisurements and of multiple comparison found the following tsk paragraph read- screening for young healthy subjects with at least 20/25 visual ing.hard copy significantly faster than QVGA:task lemer count- xcuity in each eye.No trial order effects were discovered in ad hoc ing.no differences berween displays task word search,flat panel analysis o the visual acuity data.Across all conditions,the mean significantly faster than binocalar virtual. heterophoria measurements were 2.and 095 A of exophoria at 10 cm and3m,respectively.These values compare favorably with Symploms population normative data, Acity and hetcrophuria measurements were used to tot for The mcan symptom magnitude measurements for the 22 sub- changes associated with cach disply.For cach subjcct,a norma jects on cach display condition are shuwn in Table 2. ixed vision measure for cach of the five displys was otablished by The symptom data are not normally distributed because of averaging the mean measures ross the five display cooditions.For mny rating of zero.Friedmn's test was used to determine if any each subject/display condition,a reltive vision measure was erab- differences in mean score were present acres the five displys Ifa lished by determining the difference betwren the nomalored mea significant difference was detccted (p05),post hoe pair-wise sure and the individual measure.The mean and standard devia- TABLE 2. Mean symptom magnitudes (scale 0-100)alter series o perfoxrmance tasks on each display. ye Symo1t加me Mobion Symptom Musculoskeletal Dispay Headache Fyestrain Bluery Dizziness Nausea Disorient Neck Ache Back Ache 3.d 14.9 108 95 29 1.2 5.0 3.4 23 By b.2 10.5 7.7 8.5 3.0 2.4 1.8 2.4 25 FP 2.1 7.6 5.8 4.1 1.6 2.4 1.3 3.b 3.0 HC 3.4 8.5 5.9 67 23 1.3 3.9 1.6 00 LVGA 4.5 B.D 9.2 49 3.3 2.1 6.1 2.5 02 MV,monocular virtual:BV,hinocula virtual:FP.flat pane:HC.hard copy:QVGA,small format display. Copyright American Aca

9.1% compared with flat panel, and 0.5 to 11.7% compared with hard copy. Performance speed on binocular virtual was slower by an average of 6.75% compared with the traditional flat panel and hard copy displays. Paragraph reading performance was slowest on QVGA, whereas letter-counting performance was fastest on QVGA. Statistical testing, using a 5 display x 3 task repeated measures ANOVA, was performed on the averaged (four trials) normalized logarithm values. The p value for an interaction between display and task was 0.06. Post-hoc comparisons using Tukey’s method23 of multiple comparison found the following: task paragraph read￾ing, hard copy significantly faster than QVGA; task letter count￾ing, no differences between displays; task word search, flat panel significantly faster than binocular virtual. Symptoms The mean symptom magnitude measurements for the 22 sub￾jects on each display condition are shown in Table 2. The symptom data are not normally distributed because of many ratings of zero. Friedman’s test was used to determine if any differences in mean score were present across the five displays. If a significant difference was detected (p  0.05), post hoc pair-wise comparisons of the displays were performed using the method described by Conover.24 Results are in Table 3. Symptoms of eyestrain and blurry vision were significantly higher on monocular virtual than other displays. Backache symp￾toms were significantly higher on flat panel compared with most other displays. Vision Measurements Across all conditions, the mean visual acuity measurement was 0.13 logMAR, or 20/16  1.7 letters on the 20/12.5 row. This very high acuity average results from binocular measurements and screening for young healthy subjects with at least 20/25 visual acuity in each eye. No trial order effects were discovered in ad hoc analysis of the visual acuity data. Across all conditions, the mean heterophoria measurements were 2.4 and 0.95  of exophoria at 40 cm and 3 m, respectively. These values compare favorably with population normative data.25 Acuity and heterophoria measurements were used to test for changes associated with each display. For each subject, a normal￾ized vision measure for each of the five displays was established by averaging the mean measures across the five display conditions. For each subject/display condition, a relative vision measure was estab￾lished by determining the difference between the normalized mea￾sure and the individual measure. The mean and standard devia￾FIGURE 1. Normalized performance speeds, mean and standard deviations for 22 subjects for three tasks (paragraph reading [PR], letter counting [LC] and word search [WS]) on 5 displays (monocular virtual [MV], binocular virtual [BV], flat panel [FP], hard copy [HC] and small format [QVGA] display). TABLE 2. Mean symptom magnitudes (scale 0–100) after series of performance tasks