Steady Cam Techniques in Cinematography Kian Marandi A thesis submitted in partial fulfilment of the requirements for the degree of BACHELOR OF APPLIED SCIENCE Supervisor: Dr. Paul Milgram Department of Mechanical and Industrial Engineering i Abstract Camera stabilisation is a crucial aspect of cinematography. Users of all types must deal with unwanted camera vibrations. Various stabilisation methods are available. While there is no commonly accepted performance metric for measuring stability performance, a paired comparison can be employed between various stabilisations methods to gain a sense of the relative performance between them. As such, four camera-stabilisation system setups are examined in this paper; no stabilisation assistance, on-board (OIS) assistance, simple mechanical systems, and a novel propietary software developed by Herman Lo referred to as System X. Results indicate that simple mechanical systems continue to be the best choice for vibration reduction performance, while the on-board assistance provides little noticeable improvement to vibration reduction. Finally, System X shows a lot of potential for cinematographic applications as it ranked highly among the 4 options. ii Acknowledgements Professor Paul Milgram Herman Lo Professor A.N. Sinclair The Department of Mechanical and Industrial Engineering at the University of Toronto iii Table of Contents ABSTRACT I ACKNOWLEDGEMENTS II LIST OF FIGURES V LIST OF TABLES VI INTRODUCTION 2 OBJECTIVE 3 BACKGROUND INFORMATION 4 THE MOVIE CAMERA STABILISATION SYSTEMS – THE STEADICAM® STABILISATION SYSTEMS – SIMPLE MECHANICAL SYSTEMS STABILISATION SYSTEMS – ON-BOARD SYSTEMS STABILISATION SYSTEMS – “SYSTEM X” EVALUATING STABILISATION PERFORMANCE 4 7 8 9 11 11 METHOD 13 TEST DESIGN STEP 1: FOOTAGE CAPTURE STEP 2: PERFORMANCE EVALUATION 13 14 15 MATERIALS 17 MATERIALS – STEP 1: FOOTAGE CAPTURE GENERAL: SETUP 1 (NO STABILISATION ASSISTANCE): SETUP 2 (ON-BOARD STABILISATION): SETUP 3 (SMS): SETUP 4 (SYSTEM X): MATERIALS – STEP 2: PERFORMANCE EVALUATION 17 17 18 18 18 18 19 RESULTS 20 SESSION 1 RESULTS – WITHOUT SYSTEM X FOOTAGE SESSION 2 RESULTS – WITH SYSTEM X FOOTAGE 20 22 DISCUSSION 24 CONCLUSION 25 APPENDIX A – DATA FOR EVALUATION SESSION 1 27 iv APPENDIX B – KEY FOR EVALUATION SESSION 1 28 APPENDIX C – RELATIVE PSYCHOLOGICAL DISTANCES (LARGE) (SETUP 4 OMITTED) 29 APPENDIX D – DATA FOR EVALUATION SESSION 2 30 APPENDIX E – KEY FOR EVALUATION SESSION 2 31 APPENDIX F – RELATIVE PSYCHOLOGICAL DISTANCES (LARGE) 32 v List of Figures Figure 1 – Comparison of Common Video Resolutions Figure 2 – Interlaced Video Figure 3 – The Main Parts of a Steadicam Figure 4 – Examples of Simple Mechanial Stabilisation Systems Figure 5 – Schematic of Panasonic’s MegaOIS System Figure 6 – Example of Digital Image Stabilisation Figure 7 – Example of Evaluation Session Screen Figure 8 – Panasonic AG-HVX200 Camera (left) and Firestore FS-100 Recorder (right) Figure 9 – Relative Psychological Distances (Setup 4 Omitted) Figure 10 – Relative Psychological Distances vi List of Tables Table 1 – Camera Operator and Skill Level Table 2 – Stabilisation Methods for Each Setup Table 3 – Proportion of Choice of Case Footage Based on Stability (Setup 4 Omitted) Table 4 – Z-Scores (Psychological Distance) Between Case Footage (Setup 4 Omitted) Table 5 – Case Rankings (Setup 4 Omitted) Table 6 – Proportion of Choice of Case Footage Based on Stability Table 7 – Z-Scores (Psychological Distance) Between Case Footage Table 8 – Case Rankings 2 Introduction Being able to achieve a steady shot without unwanted vibrations while filming is important for cinematographic applications. Professional cinematographers and amateur home-movie makers alike must cope with the same set of physical challenges when using their movie cameras. Professional filmmakers use heavier cameras and thus will often mount their cameras onto expensive and elaborate mechanical systems to steady their filming [1]. Consumer level cameras, on the other hand, will make use of on-board systems that are based either purely on software algorithms, or simple mechanical systems that work in tandem with on-board software processing [2]. While on-board software solutions can be cheaper and provide for a lighter and less cumbersome camera setup, there is a level of control that the camera operator is surrendering to the judgments of the software’s algorithms. For the professional cinematographer, ultimate control of the captured image is paramount. Thus, the software and mechanical based solutions that exist in consumer video camera models today may be suitable for consumer applications where manufacturers are aiming to create smaller and lighter cameras, but it is still problematic for the professional cinematographer. A trade-off also occurs with usability. Whereas software systems may not perform as well as their mechanical counterparts, their use is fairly simple. Using the advanced mechanical systems often requires training. Herman Lo and Dr. Paul Milgram at the University of Toronto have developed a novel proprietary image processing software that approaches the problem of stabilising video in a different way than the aforementioned existing software systems. As such, this 3 new software has potential for cinematographic applications and is worth investigating. Due to legal reasons, the software system will henceforth be referred to as “System X”. Objective The overall purpose of this thesis project is to use a systematic approach to evaluate existing techniques for stabilising video cameras as well as “System X” which was developed in the ETC Laboratory at the University of Toronto and to determine its feasibility for cinematographic applications. 