Curiosity Cloning Image Viewer User`s Manual
Contents
1. These two edit boxes allow the user to specify the dimensions respectively width and height in pixels of the images contained in the dataset Actual size of the images may be different from specified but please note that bigger images will be cropped to the frame size left top part of an image is cut and smaller images will be placed in the left top part of the frame 3 4 2 Image buffer size In this edit box the user may specify the number of images to be pre loaded to the system memory before starting the actual image display Approxi mation of the amount of the memory required in megabytes is displayed on a caption next to the edit box Please note that both frame dimensions and the number of images affect the amount of memory required 2008 Advanced Concepts Team European Space Agency All rights reserved 11 Chapter 4 Usage Guidelines 4 1 Video Mode Selection There are various factors affecting the optimal selection of the display mode The first one is the size of the images displayed during the experiment Ideally width and height of the screen in selected display mode should be identical to the dimensions of images being displayed However it is very often not possible In such case the best video mode would be the one with the smallest dimensions which guarantee that entire images will be displayed In such case the images will be rendered in the centre of the screen For example if the dataset consi2. Curiosity Cloning Image Viewer User s Manual Marek Ruci ski Advanced Concepts Team European Space Agency marek rucinski esa int ACT Technical Report ACT MAN 5100 CCIVUMO1 November 18 2008 Contents 1 Introduction VOL Features pr tairi grene E A ti A a 1 2 Software Requirements e 1 3 Legal Information 0 20200 0004 2 Principles of Operation 2A Terminology iaa Bab he eS kee eae 2 2 Program Operation pri sox e Re 3 Configuration Dialog 3 1 Group Display as eae oe a ee te Es 3 1 1 Display adapter 2000 3 1 2 Display mode uri ua e a a e 3 1 3 Fullscreen 3 2 Group Datasets ee O26 image folder g a fee ego ke Song OE a eae Ge Rt 3 2 2 Use random image sequence s Use an experi ment definition file 00 3 2 3 Number of images per run 3 2 4 Random seeds 2 0000020 eee 3 2 5 Experiment definition file not labelled 3 3 Group Exposition 2 20 0 2 000 a a a ee 3 3 1 Attempt to achieve arbitrary durations Strictly synchronize with monitor s vertical refresh 3 3 2 IDP duration 2 0000085 Joo TER duration eo tte vy ate Bs eee oe 3 3 4 Synchronize with monitor s vertical refresh 3 3 5 Allow shorter IDP durations 3 3 6 Allow shorter IIP du
3. 20 Outcome of the synchronised timing method 21 2008 Advanced Concepts Team European Space Agency All rights reserved Ill List of Tables 4 1 Exposition times for popular refresh rates 21 2008 Advanced Concepts Team European Space Agency All rights reserved IV Chapter 1 Introduction Curiosity Cloning Image Viewer further referred to as CCViewer is an application which has been designed to display images with very high timing accuracy The application has been initally developed by the Advanced Concepts Team European Space Agency in order to provide a tool to be used by the universities participating in the Curiosity Cloning Neural Modelling for Image Analysis project The software is intended to be used to display series of images to a human subject during which it s EEG signals and possibly other biometric measurements are taken 1 1 Features The software allows the user to e Choose the display adapter and it s display mode e Set up series of experiments generated pseudo randomly or using an user provided script file e Specify image exposition duration using multiple timing methods The software is able to load images in the following formats e Windows bitmap file format BMP e Joint Photographics Experts Group compressed file format JPEG e Truevision image file format Targa or TGA e Portable Network Graphics file format PNG recommended e DirectDraw surf
4. every image is displayed not more than once e One image may appear in multiple runs e Sequences of images are independent from each other as the input permutation is always the same determined by the lexicographical order of the file names Note that due to the nature of the pseudo random number generators generated image sequences may be considered random but are completely determined by the random seeds used Thus in order to repeat exaclty the same experiment one just has to use identical random seeds Please also note that the program has no knowledge about the paradigms of experiments involving the display of visual stimuli it just displays the 2008 Advanced Concepts Team European Space Agency All rights reserved 13 0 mon aa Pt WN SB rer N FO sequences of images It does not discern between background oddball and distraction images This makes the random sequence generator rather of small use in for instance oddball paradigm experiments The second method of defining the experiment is to provide an experi ment definition file This file contains the descriptions of sequences of images to be displayed The format of the file is the following The file should be a plain ASCII text file The first line of the file should contain the number of file sequences defined in the file i e the number of experiment runs Following should be the appropriate number of file sequence definitions separated by one blan