on each display. Display Eye Symptoms Motion Symptoms Musculoskeletal Headache Eyestrain Sore eyes Blurry Dizziness Nausea Disorient Neck Ache Back Ache MVa 3.4 14.9 10.8 9.5 2.9 1.2 5.0 3.4 2.3 BV 6.2 10.5 7.7 8.5 3.0 2.4 1.8 2.4 2.5 FP 2.1 7.6 5.8 4.1 1.6 2.4 1.3 3.6 3.0 HC 3.4 8.5 5.9 6.7 2.3 1.3 3.9 1.6 0.0 QVGA 4.5 8.0 9.2 4.9 3.3 2.1 6.1 2.5 0.2 a MV, monocular virtual; BV, binocular virtual; FP, flat panel; HC, hard copy; QVGA, small format display. TABLE 3. Results from Friedman’s test comparing symptom score across the 5 displays. Symptom p Value Post Hoc Results Headache 0.5976 Eyestrain 0.0131 MVa significantly different (greater) than all other displays. Sore eyes 0.3390 Blurry vision 0.0382 MV significantly different (greater) than FP. Dizzy 0.8039 Nausea 0.6515 Disorientation 0.0829 Neck ache 0.4779 Back ache 0.0404 FP significantly different (greater) than MV, HC, and QVGA. a MV, monocular virtual; BV, binocular virtual; FP, flat panel; HC, hard copy; QVGA, small format display. Near-Eye Computer Displays—Sheedy and Bergstrom 309 Optometry and Vision Science, Vol. 79, No. 5, May 2002

310 Near-Fye Computer Displays-Sheedy and Bergstrom 25 symptoms became intolerable.No subject chose to make such indication. 16 Sympeom measurements were obeained after body movements with and without wearing the hinoaar virual display.The mean 05 symptom mignitudes are shown in Tahle 4. As beforc.the symptom data are noe normally distrihuted and nonparamctric statistical testing was usod to tot for difermnco betwcen symptum magnitudes with and without wearing the hin- -15 ocular virrual display.The Wilcoooa signed ranks test (p<0.05) for each symptomn showed no statistically significant differencs HC berween wearing and not wearing the binocular virtual display. Dispbry FIGURE 2 DISCUSSION orular virtial [MVI,binncular virtal [BVI.at panel [FPL hard comy IHCl Performance and small format [QVGAl displayl. The greatst deviation from normaliaed performance was +63%(paragraph reading on hard copy),and the greatest differ 08 ence between any two displays for a single task was 12.5%6 (para 0 ◆400m graph reading faster an hand eopy than QVGA).a difference that 04 ■3M was statistically significant.Thebetter performance n hard copy is posibly the result of the increaed resolution of a laser-printed 02 ima oompred with one fumed with scren pixels.However.the 0 QVGA display diffiered from all athers testod in this study insofr 02 the format of the display was different,Le.,there was more 04 frequent line wrap.The increased number oflines (albeit the sme 06 amount of tes)may have comtributed to the poorer performance speed on QVGA.However,it should alo be noted that lter 08 counting performance on QVGA was faster then normalized per- 1 FP HC OVCA formance by 5.8%indicating that more frequent line wrap may 5 have assisted the letter counting task.Because of the format differ ence of the QVGA display compared with the other four displays. FIGURE 3. performance differences between QVGA and the other displays Heerophoria meas.remenb lin prbm dfopknl nalive lu nommive cannot necessarily be artribated to image qualiry. Imonocular vimual (Mvl.hinccular vimol lvL flat panel Il1.hard copy Performance on the monocular virtual ncar-ye display was [HC]and snall fonnat QNGA]diplay). guod,including cond-bet performanoe for paragraph reading tions of the relative viasl acuity mrasrements and heteruphori and word search.Performanxce on monocu山ar virtual was generall与 mcasurements are shown in Figs.2 and 3,respectively. comparable with performan flat panel and hard copy,prob- The dara in Figs.2 and 3 show the differences in vision mea ably reflective of the quality of the virtal image.Performance on surements tobe small compared with the stadad deviations.The the moeorlar9 rtual near-<e displry was unesp区：tny白ster an significance of all differences was tested with the paired t-test.No all tasks than on the hinoculr virtual near-cye displary,although statisrically significant differences in vision meaares were the differenoes were not signifcant.One explanarion for this dif determined ference cuuld be the brger esit pupi in the munocular virtual device,and the fact that it was hand held by the subict so that it Experiment 2 lilcely remained relatively centered on the line of sight.In the benocular virtual device.the relationhip berwren the exit pupil Subjects were ieructed o verhally indicate,at any time during and the line of sight of each eye was dependent upon the Fitting the testing,if there were a naticeable increase in symptoms,oe if characteristies of the frame on the face and also upon the interpu- TABLE 4. Mean symptom magnitudkes (scale 0-100)after hody movements with and without wearing the binocular virtual (BV) display. es号mpt6 Motion Symptoms Musculoskoletal llcadache Eyestrain Sore Eves Blurry Dizziness Nausca Disaricnt Nock Ache Back Ache With BV 6.8 8.0 7.4 6.9 4.9 2.4 3.3 1.1 1.1 Withou BV 4.8 7.D 6.5 6.6 5.2 1.5 2.7 1.1 1.3 Opdometry aud Viaiow Srimnre,Vol.79.No.5.May 2002 Copyright American Academy of Optometry.Unauthorized reproduction of this article is prohibited