4 Background Information Since the aim of this thesis is to examine the effective stabilisation performance of various systems, it follows that a basic discussion regarding these systems should be present. Furthermore, these systems all work to assist a camera while it captures footage. Thus, an introduction to the various camera types and technologies is also presented. The Movie Camera Many camera types are available today fulfilling the requirements of various user groups. Professional filmmakers working on large-scale productions will often use a 35mm film camera that captures an image onto a film that moves at 24 frames per second [3]. Popular manufacturers of these types of cameras include Arriflex and Panavision. The needs of the professional cinematographer require the camera system to be capable of accommodating various lenses, film cartridges, and grip systems. The compromise with having professional levels of image capture performance and modularity is that these camera systems are expensive, heavy, and require experience to operate. The advent of the digital video era technologically provided the means to improve the performance of more affordable cameras and thus gave rise to the “SemiProfessional” camera market segment. Semi-professional, or “prosumer”, users include event videographers and amateur filmmakers. Instead of capturing the image onto a physical film, the digital video camera works by converting the images captured by the image sensor into a digital format that can be stored on various types of storage media. Storage media include digital tape, hard-drives, and solid-state memory. The advantages 5 of the cameras that fall into this market segment include increased user friendliness and decreased cost and physical weight. These improvements are gained at the expense of image quality when compared to professional film cameras, and the weight and size of these cameras are still greater than those of consumer video cameras. Many camera models that fall into this market segment are capable of capturing High-Definition video, which currently exists in three common formats; 720p, 1080i, and 1080p [4]. A full discussion of these video standards is beyond the requirements of this project, however it is useful to know that the numerical value (720, 1080) represents the number of horizontal lines of resolution of the captured image (see Figure 1 below). The appended letter, “p” or “i”, designate “progressive” and “interlaced” respectively. A simple explanation of the two terms would be that with a progressive video source, the entire frame of the viewed image is recorded and subsequently presented to the viewer at the same frame rate that the camera is set to record at. Thus, if a camera is set to record at 24 fps (frames per second) using a progressive scan pattern, upon playback the viewer will be presented with 24 separate and complete frames every second. An interlaced video source, on the other hand, presents each frame as two separate fields that were originally captured one after the other wherein each field is only one half of the entire frame (see Figure 2 below) [5]. Thus, if a viewer was presented with an interlaced video source at 60 fields per second, the viewer is only actually presented with 30 complete frames every second [5]. However, since each field is captured chronologically in succession, the viewer will perceive the smoother motion that is characteristic of a video source that is truly 60 frames (and not fields) per second due to the persistence of vision. 6 Figure 1 – Comparison of Common Video Resolutions (Source: http://upload.wikimedia.org/wikipedia/commons/thumb/a/a2/Common_Video_Resolutions_2.svg/700pxCommon_Video_Resolutions_2.svg.png) Figure 2 – Interlaced Image (Source: http://www.anchorbaytech.com/_media/images/support/interlaced-scan.jpg) In addition to the ability of capturing high-definition video, some semi-professional cameras are capable of capturing video at user selectable frame rates. While 60 fps provides the common look and “feel” of motion of what is referred to as “home video”, 30 fps provides the look of most television programs, and 24 fps provides the look that is common to professional, “Hollywood” films [6]. 