5. a symmetrical probability distribution In the CCViewer GUI frame display delay compensation is enabled using two options in the Exposition controls group namely Allow shorter IDP and Allow shorter IIP checkboxes These options allow selection of the due time calculation method for IDP and IIP independently thus allowing for instance forcing the display of the images to last for the desired value while allowing shortening the duration of the blank screen between two images in order to compensate the drift of the average image presentation rate Practical difference between the two presented timing options is illus trated on the histograms on figure 4 5 Data has been gathered for a sample 2008 Advanced Concepts Team European Space Agency All rights reserved 17 Frequency Frequency i Mi a 1032 1033 1034 1035 1036 1037 1038 1039 104 104 1 1042 More 99 5 997 999 1001 1003 1005 1007 1009 More Duration ms Duration ms Figure 4 5 Timing histograms for arbitrary timing without left and with lag compensation right consisting of 100 images displayed with desired exposition interval of 100 milliseconds The left hand histogram shows the distribution of the ac tual measured frame display time when the due time of the next frame is calculated relatively to the previous frame s actual display time the basic version the right hand histogram when it is calculated relat
6. on figure 4 2 As shown on the figure actual frame display time succeeds the sched uled frame time by a random unpredictable amount of time indicated by redrectangles on the time axis Timing strategies mentioned in the begin ning of this section differ in the way the next frame due time is calculated 2008 Advanced Concepts Team European Space Agency All rights reserved 15 Scheduled time Scheduled time Scheduled time _ Time a wu ul Actual time Actual time Actual time Figure 4 2 The basic timing mechanism and in the method the frame display is triggerred Details of these strategies are described in the following sections 4 3 2 Arbitrary Timing The first method assumes that the user specifies an arbitrary duration of a frame display time The program tries to meet user s expectations as closely as possible In the basic version of this method the due time of the next frame is calculated relatively to the actual display time of the previous frame If d denotes the desired frame duration program operation may be visualised as on figure 4 3 Because of the unpredictable delay between scheduled and actual frame display time resulting time intervals on the figure denoted as d and d between actual frame display times are always different from and not shorter than the desired interval d The advantage of this approach is that for every frame it is guaranteed that it will be displayed for at least d time
7. Agency All rights reserved 6 Chapter 3 Configuration Dialog Curiosity Cloning Image Viewer Intel R 82945G Express Chipset Family Figure 3 1 CCViewer configuration dialog Configuration dialog of the CCViewer is shown on figure 3 1 All dialog controls and their meanings are described below 2008 Advanced Concepts Team European Space Agency All rights reserved 7 3 1 Group Display This group contains controls related to the display adapter 3 1 1 Display adapter This control allows the user to select the display adapter which will be used to display the images This option is useful in computer systems in which a separate graphics adapter is used to handle the high definition display or when different interfaces of the graphics adapter are handled as if they were separate adapters this approach has been commonly used by ATI 3 1 2 Display mode This control allows the user to select the video mode to be used to display images For guidelines on video mode selection please refer to the section 4 1 3 1 3 Fullscreen This checkbox toggles between full screen and windowed program operation Windowed mode is made available mainly for software developing purposes During normal program operation full screen mode should be used as it makes use of entire screen no other program windows are visible and may very likely provide better performance 3 2 Group Datas
8. ace file format DDS For details about the Curiosity Cloning Neural Modelling for Image Analysis project visit the Advanced Concepts Team website http www esa int act 92008 Advanced Concepts Team European Space Agency All rights reserved 1 e Portable pixmap file format PPM e Windows device independent bitmap file format DIB e High dynamic range file format HDR e Portable float map file format PFM 1 2 Software Requirements In order to use CCViewer following hardware and software is required e PC with Microsoft Windows XP operating system Windows Vista should work but was not tested e Microsoft DirectX 9 0c runtime libraries update August 2007 version 9 24 1400 or newer e Microsoft C Runtime Library CRT version 7 1 or newer e Microsoft DirectX 9 0c compatible video adapter In order to make use of the full software potential following hardware is recommended e At least 2 GB of RAM the more the better e High performance video adapter and display capable of handling HD display resolutions with high refresh rates 100Hz without interlacing using a digital interface like DVI D or HDMI e Hard disk storage with very high transfer rates and very short access time 1 3 Legal Information The names of actual companies and products mentioned in this Manual may be the trademarks of their respective owners Initial release of this Manual has been published in 2008 by the A