tions of the relative visual acuity measurements and heterophoria measurements are shown in Figs. 2 and 3, respectively. The data in Figs. 2 and 3 show the differences in vision mea￾surements to be small compared with the standard deviations. The significance of all differences was tested with the paired t-test. No statistically significant differences in vision measures were determined. Experiment 2 Subjects were instructed to verbally indicate, at any time during the testing, if there were a noticeable increase in symptoms, or if symptoms became intolerable. No subject chose to make such indication. Symptom measurements were obtained after body movements with and without wearing the binocular virtual display. The mean symptom magnitudes are shown in Table 4. As before, the symptom data are not normally distributed and nonparametric statistical testing was used to test for differences between symptom magnitudes with and without wearing the bin￾ocular virtual display. The Wilcoxon signed ranks test (p  0.05) for each symptom showed no statistically significant differences between wearing and not wearing the binocular virtual display. DISCUSSION Performance The greatest deviation from normalized performance was 6.3% (paragraph reading on hard copy), and the greatest differ￾ence between any two displays for a single task was 12.5% (para￾graph reading faster on hard copy than QVGA), a difference that was statistically significant. The better performance on hard copy is possibly the result of the increased resolution of a laser-printed image compared with one formed with screen pixels. However, the QVGA display differed from all others tested in this study insofar as the format of the display was different, i.e., there was more frequent line wrap. The increased number of lines (albeit the same amount of text) may have contributed to the poorer performance speed on QVGA. However, it should also be noted that letter counting performance on QVGA was faster then normalized per￾formance by 5.8%, indicating that more frequent line wrap may have assisted the letter counting task. Because of the format differ￾ence of the QVGA display compared with the other four displays, performance differences between QVGA and the other displays cannot necessarily be attributed to image quality. Performance on the monocular virtual near-eye display was good, including second-best performance for paragraph reading and word search. Performance on monocular virtual was generally comparable with performances on flat panel and hard copy, prob￾ably reflective of the quality of the virtual image. Performance on the monocular virtual near-eye display was unexpectedly faster on all tasks than on the binocular virtual near-eye display, although the differences were not significant. One explanation for this dif￾ference could be the larger exit pupil in the monocular virtual device, and the fact that it was hand held by the subject so that it likely remained relatively centered on the line of sight. In the binocular virtual device, the relationship between the exit pupil and the line of sight of each eye was dependent upon the fitting characteristics of the frame on the face and also upon the interpu￾FIGURE 2. Visual acuity measurements on each display relative to normative (mon￾ocular virtual [MV], binocular virtual [BV], flat panel [FP], hard copy [HC] and small format [QVGA] display). FIGURE 3. Heterophoria measurements (in prism diopters) relative to normative (monocular virtual [MV], binocular virtual [BV], flat panel [FP], hard copy [HC] and small format [QVGA] display). TABLE 4. Mean symptom magnitudes (scale 0–100) after body movements with and without wearing the binocular virtual (BV) display. Eye Symptoms Motion Symptoms Musculoskeletal Headache Eyestrain Sore Eyes Blurry Dizziness Nausea Disorient Neck Ache Back Ache With BV 6.8 8.0 7.4 6.9 4.9 2.4 3.3 1.1 1.1 Without BV 4.8 7.0 6.5 6.6 5.2 1.5 2.7 1.1 1.3 310 Near-Eye Computer Displays—Sheedy and Bergstrom Optometry and Vision Science, Vol. 79, No. 5, May 2002