7 Consumer digital video cameras, available in most electronics stores, are more compact and lightweight than their semi-professional counterparts. Since consumers typically do not require the same performance from a camera as professionals and semiprofessionals, the cameras will subsequently also be much less expensive. The method of recording is the same as the digital semi-professional cameras mentioned above. However, consumer cameras will use fewer image sensors, or sensors of lower quality (i.e. smaller size, resolution and colour depth). Stabilisation Systems – The Steadicam® The Steadicam® was invented by Garrett Brown in the 1970’s and continues to be an effective method of image stabilisation today [7]. The Steadicam® is an apparatus that is worn by the camera operator and consists of three main components; the camera sled, arm, and vest [1] (see Figure 3 below). The camera sled acts as a mounting point for the camera and any additional equipment that is necessary such as batteries or monitors. The sled is connected to the vest via the arm. Figure 3 – The Main Parts of a Steadicam ® (Source: http://entertainment.howstuffworks.com/steadicam1.htm) 8 Together, the system provides smooth footage even during shots where the camera operator is travelling [1]. Operating the Steadicam® however requires training [1], and is clearly a system that is elaborate and cumbersome. Some of the early movies that showcased the effectiveness of the Steadicam® system include The Return of the Jedi and The Shining. In the The Return of the Jedi, the system was used to film a walkthrough shot of a forest to be used as the background footage used to create the fast-moving speeder-bike sequence [8]. In Stanley Kubrick’s, The Shining, the Steadicam® was used to film the famous scene where the camera is following the young boy on his tricycle as he quickly pedals his way through the halls of the Overlook Hotel [9]. Unfortunately, Steadicam® systems are costly and consequently are not readily accessible to prosumers and consumers. Stabilisation Systems – Simple Mechanical Systems The term “steady cam”, spelled with a “y” is a generic term used in the industry that can refer to any mechanical device that a camera is mounted on to provide stabilisation support. However, to avoid confusion with the trade name “Steadicam®”, other methods of image stabilisation via a mechanical system will be henceforth referred to as SMS (simple mechanical systems). Various types of SMS exist, however they all function using a similar principle; the system has a centre of gravity (COG) that is in a different location than that of the bare camera, such that either the operator is holding the apparatus at the COG or the COG exists directly in-between both of the operator’s hands (see Figure 4 below). 9 Figure 4 – Examples of Simple Mechanical Stabilisation Systems (Sources: http://www.newdaypictures.com/images/Steadicam_Merlin_Hire.gif and http://www.dv.com/dv/magazine/2007/July/goodman0707figrig_r.jpg ) These types of systems are popular with prosumers, since they are cost-effective and easy to use. As mentioned above, for cinematographic applications ultimate control over what the camera is recording is important. These systems provide stabilisation support while maintaining the level of control available to the operator. Stabilisation Systems – On-Board Systems Many cameras available today, prosumer and consumer alike, will often have an on-board feature that can assist in the stabilisation of the recorded footage. These systems are available in two varieties; optical image stabilisation, and digital image stabilisation. Optical image stabilisation (OIS) is a term used in reference to any system wherein the camera mechanically adjusts the angle of the lens with respect to the image sensor such that unwanted vibrations are reduced [2]. The OIS system employed by Panasonic (see Figure 5 below) is controlled via software algorithms that respond to the inputs provided by on-board sensors that detect camera vibrations [2]. 10 ® Figure 5 – Schematic of Panasonic’s MegaOIS System (source: http://www2.panasonic.com/webapp/wcs/stores/servlet/MegaOISExplained?storeId=15001) Digital image stabilisation (DIS), on the other hand, executes all the adjustments digitally. The image sensor captures an oversized frame (the blue frame in Figure 6 below), and what the operator observes through the camera, and subsequently what the camera records, is a smaller frame within the larger one (the red frame in Figure 6 below). The area of the captured frame that is not visible to the operator acts as a buffer against vibrations. Thus, as the camera vibrates, the smaller frame will move within the larger frame in order to keep the image centred and stable (as indicated by the arrows in Figure 6 below). Figure 6 – Example of Digital Image Stabilisation 11 Of course, both of these systems are subject to the decision making of the built-in software algorithms. However, while the user is sacrificing image control to the built-in system, the system is very easy to operate considering that it usually consists of merely pressing a button to activate it. Stabilisation Systems – “System X” System X, as mentioned above, is a novel proprietary image processing system developed by Herman Lo and Dr. Paul Milgram in the ETC Laboratory at the University of Toronto. Unfortunately due to legal restrictions, the working principles of System X must be omitted from this report. However, it is hypothesized that in a fully implemented form, System X will provide a new approach for real-time on-board video stabilisation. Evaluating Stabilisation Performance While no formal and widely accepted performance metric has so far been discovered within the literature review for evaluating stabilisation systems, a study by Borys Golik entitled Development of a Test Method for Image Stabilizing Systems is worth mentioning [10]. Golik’s study focuses on the stabilisation systems used by leading manufacturers of digital still cameras. During the study, Golik proposes using a test rig that a camera can be mounted on that would shake at a user-defined frequency and