9. always improve the situation because the delay caused by the processing time is random Depending on the relation between the desired frame exposition duration d and the refresh rate of the display device resulting actual frame exposition time may be sig nificantly different for different frames The example resulting distributions of the frame exposition times are illustrated on histograms on figure 4 7 Frame display was in this case synchronised with the vertical refresh rate of an LCD monitor 60 Hz meaning roughly 16 67 milliseconds between two vertical refresh periods Again on left hand histogram the compensation is disabled on the right hand enabled It is clear that apart from increasing the image presentation quality tearing has been eliminated one has gained significant improvement of the frame exposition time variance This is the consequence of eliminating at least partially the delay resulting from the random frame rendering time The drawback is that the actual frame exposition time may very well be significantly different from user s demands It may also happen that if the relation between desired exposition time and device s refresh rate 2008 Advanced Concepts Team European Space Agency All rights reserved 19 Frequency Frequency 1166 11661 11662 116 63 116 64 116 65 More 99 93 99 94 99 95 99 96 99 97 90 98 More Duration ms Duration ms Figure 4 7 Timing histograms for
10. dvanced Concepts Team European Space Agency in the form of a technical report number ACT MAN 5100 CCIVUMO1 If you find CCViewer software use ful in experiments you conduct please give proper credit to the CCViewer authors by quoting this technical report in related papers references For your convinience we provide the appropriate BibTeX entry on listing 1 1 2008 Advanced Concepts Team European Space Agency All rights reserved 2 ROO Ne VD N Q A Listing 1 1 This Manual s BibTeX entry Qtechreport CCIVUMO1 Author Rucinski M Title Curiosity Cloning Image Viewer User s Manual Institution European Space Agency the Advanced Concepts Team Year 2008 Number CCIVUM01 Note Available on line at http www esa int act Url http www esa int gsp ACT doc INF pub ACI MAN 5100 CCIVUMO1 pdf CCViewer software as well as this Manual are distributed under the following conditions Copyright 2008 Advanced Concepts Team European Space Agency All rights reserved Redistribution and use in source and binary forms with or with out modification are permitted provided that the following con ditions are met e Redistributions of source code must retain the above copy right notice this list of conditions and the following dis claimer e Redistributions in binary form must reproduce the above copyright notice this list of conditions and the following dis claimer in the docu
11. ets This group contains controls related to the image datasets 3 2 1 Image folder This control allows the user to point to the directory which contains the images to be displayed Folder picking dialog is displayed using the button 3 2 2 Use random image sequence s Use an experi ment definition file These two radio buttons allow the user to choose between the two options for defining an experiment scenario i e the image sequence s to be displayed See section 4 2 for more information on defining experiment scenarios 2008 Advanced Concepts Team European Space Agency All rights reserved 8 3 2 3 Number of images per run In this edit box the user can specify how many images should be contained in one experiment run This number must be less than or equal to the number of images available in the dataset folder The value 0 means that all images in the image folder should be used 3 2 4 Random seeds This edit box allows the user to specify random seeds which will be used to generate image sequences for every experiment run In consequence the number of entered random seeds determines the number of experiment runs If there is more then one experiment run consecutive random seeds should be separated by a space Random seeds may be entered manually or generated pseudo randomly using the Draw button 3 2 5 Experiment definition file not labelled When using an e
12. explained 2 1 Terminology Dataset a set of image files used in an experiment Experiment run a display of a sequence of a subset of images from a dataset Experiment a sequence of one or more consecutive experiment runs Image Display Period IDP a period during which a single image is being displayed on the display device Inter Image Period IIP a period between two consecutive IDPs dur ing which no image is displayed on the display device 2 2 Program Operation CCViewer is designed to conduct series of displays of sequences of images from an image dataset When the program is run a configuration dialog is displayed giving the user the opportunity to set up the experiment and different image presentation parameters Program configuration is stored in a file named ccviewer ini located in the same directory as the program executable This allows preserving program configuration from one program run to another The configuration file is written every time the user presses the OK button in the configuration dialog After the experiment is set up pressing the OK button in the config uration dialog starts the experiment An experiment consists of at least one 2008 Advanced Concepts Team European Space Agency All rights reserved 5 display of a sequence of images Displayed sequences may be different but they may use only images available in the dataset Every image sequence display consi