Near-Fye Compuner Displays-Sheedy and Bergstrom 311 pillary distance of the subject.(The displays in the binocular virtual sure the size of the real-world view while wearing the binocular device had a separation of 63.5 mm.whereas the incerpupillary virtual device.Each of the 22 subjects in this study fixated the distances of subjects ranged from 61 to 66 mm.)Other factors center of the binocular virtual display with their head placed in the favoring the monecular virtl disphy area6%larger angar perimcter arc.The first supcrior and inferior appcaranoe of a 1" and a slightly beter MTF.The binoguar virtsl displays werc white stimulus was measund in the vertical mcridian.The mcan smaller and lighter than the monoculr virual display to bettur supcriur and inferior values were 56"and 76",ropcctivdly,indi- enabk them to be womn on the head. cating periphcral view of the real-worid envirunment beyond these P'erformance spoeds on binocular virtual were generally sower valucs.This peripheral viw of the real environment prowides a than foe ather displays Overall.perfoemance speeds on the binoc- veridical sensory referene that can mitigae the sensory mismatch ular virtual display were about 5 sower than normalized perfor- created by nonveridical mowement of the virtual image and there maner,6.7596 swver compared wich the traditona fat panel and fare lessen the risk of motion-related symptams. hard copy displays Hoever,the only isticay significant dif Anocher differenee is that the current near-cye virtual displays ference was for the wurd xcarch task for which hineeular virtual was had significantly hetter nixel resolution (2 min/pixel)compired slower than flat panel. with the peevostudesThe binocular virtuaeyedis The functional significance of the speed differences in the cur- play in the cunent study ao had proper concordanxce between rent srudy can be compared with performance differences mea- ccommodative and vergence stinli,such was ldaing in some sured in previous sudies that ued similar techniques. previous studies.Proper interpupilry distance (IPD)was ob. Occluding a eye reduces reading and letter counting by 3.7 and tained in the curentscudy by selecting subjeces with an IPD within 2.0%,respectively.Presbyopic contact lens corrections such as the acceptabl range of the device (61-65 mm). monovision,concentric design lenses,and diffractive deign lenses In the current study.subjects pertored text tasks on static displays have been shown to reduce ask perfoce by3 to and also a shortterm task designed to phce subjecrs at risk for dsori Increasing the resolution of a display from 75 dpi to 115 dpi ention.It is posible,however,that if the neareye diplays were used improves reading perfocmance by 17.and using gray scale dfferenty,motion related sympcoms migh:ensue.Video tisks.long improves reading performance by 4 to 20%depending on the pixcl c-trm tasks,or tasks with body tracking and bocly-dirccnod mocion density.The performance differenoes measund berween displays on the disply hne no boen toaod on thes devios Usg:n a dark- in this study are similar in magnitude to pertormance decremants ed room wuuld negac the partial immerion achantage and would associated with wearing the various forms of bifocal contact lens abo incrcase the risk of motion symptoms correction and less than the periormance improvements attained with beter display image For perspective.many people succes fully wear the hifocal contact lenesing that the 59 mag Eye and Vision Symptoms ninade of decreased speed measurd on the binocular viral dis Fyrstrain symptomns weresignificantly higher ne moneeular vir- play will nos persent a lange problem for many people. tual compared with alocher displys blurry vision symptoms were ako signifiantly higher on manocular virtual compared with flat Motion Symptoms and Vision Changes panel.For the monoolr viral diphy.onc cye was ooclded with an arached plastic occluder that extended across the brdge of Mocioo-relted sympcoms were significant findings in many the noce to before the unused eye.It completely occuded fowveal peevios sudes n HMDs,wheres they were oted and most perpheral vision.but some peripheral vision remained, to any significant effect in either experiment I or 2 Likewise. and some illuminrion of the eye-ward side of the dask plastic changes in visul acuiry and binoculr alignment were mot me- occluder nccurred.Te is possihle that monocularity.either hy itself sured in the present study as in previous sudies.1.Several or because of the less than perfect ocdusion,contributed to the conditions were different in the present study compared with pre increased symptoms with the monocular virtual device.The un vious studies in which mocion related symptoms andor vision steadiness of a hand held device may also have contributed to changes were identified. increased sympcoms with monocular virtual Eyesympcom ratings Ooe significant difference is that the current study used were generally higher with the binocular virtual device compared with the other devioes (excepe for monooar virtual),but the dif- ferences were not significant. tasks in which some form of body tracking created mowement on the virtual display.Movement tasks,especially those in which dis CONCLUSIONS play moement is interaive with body movements.have greaer potential tochcit motion-edated symptoms than the crrent sudy In summary,performanee sperd on the monocular near-cyr that had no movement on the sereen and no interaction with hody dispby used in this study was compurable with flat panel and hard movement.Secral of the prios studicrportod the copy diys.Performan speeds with the hinocular near-cyr oomplicating fatrur of a time delay bctwoen body and scroen mowe- dispby wire redeod by about 5%cumpand with normalixcd ment that almost certainly contributed to the mutiun-rdaned perfor▣nce and by6,7S%oomp四rod with the tradicional升am symptoms. panel and hard copy displys The moat likely reaon is becse of Another imporrant difference is that the binocular virrual device less-than-optimal alignment of the exit pupils of the displys with in the cutrent study did not hive full immersion as the devices in the lines of sight.Sympeoms of eyestrain and blurry vision were the previaus sdA perimeter arewd omesignificanely higher on moncular virtl thnher dpy.This Optomery and Viuioe Srimre,Vol.79.No.$.May 2002 Copyright American Academy of Optometry.Unauthorized reproduction of this article is prohibted

pillary distance of the subject. (The displays in the binocular virtual device had a separation of 63.5 mm, whereas the interpupillary distances of subjects ranged from 61 to 66 mm.) Other factors favoring the monocular virtual display are a 6% larger angular size and a slightly better MTF. The binocular virtual displays were smaller and lighter than the monocular virtual display to better enable them to be worn on the head. Performance speeds on binocular virtual were generally slower than for other displays. Overall, performance speeds on the binoc￾ular virtual display were about 5% slower than normalized perfor￾mances, 6.75% slower compared with the traditional flat panel and hard copy displays. However, the only statistically significant dif￾ference was for the word search task for which binocular virtual was slower than flat panel. The functional significance of the speed differences in the cur￾rent study can be compared with performance differences mea￾sured in previous studies that used similar techniques.6, 8, 26–29 Occluding an eye reduces reading and letter counting by 3.7 and 2.0%, respectively.26 Presbyopic contact lens corrections such as monovision, concentric design lenses, and diffractive design lenses have been shown to reduce task performance by 3 to 8%.27–29 Increasing the resolution of a display from 75 dpi to 115 dpi improves reading performance by 17.4%,6 and using gray scale improves reading performance by 4 to 20% depending on the pixel density.8 The performance differences measured between displays in this study are similar in magnitude to performance decrements associated with wearing the various forms of bifocal contact lens correction and less than the performance improvements attained with better display images. For perspective, many people success￾fully wear the bifocal contact lenses, suggesting that the 5% mag￾nitude of decreased speed measured on the binocular virtual dis￾play will not present a large problem for many people. Motion Symptoms and Vision Changes Motion-related symptoms were significant findings in many previous studies on HMDs,10–16 whereas they were not measured to any significant effect in either experiment 1 or 2. Likewise, changes in visual acuity and binocular alignment were not mea￾sured in the present study as in previous studies.11, 13, 16–19 Several conditions were different in the present study compared with pre￾vious studies in which motion-related symptoms and/or vision changes were identified. One significant difference is that the current study used common work-type tasks. Most previous studies on HMDs10, 11, 13–17, 19–21 used video movement or virtual reality tasks in which some form of body tracking created movement on the virtual display. Movement tasks, especially those in which dis￾play movement is interactive with body movements, have greater potential to elicit motion-related symptoms than the current study that had no movement on the screen and no interaction with body movement. Several of the previous studies10, 11, 14 reported the complicating factor of a time delay between body and screen move￾ment that almost certainly contributed to the motion-related symptoms. Another important difference is that the binocular virtual device in the current study did not have full immersion as the devices in the previous studies.10–16, 18–21 A perimeter arc was used to mea￾sure the size of the real-world view while wearing the binocular virtual device. Each of the 22 subjects in this study fixated the center of the binocular virtual display with their head placed in the perimeter arc. The first