13. h rates one can choose among much more options Special care has been taken by the CCViewer developers to ensure as reliable actual frame exposition duration as possible Unfortunately Mi crosoft DirectX 9 programming interface used by CCViewer does not allow the software to reliably count the number of vertical refresh periods that have passed during certain period of time Thus Skipping the desired number of vertical refresh periods has to be obtained by calculating the duration of the interval between vertical refresh periods basing on the known refresh rate and scheduling the next frame shortly after one but last vertical refresh period This allows the software to have enough time to render the frame 2008 Advanced Concepts Team European Space Agency All rights reserved 20 Vblanks Monitor Refresh Rate Hz per frame 50 60 70 75 85 100 1 20 00 16 67 14 29 13 33 11 76 10 00 2 40 00 33 33 28 57 26 67 23 53 20 00 3 60 00 50 00 42 86 40 00 39 29 30 00 4 80 00 66 67 57 14 53 33 47 06 40 00 5 100 00 83 33 71 43 66 67 58 82 50 00 6 120 00 100 00 85 71 80 00 70 59 60 00 T 140 00 116 67 100 00 93 33 82 35 70 00 8 160 00 133 33 114 29 106 67 94 12 80 00 9 180 00 150 00 128 57 120 00 105 88 90 00 10 200 00 166 67 142 86 133 33 117 65 100 00 11 220 00 183 33 157 14 146 67 129 41 110 00 12 240 00 200 00 171 43 160 00 141 18 120 00 Table 4 1 Ex
14. hether it is an image a fixation screen or a blank screen before they are displayed are assigned a time moment at which they should be displayed The program runs in a loop checking the value of a high precision system timer When current system time matches or exceeds due time of the next frame the frame is rendered to the screen After this is done next frame to be rendered is scheduled This algorithm is shown on figure 4 1 a m once EN Yes Figure 4 1 Frame rendering algorithm Using the high performance timer mechanism QueryPerformanceFre quency and QueryPerformanceCounter Windows API functions to measure the time pass is the most accurate timing mechanism available in the Mi crosoft Windows XP operating system Other available timing mechanisms like for instance Queue Timers do not provide satisfying accuracy This timing solution is not perfect though There exists unavoidable and unpredictable difference between frame due time and the time at which the frame is actually displayed Firstly the program queries the timer with a finite resolution usually significantly lower than the timer resolution In consequence the program usually detects that the scheduled frame time has already passed rather than that the time is now Secondly after the condition is detected rendering the frame also takes some unpredictable amount of time Thus the outcome of using the basic timing mechanism may be visualised as
15. ively to the previous frame s due time the second version It is clear that for the former method no frame is displayed for a time shorter than desired 100ms For the latter the distribution is concentrated around the desired value of 100ms most measurements fall between 99 9ms and 100 1ms Another very important practical issue related to the arbitrary timing method is synchronisation of the frame display with so called vertical refresh period of the display device For CRT monitors refresh rate is directly related to the trajectory of the electron beam inside the kinescope For LCD displays one of the most important parameters determining the device s refresh rate is the rate of on off pulses of the monitor s backlight When the software displays an image on the device without taking into account it s refresh rate it is very likely that the screen contents will change while the device is actually drawing previous screen contents The result will be an image consisting of a part of the previous constents in the upper part and the new contents in the lower part This effect is often referred to as tearing or flickering and affects both CRT and LCD displays It is unfortunately both very easily noticeable and quite disturbing for the spectator In order to increase the quality of the exposition the update of the screen contents must be allowed only during so called vertical refresh period i e during the time period when
16. k line Each file sequence definition consists of names of files making up the sequence one per line File names are relative to the image folder directory Before launching the experiment the program verifies if all the files referred to in the experiment definition file are actually present in the image dataset folder Example of an experiment definition file can be faound on listing 4 1 Listing 4 1 Example experiment definition file 3 Filel png File2 png File3 png File4 png File5 png File6 png File5 png File4 png File2 png Experiment definition files have the following advantages over pseudo randomly generated sequences e The user has the full control over the sequences a priori e Experiment runs can be of different length e One image may appear multiple times in one experiment run e Experiment definitions may be generated using external tools for in stance designed for modelling the oddball paradigm 4 3 Timing Options There are two basic timing strategies available for the user of the CCViewer They offer different levels of flexibility accuracy and presentation quality 2008 Advanced Concepts Team European Space Agency All rights reserved 14 When selecting an appropriate timing method a trade off must be done between these three properties 4 3 1 Timing Basics The basic mechanism behind image display timing in CCViewer is the frame scheduler All frames any contents of the screen w