superior and inferior appearance of a 1° white stimulus was measured in the vertical meridian. The mean superior and inferior values were 56° and 76°, respectively, indi￾cating peripheral view of the real-world environment beyond these values. This peripheral view of the real environment provides a veridical sensory reference that can mitigate the sensory mismatch created by nonveridical movement of the virtual image and there￾fore lessen the risk of motion-related symptoms. Another difference is that the current near-eye virtual displays had significantly better pixel resolution (2 min/pixel) compared with the previous studies.10–21 The binocular virtual near-eye dis￾play in the current study also had proper concordance between accommodative and vergence stimuli, such was lacking in some previous studies. Proper interpupillary distance (IPD) was ob￾tained in the current study by selecting subjects with an IPD within the acceptable range of the device (61–65 mm). In the current study, subjects performed text tasks on static displays and also a short-term task designed to place subjects at risk for disori￾entation. It is possible, however, that if the near-eye displays were used differently, motion-related symptoms might ensue. Video tasks, long￾er-term tasks, or tasks with body tracking and body-directed motion on the display have not been tested on these devices. Usage in a dark￾ened room would negate the partial immersion advantage and would also increase the risk of motion symptoms. Eye and Vision Symptoms Eyestrain symptoms were significantly higher on monocular vir￾tual compared with all other displays; blurry vision symptoms were also significantly higher on monocular virtual compared with flat panel. For the monocular virtual display, one eye was occluded with an attached plastic occluder that extended across the bridge of the nose to before the unused eye. It completely occluded foveal and most peripheral vision, but some peripheral vision remained, and some illumination of the eye-ward side of the dark plastic occluder occurred. It is possible that monocularity, either by itself or because of the less-than-perfect occlusion, contributed to the increased symptoms with the monocular virtual device. The un￾steadiness of a hand-held device may also have contributed to increased symptoms with monocular virtual. Eye symptom ratings were generally higher with the binocular virtual device compared with the other devices (except for monocular virtual), but the dif￾ferences were not significant. CONCLUSIONS In summary, performance speed on the monocular near-eye display used in this study was comparable with flat panel and hard copy displays. Performance speeds with the binocular near-eye display were reduced by about 5% compared with normalized performance and by 6.75% compared with the traditional flat panel and hard copy displays. The most likely reason is because of less-than-optimal alignment of the exit pupils of the displays with the lines of sight. Symptoms of eyestrain and blurry vision were significantly higher on monocular virtual than other displays. This Near-Eye Computer Displays—Sheedy and Bergstrom 311 Optometry and Vision Science, Vol. 79, No. 5, May 2002

312 Neur-Eye Cumpuler Displays-Sheedy arl Bergstrom finding may be related to the effects of oocluding an eye or hand ssympioms when an HMD is used a penscedl viewing system. holding the device.No significant changes in visual acuiry o het Dwpl%1997l8:17-16 erophoriaoccuned with any of the displays.Motion-related symp 15.Morse SE.Jiang BC.Oculomoooe function after wirual reality uie toms with the head mountod near-cye display were not diesympomi fom symptomic indidas.Optom significantly different than with other displays toted.T'asks that V5-i19976637-42 invohe movement.body tracking.or utilization in a dark environ- 14.Howacth PA.Finch M.The nausmopreiciry of tum mehods of savi- ment would increase the risk of symptoms and should be tested. gaing wichin a virud cavirunment.Appl Eno 199910:39-5. 15.Hill KJ.Howarh PA.Habirgation ro the side effects ofimmertion in The better performance and comtort on near-eye displays in this a vimeal envimnment.Displays 2000:21:25-30 stuady compared with mast peevions srudiest with HMDs is 16.Huwarth PA.Cootello Pl.Viual effects of immepion in virtual likely due to lck of mavement in the task,parrial instead of full evronmens:Inserim rois fiom the UK Heah and Safery Exocutive immersion,herter digply realution,and concordnce of the ac- commodntive and vergence stimuli.The findings of this study. D%4 ot uf Toclnical4ul8XVl.h,1明6，SaXu,CA.Sa along with tho of Rushtonand Peli,indicate that opti- lu:.CA:Socicty fer letunatiun Dopbey.19968858. mizing the display and/or task conditions can enable safe and ef fective use of near-eye displays. mounted displays and their rebtion to in-flight symptoms.Hum .1un1995:37699-710. ACKNOWLEDGMENTS displys In:Rares J.Rarletr CT.Delahagita PA.Encarnxao JL Iabiryan NV,Inaris P,Wocl A民ek.Prucrrling o SPl上 应a pparied与hwC小aiae.atbor ES aened a lagieg Soo and Diplay Tuheulgies vel.2949.Fcbruary, ukw的你eng成5同giv nalrmu却8uaaw 996.San Diego.CA.Belinghom WA:SPTE 19162-71. ircbeicef auitome avd guislaeer We aby poome lan Paily fr auitemr 19.Hwath PA.Cuulumolur dn witlin virtual eviroum avsl powwe af rive touir wwd'in thie mudr and prve ircl r steeinnaf A4l上un1999:309-67. swanimn and annou Reeriud fuw 7.:ririox reriun Ctoher1.3M) 20.Rushton S.Mon Williams M.Wann JP.Binoculr vision in a bi- ocularworld:newgnrion head-moureed diplays aid casing viual delisit.Diplay 1994:15-255-260. 21.Pel E The vis cffet of bed-muumed dilay (HMD)a nut REFERENCES distingishablefom those ofdesk-top compurer diply.Vison Res 1.Paterson DG.Tinker,MA.Srudies of ypogriphical facroes influenc 1998:38:205-66 ing speed of reading J Aopl Pyychol 1929:13:120-30. 22.Bailey IL.Lunir JE.New doip princido fur vioudl y letter 2.Lnckirsh M.Mna FK.The Scirntr af Socirg.New York:Van Now- chars.AmJOptom Physiol Opt 197653:740-5. cTnd,1937. 23.Daiel WW.Biosrarirics:A Foandarioe foe Analysis in the Health 3.Muter P,Latremnuille SA.Tncuerict C,Beam P.Exeeded reading nf Suirnun.Sth o New Yurk:Wiry.1991-292-4. cnntimaou srxt on tdleviion scrmns Hus Fa:tnx 198224-501-8. 24.Comover ]Practical Nonparametric Seatistics.3nd ed.Neu York Goul ID.Grischkowsky N.Doing the same work with hard copy WT年，1. dwith cathode-ray tbe (CRT)compuier terminals Hm Facoes 25.Sherdly JE,Saladin J).Validiy uf danontic crieria and cue aalyis 198426323-37 ibinncubr viaice diwerden.In:Schor CM,Ciuflrod KI,ctle.Ver 5.Goul JD.Alfaco L.Baraes V.Pinn R.Grischlowsy N.Minuto A. gence Eye Movemens Baskc and Cical Aspects.Boston:Burter. Reading is slower from CRT displrys than fiom paper sctempts to worth,183:517-4n isdane a single-varihe explancrioe Hum Factoex 198719:269-99 26.Shrrdy Hailry IL.Buri M,Boa E Binocalar v.moeccular t 6.Sheedy.Reading perfomince and wiul cnmfeet an a high molu performance.Am J Opoom Physiol Ope 1986x6t:8s9-i6. con moairor compured to a VGA monaoe.Electroni Imag 1992:1: 2/.Sheedy JE.Harris MG,Busby L Chan E.Koga L Monovision con 405-1a eatt lsa wear and outupatiunal tok perloemanr.Am J Optom 7.Naiman AC.MakousW.Undrtrered gray strips digplacr percrivrd Physol Op阿1s865:14-8. odges nonlinearly.Opt Soc Am A 1993:10:791-803. 28.Sheedy JE.Harris MG,Bronge MR.Joe SM.Mook MA.Task and 8.Sheedy IE.McCarthy M.Reading perfoumance and visual comfort vial perlunance with cuncrntricbilucal cuntact linse.(pcoe Vis wirh arale m gray compared wich black and white scanned peint. 5i1991:68:537-41. T%nlm1994l5:27-30. 29.Hamris MG.Sheedy JE.Gan CM.Vision and rask performance with 9.laschirski-Krues W.Vhual stain during VDU woek:the effec of monticon and dilfractive biocd contact lenu.Opnm Vis Sci virwing digance and dark focus.Frgnnemnics 1988:31:1449-65. 199269:609-14 If.Mon-Willree M.Wann JP.Ruchron 5.Birecular vision in a vietral wur达viual defwin th:wearing of a head-mounedd山s James E.Sheedy shy.Oh山山c Pivsiol0w1明3：1ki87-9l. /Ohio Stare Uuisersi的 11.Regan FC.Price KR.The frepeecy of occamenoe and srwrity nf Cx时fm side-cffocis of immersion virtual realiry.Avat Space Envieon Med 320 Wot Teth Ave 19965:527-30. Colsmaour,OH 43210) 12.Howerth PA,Corello PI.The ecarrence of rirmal simultinn sick- rma业e的am.rd Optomerry and Vhiae Srimare,Vad?以Na天M事2Ue2 Copyright American Academy of Optometry.Unauthorized reproduction of this article is prohibited

finding may be related to the effects of occluding an eye or hand￾holding the device. No significant changes in visual acuity or het￾erophoria occurred with any of the displays. Motion-related symp￾toms with the head mounted near-eye display were not significantly different than with other displays tested. Tasks that involve movement, body tracking, or utilization in a dark environ￾ment would increase the risk of symptoms and should be tested. The better performance and comfort on near-eye displays in this study compared with most previous studies10–19 with HMDs is likely due to lack of movement in the task, partial instead of full immersion, better display resolution, and concordance of the ac￾commodative and vergence stimuli. The findings of this study, along with those of Rushton et al.20 and Peli,21 indicate that opti￾mizing the display and/or task conditions can enable safe and ef￾fective use of near-eye displays. ACKNOWLEDGMENTS This research was supported by InViso Corporation. Author JES served as a consultant to InViso during the project. We give special thanks to Ron Roncone for technical assistance and guidance. We also thank Ian Bailey for assistance with some of the tasks used in this study and Lynn Mitchell for statistical consultation and analysis. Received June 7, 2001; revision received October 1, 2001. REFERENCES 1. Paterson DG, Tinker, MA. Studies of typographical factors influenc￾ing speed of reading. J Appl Psychol 1929;13:120–30. 