17. mages displayed e Set the image size to be exactly equal to the size of the images being displayed not bigger e If the required amount of memory is too high specify largest possible buffer size and use the fastest disk storage available possibly even a RAID matrix 2008 Advanced Concepts Team European Space Agency All rights reserved 23 e Make sure no other software is running in the background during the experiment Especially disable any anti virus software if present as it may very likely scan all files loaded by the CCViewer against viruses e Make sure to disable any system services which are not necessary as they consume system memory 4 5 Logging In order to trace the conducted experiments CCViewer creates a log file in which much information about the program usage is stored The log file is created in the same directory as the CCViewer s executable file The log file is a plain ASCII text file structured in the following way 1 A header containing complete experiment configuration 2 For every experiment run a The sequence of the file names of the images being displayed b Start time of the image display in human readable form c Image display timing information that is i Frequency of the high performance system timer li Basic statistics for IDP and IIP measurements ii Full table of measurements for every displayed image for both TIP and IDP in both raw timer units and milliseconds in b
18. mentation and or other materials pro vided with the distribution e Neither the name of the Advanced Concepts Team nor the name of the European Space Agency nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS AS IS AND ANY EXPRESS OR IMPLIED WARRANTIES INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PAR TICULAR PURPOSE ARE DISCLAIMED IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT INDIRECT INCIDENTAL SPECIAL EXEMPLARY OR CONSEQUENTIAL DAMAGES INCLUDING BUT NOT LIMITED TO PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES LOSS OF USE DATA OR PROFITS OR BUSINESS INTERRUPTION HOW EVER CAUSED AND ON ANY THEORY OF LIABILITY WHETHER 2008 Advanced Concepts Team European Space Agency All rights reserved 3 IN CONTRACT STRICT LIABILITY OR TORT INCLUDING NEGLI GENCE OR OTHERWISE ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE 2008 Advanced Concepts Team European Space Agency All rights reserved Chapter 2 Principles of Operation In this chapter important terms referring to the concepts important for the understanding of the operation of the CCViewer software are introduced and the principles of the software operation are
19. no contents are actually being drawn on the display device in CRT monitors this is the time when the electron beam travels from the bottom right corner to the upper left corner of the screen This is usually achieved at the hardware level by delaying the actual frame display until the next vertical refresh period The outcome of synchronising the image diplay with the vertical refresh period of the display 2008 Advanced Concepts Team European Space Agency All rights reserved 18 device is illustrated on figure 4 6 Scheduled time Scheduled time Scheduled time EIT ha ES A A d e d WwW Sa A ul Actual time Actual time Actual time Figure 4 6 Outcome of the arbitrary timing method with vertical refresh synchronisation On the illustration vertical refresh periods of the display device have been depicted as vertical yellow bars As clearly seen synchronisation of an arbitrary frame duration with the vertical refresh period of the display de vice may have a dramatic impact on the actual frame exposition time which is now forced to be a multiplicity of the length of the interval between two consecutive vertical refresh periods of the display device In consequence the delay between frame s due time and actual display time is further in creased by the time from the moment when the frame has been rendered to the next vertical refresh period depicted as yellow rectangles on the time axis Enabling lag compensation does not
20. nstance when using a laptop and a projector disable the main display 4 4 Performance Considerations Special care has been taken that the CC Viewer operates as reliably as possi ble Nevertheless the user must also keep in mind that certain things affect the software s performance Most crucial parameter for achieving high performance of the CCViewer is the amount of available system memory In order to display an image it has to be transferred from a mass storage usually a hard disk to the system 2008 Advanced Concepts Team European Space Agency All rights reserved 22 memory and then to the graphics adapter memory The throughput of the mass storage is usually not sufficient for displaying images with very fast rate directly from it In order to achieve high image display rates CCViewer allows pre loading of a certain number of images to the system memory before starting the display The highest reliability is of course obtained when all images to be displayed are pre loaded slow hard disk access is completely eliminated during the image display phase The number of pre loaded images is set in the CCViewer GUI using the Image buffer size edit box in the Performance control group Program memory requirements are directly proportional to the number of pre loaded images and to the size of the images themselves Images have to be kept in the computer memory in an uncompressed bitmap format thus the amount of memory
21. one with biggest screen dimensions supported by the device Support for the other modes is in most cases only emulated i e the graphics hardware automatically resizes rendered image to the native size of the monitor what has a very significant impact on the image quality usually heavy blur Thus on LCD devices the best choice most likely will be to use the native resolution regardless of the size of displayed images 4 2 Setting up the Experiment There are two ways in which the user may define the experiment i e the sequences of the images to be displayed to the subject The first method is to generate the image sequences using a pseudo random number generator The second one is to provide an experiment definition file The pseudo random method uses the following algorithm to generate the image sequences to be used in the experiment 1 For every provided random seed a Take all images in the dataset folder in the order determined by the lexicographic order of their file names b Generate a random permutation of the files basing on the current random seed c Take first N images from the sequence where N is a configuration parameter being the number of images per experiment run or the whole sequence if N is equal to 0 Image sequences generated in this way have the following properties e The number of sequences experiment runs is equal to the number of random seeds provided by the user e During each experiment run
22. oth cumulative and non cumulative form d Basic statistics for image loading times Basic statistics include the sample count mean and standard deviation 2008 Advanced Concepts Team European Space Agency All rights reserved 24
23. plier This track bar allows the user to define the duration of the Image Display Period as a multiplicity of the vertical refresh period of the display device Resulting IDP duration in milliseconds is displayed on an appropriate cap tion below the track bars 3 3 8 IIP multiplier This track bar allows the user to define the duration of the Inter Image Period as a multiplicity of the vertical refresh period of the display device Resulting IIP duration in milliseconds is displayed on an appropriate caption below the track bars The multiplier of 0 completely disables the Inter Image Period 3 3 9 Show fixation cross over images This checkbox toggles display of the fixation cross over the images during the experiment The cross is partially transparent 3 3 10 Fixation rest duration This edit box enables the user to enter the desired duration of both eye fixation screen a blank screen with a fixation cross and countdown shown before the actual experiment run and the eye rest screen a blank screen shown after the experiment run is finished The time is specified in seconds 3 4 Group Performance This group contains controls which allow the user to tune the performance of the application For detailed information regarding the application s per formance please refer to the section 4 4 2008 Advanced Concepts Team European Space Agency All rights reserved 10 3 4 1 Frame dimensions
24. position times for popular refresh rates in milliseconds before the target vertical refresh period comes However as experiments with the software on different hardware platforms showed that the actual refresh rate of a display device may be somewhat different from the nominal one If the actual refresh rate is a bit lower than the nominal one it may happen that the frame will be displayed too early during the one but last vertical refresh period Because no official tolerance specification standards have been found CCViewer authors introduced a margin of around 16 6 of the nominal between refresh period which turned out to be suitable for all tested devices Finally program operation in the synchronised timing mode is illustrated on figure 4 8 Scheduled time Scheduled time Scheduled time 3 1 D e 3 1 D e gt D Time 2 ra D A 4 d 3 a d 3D re di di di Actual time Actual time Actual time Figure 4 8 Outcome of the synchronised timing method On the figure D denotes the duration of the time interval between ver 2008 Advanced Concepts Team European Space Agency All rights reserved 21 tical refresh periods 3 is the desired duration of the frame exposition time expressed in the number of intervals and e denotes the assumed safety mar gin It is clearly visible that if the worst case rendering time of a frame is small enough to fit between two vertical refresh periods the resulting actual timing is ve