2. Luckiesh M, Moss FK. The Science of Seeing. New York: Van Nos￾trand, 1937. 3. Muter P, Latremouille SA, Treurniet WC, Beam P. Extended reading of continuous text on television screens. Hum Factors 1982;24:501–8. 4. Gould JD, Grischkowsky N. Doing the same work with hard copy and with cathode-ray tube (CRT) computer terminals. Hum Factors, 1984;26:323–37. 5. Gould JD, Alfaro L, Barnes V, Finn R, Grischkowsky N, Minuto A. Reading is slower from CRT displays than from paper: attempts to isolate a single-variable explanation. Hum Factors 1987;29:269–99. 6. Sheedy, JE. Reading performance and visual comfort on a high resolu￾tion monitor compared to a VGA monitor. J Electronic Imag 1992;1: 405–10. 7. Naiman AC, Makous W. Undetected gray strips displace perceived edges nonlinearly. J Opt Soc Am A 1993;10:794–803. 8. Sheedy JE, McCarthy M. Reading performance and visual comfort with scale to gray compared with black and white scanned print. Displays 1994;15:27–30. 9. Jaschinski-Kruza W. Visual strain during VDU work: the effect of viewing distance and dark focus. Ergonomics 1988;31:1449–65. 10. Mon-Williams M, Wann JP, Rushton S. Binocular vision in a virtual world: visual deficits following the wearing of a head-mounted dis￾play. Ophthalmic Physiol Opt 1993;13:387–91. 11. Regan EC, Price KR. The frequency of occurrence and severity of side-effects of immersion virtual reality. Aviat Space Environ Med 1994;65:527–30. 12. Howarth PA, Costello PJ. The occurrence of virtual simulation sick￾ness symptoms when an HMD is used as a personal viewing system. Displays 1997;18:107–16. 13. Morse SE, Jiang BC. Oculomotor function after virtual reality use differentiates symptomatic from asymptomatic individuals. Optom Vis Sci 1999;76:637–42. 14. Howarth PA, Finch M. The nauseogenicity of two methods of navi￾gating within a virtual environment. Appl Ergon 1999;30:39–45. 15. Hill KJ, Howarth PA. Habituation to the side effects of immersion in a virtual environment. Displays 2000;21:25–30. 16. Howarth PA, Costello PJ. Visual effects of immersion in virtual environments: Interim results from the UK Health and Safety Executive Study. In: Society for Information Display International Symposium Digest of Technical Papers, vol. XXVII. May, 1996, San Diego, CA. San Jose, CA: Society for Information Display, 1996:885–8. 17. Kotulak JC, Morse SE. Oculomotor responses with aviator helmet￾mounted displays and their relation to in-flight symptoms. Hum Factors 1995;37:699–710. 18. Kooi FL. Visual strain: a comparison of monitors and head-mounted displays. In: Bares J, Bartlett CT, Delabastita PA, Encarnacao JL, Tabiryan NV, Trahanias P, Weeks AR, eds. Proceedings of SPIE: Imaging Sciences and Display Technologies, vol. 2949. February, 1996, San Diego, CA. Bellingham, WA: SPIE 1996:162–71. 19. Howarth PA. Oculomotor changes within virtual environments. Appl Ergon 1999;30:59–67. 20. Rushton S, MonWilliams M, Wann JP. Binocular vision in a bi￾ocular world: new generation head-mounted displays avoid causing visual deficit. Displays 1994;15:255–260. 21. Peli E. The visual effects of head-mounted display (HMD) are not distinguishable from those of desk-top computer display. Vision Res 1998;38:2053–66. 22. Bailey IL, Lovie JE. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976;53:740–5. 23. Daniel WW. Biostatistics: A Foundation for Analysis in the Health Sciences. 5th ed. New York: Wiley, 1991:292–4. 24. Conover WJ. Practical Nonparametric Statistics. 3rd ed. New York: Wiley, 1999. 25. Sheedy JE, Saladin JJ. Validity of diagnostic criteria and case analysis in binocular vision disorders. In: Schor CM, Ciuffreda KJ, eds. Ver￾gence Eye Movements: Basic and Clinical Aspects. Boston: Butter￾worth, 1983:517–40. 26. Sheedy JE, Bailey IL, Buri M, Bass E. Binocular vs. monocular task performance. Am J Optom Physiol Opt 1986;63:839–46. 27. Sheedy JE, Harris MG, Busby L, Chan E, Koga I. Monovision con￾tact lens wear and occupational task performance. Am J Optom Physiol Opt 1988;65:14–8. 28. Sheedy JE, Harris MG, Bronge MR, Joe SM, Mook MA. Task and visual performance with concentric bifocal contact lenses. Optom Vis Sci 1991;68:537–41. 29. Harris MG, Sheedy JE, Gan CM. Vision and task performance with monovision and diffractive bifocal contact lenses. Optom Vis Sci 1992;69:609–14. James E. Sheedy The Ohio State University College of Optometry 320 West Tenth Ave Columbus, OH 43210 e-mail: sheedy.2@osu.edu 312 Near-Eye Computer Displays—Sheedy and Bergstrom Optometry and Vision Science, Vol. 79, No. 5, May 2002