25. rations 3 3 7 IDP multiplier o 3 3 8 IIP multiplier 0 3 3 9 Show fixation cross over images 3 3 10 Fixation rest duration apor dis ta S 2008 Advanced Concepts Team European Space Agency All rights reserved 3 4 Group Performance oaa 3 4 1 Frame dimensions 3 4 2 Image buffer size oaoa 4 Usage Guidelines 4 1 Video Mode Selection ooa aa a 4 2 Setting up the Experiment aoaaa a 43 Timing Options eps ira a da 403 1 Limine Basics moros iaa a Bes 4 3 2 Arbitrary Timing 4 3 38 Synchronised Timing 4 34 Final Remarks 4 4 Performance Considerations AB bogging on ke Qn ee Ra De ee De 2008 Advanced Concepts Team European Space Agency All rights reserved II List of Figures 3 1 4 1 4 2 4 3 4 4 4 5 4 6 4 7 4 8 CCViewer configuration dialog 7 Frame rendering algorithm 15 The basic timing mechanism 16 Outcome of the arbitrary timing approach 17 Due time relative arbitrary timing approach 17 Timing histograms for arbitrary timing 18 Outcome of the arbitrary timing method with vertical refresh synchronisation i oac piina A eg E a eee ee 19 Timing histograms for synchronised timing
26. required for an image grows rather fast with the image dimensions For example 2 gigabytes of available memory would allow storing only a bit less than 540 images of dimensions 1000 by 1000 pixels On 32 bit processors Microsoft Windows operating system does not allow any program to use more than 2 gigabytes of system memory Moreover it is very unlikely that such amount of memory will be actually available for the program Even if the program is allowed to run part of it s memory will be swapped i e put to the hard disk neutralising the performance gain achieved with the buffer CCViewer allows the user to specify what is the size of images being dis played using Frame dimensions edit boxes in the Performance control group The given dimensions do not have to match actual dimensions of displayed images If they are smaller they will be loaded in the upper left part of the frame buffer If they are bigger the left upper part of the image will be cropped and displayed For the user s convenience in the Perfor mance control group an estimate of required memory amount is displayed and updated in real time when either image dimensions or buffer size is changed Summarising in order to obtain maximum performance e Use system with as much random access memory as possible but note that 32 bit version of Microsoft Windows XP supports maximum 3 gigabytes e Specify the image buffer size equal to the number of i
27. ry precise equal to desired 3 D 4 3 4 Final Remarks In this subsection several guidelines for timing method selection have been gathered e Rather avoid using the arbitrary timing method The only advantage flexibility comes at the cost of the presentation quality Synchro nising with the refresh rate of the display device removes the only advantage over the synchronised mode while the latter is much more reliable e In order to increase the flexibility use display devices which support multiple refresh rates not by emulation Keep in mind that using devices with very high refresh rates requires more robust display hardware This is because the software is given less time to render the frame between consecutive vertical refresh peri ods If desired exposition interval is obtainable using multiple refresh rates choose display mode with the lower refresh rate it will be more reliable Make sure that the display hardware does not make any tricks with the refresh rate It has been verified by the program authors that some display hardware when using multiple displays applies different refresh rates for both display devices and the software The result was that actual exposition intervals were completely different from what was calculated basing on the refresh rate of selected display modes The conclusion of the experiments is to avoid using multiple displays at once If an auxiliary video output has to be used for i
28. sts of images of size 800 by 800 pixels and the monitor supports display modes with resolutions 800 by 600 1024 by 768 and 1280 by 1024 the latter should be chosen as it is the only mode with height greater than or equal to 800 The user may decide however to use a display mode which is smaller than image dimensions In such case the central part of the image will be displayed The second very important factor affecting the video mode selection is the mode s refresh rate Refresh rate defines how many times per second the screen is completely redrawn This parameter is important because it has a big impact on display timing The higher the refresh rate the more flexibility the user has when defining timing parameters while keeping the high quality of the displayed content Last but not least optimal display mode selection is different for different types of display devices One of the major practical differences between CRT and LCD monitors is that the former display high quality image regardless of the video mode being selected That means that the user can freely select the video mode basing one the criteria mentioned above appropriate dimensions with as much refresh rate as possible In contrary LCD devices obtain optimal display quality only when working in the display mode which is native for the device Usually this is the highest display mode i e the 2008 Advanced Concepts Team European Space Agency All rights reserved 12
29. sts of the following parts 1 Pre loading of images from a mass storage to the system memory in order to improve performance Because it may take considerable time during this phase a progress bar is displayed to the user allowing monitoring the progress and preparation to the experiment run itself Eye fixation screen This is a blank screen with neutral background with a fixation cross displayed in the centre of the screen allowing the user to fully concentrate right before the experiment run In addition to that a countdown counter is displayed allowing the user to precisely anticipate the actual experiment run start what is supposed to reduce the surprise effect Duration of the fixation screen is configurable Actual experiment run The images are displayed to the user in ap propriate sequence Optionally images may be interleaved with Inter image Periods Duration of an image exposition and IIPs may be configured independently Also a semi transparent fixation cross may be kept on the screen during this phase Eye rest screen After the last image is presented a blank screen with neutral background is presented to the user for the same duration as the fixation screen in order to reduce the surprise effect at the experiment run end No fixation cross nor countdown is displayed If there is more than one experiment run the experiment continues from the point 1 2008 Advanced Concepts Team European Space
30. synchronised timing without left and with lag compensation right is inconvenient resulting distribution will be multimodal All this is the consequence of synchronising to the fixed refresh rate it implies that the exposition time may be no longer selected arbitrarily This observation is the fundamental idea behind the second timing method available in CCViewer 4 3 3 Synchronised Timing As indicated in the previous section if the image presentation quality must be kept at the highest level frame display must be synchronised with the refresh rate of the display device This implies that the frame exposition time may be selected only as a certain multiplicity of the time period between two consecutive vertical refresh periods of the display device Thus in the second timing method the frame exposition time is no longer specified directly by the user but calculated as a pre defined number of such intervals Obviously the higher the refresh rate the shorter the interval between vertical refresh periods and the more flexibility the user has when selecting the exposition time Table 4 1 shows example exposition times obtainable for most common refresh rates of the display devices available in the consumer market As clearly seen even for the high end refresh rates the flexibility of the frame duration selection is significantly reduced in comparison to the arbitrary method However if available display device supports multiple refres
31. units The drawback is that the average frame presentation rate is always lower than expected i e lower than the rate calculated basing on arbitrary frame duration given by the user If obtaining accurate presentation rate is more important than achieving a guaranteed exposition time one may try to compensate the error intro duced by the random delay by changing the reference point from which the due time of the next frame is calculated In order to achieve expected frame presentation rate the due time of the next frame must be calculated relatively to the due time of the previous frame as illustrated on figure 4 4 Please keep in mind that the term frame means here any contents of the screen including a blank screen so there is no need to discern between IDP and IIP 2008 Advanced Concepts Team European Space Agency All rights reserved 16 Scheduled time Scheduled time Scheduled time d d E T E G a Actual time Actual time Actual time Figure 4 3 Outcome of the arbitrary timing approach Scheduled time Scheduled time Scheduled time 8 i IE Time di d d Ll Actual time Actual time Actual time Figure 4 4 Due time relative arbitrary timing approach Although actual frame exposition times d and d are still different from d they now oscillate around its value In other words some frames are displayed for a time longer than d and some for shorter hopefully with
32. xperiment definition file to define the image sequences to be displayed the button allows the user to point to the file to be used Format of this file is described in section 4 2 3 3 Group Exposition 3 3 1 Attempt to achieve arbitrary durations Strictly synchronize with monitor s vertical refresh This pair of radio buttons allows the user to choose between available timing options For more information about timing options and differences between them see section 4 3 3 3 2 IDP duration This edit box allows the user to enter the desired duration of the Image Display Period in milliseconds 3 3 3 IIP duration This edit box allows the user to enter the desired duration of the Inter Image Period in milliseconds The value 0 will cause that the Inter Image Period will be completely omitted 3 3 4 Synchronize with monitor s vertical refresh This checkbox toggles the synchronisation with the vertical refresh of the display device 2008 Advanced Concepts Team European Space Agency All rights reserved 9 3 3 5 Allow shorter IDP durations This checkbox toggles the frame display time correction during the Image Display Period Please refer to the section 4 3 for details 3 3 6 Allow shorter IIP durations This checkbox toggles the frame display time correction during the Inter Image Period Please refer to the section 4 3 for details 3 3 7 IDP multi
Download Pdf Manuals
Related Search