amplitude to simulate the vibrations typically caused by physiological tremor. The application of Golik’s method to this project is unfortunately limited. The study focuses on digital still cameras, which poses a different set of physiological challenges to the camera operator. Where using a movie camera would involve controlled panning, tilting, translation, and even walking, a digital still camera requires the operator to not move at 12 all. Thus, not only does the product from using a movie camera involve greater degrees of freedom, the biomechanical requirements of the operator are also different. Therefore, a test rig that vibrates only to simulate the physiological tremor experienced while holding an object still cannot adequately simulate the wide range of physical challenges encountered by movie camera operators. Furthermore, Golik’s test is designed to treat a single still image, whereas video requires the analysis of many still images being viewed in succession. 13 Method Test Design The testing procedure used for evaluating the stabilisation techniques included having three camera operators film a target recording using four different camerastabilisation system combinations. The three camera operators were selected such that their experience with operating a camera was varied: Table 1 – Camera Operator and Skill Level Camera Operator Experience Level A Experienced B Amateur C Novice The four camera-stabilisation system combinations that were evaluated are as follows: Table 2 – Stabilisation Methods for each Setup Setup Stabilisation Method 1 No stabilisation assistance 2 On-board stabilisation assistance – (OIS) 3 SMS 4 System X Thus, in total there were 12 cases (4 setups for each of the 3 camera operators). After the footage was prepared, the 12 cases were judged by numerous test subjects using a pair 14 comparison test such that they can be ranked relative to one another on a single continuum. Step 1: Footage Capture For each setup, the operators used the camera to capture a short 30-second predefined “target” recording. The target recording involved having the operators walk down a well-lit hallway while at a set pace. The operator’s walking pace was controlled using a digital metronome that provided an auditory signal via its built-in speaker. During the setup of the recording process, the camera operators decided that a signal of 100 bpm from the metronome provided a comfortable target walking pace. Setup 1 (no stabilisation assistance) had the operators capturing the target footage using only the bare camera with no on-board or mechanical assistance. Setup 2 (on-board stabilisation assistance) was done with the camera’s on-board OIS system activated. Setup 3 (SMS) had the camera mounted on a SMS with the on-board OIS turned off. Finally, Setup 4 (System X) involved the use of System X to process the video captured during Setup 1. For all setups, the camera’s auto-focus feature was turned off to eliminate the differences between recordings that may arise from the camera’s auto-focus algorithms. During preliminary pilot testing, it was determined that the amount of zoom used during recording has a strong effect upon the visible amount of shake. An increase in telephoto will increase the amount of visible camera vibrations. Thus, the zoom was kept at a moderate level of 25% telephoto during all of the actual target recordings. Furthermore, the different ways to grip the camera also proved to impact the results of the 15 footage during pilot testing. The overhand grip provided footage that was more stable. However, since the overhand grip handle is a feature unique to the HVX200 (see Figure 8 below) and similar larger cameras, the camera operators were instructed to use a standard grip in order to more comprehensively represent the ergonomics of the majority of camera models available. Finally, as a control case for the performance evaluation phase, the camera was mounted on a tripod, which was then placed on a dolly so that a “perfectly smooth” version of the target recording was captured by pushing the dolly down the hallway. Step 2: Performance Evaluation After the target recordings were captured, they were prepared for evaluation using Final Cut Pro, a non-liner editing software suite. The method of analysis used is a pair comparison as described in Trygg Engen’s, Psychophysics: II. Scaling Methods [12]. To do so, a group of test subjects were required to observe 66 pairs of recordings such that each case was paired with every other case once. During which, they were required to indicate which recording of each pair (the left side or right side) was more stable according to their own perceptions. However, before the test subjects began the 66 trials, they were presented with the control case mentioned above. This was done so that the subjects understand what an ideal target recording looks like. Initial plans involved having the entire evaluation process completed online, such that the test subjects can view the 66 trials on YouTube, and indicate their selection on an online form. However, due to time constraints, the evaluation procedure was carried out by having a group of test subjects observe a video screen in a room while indicating their 16 selections on a form. Figure 7 below is an example of what the video screen was displaying. Figure 7 – Example of Evaluation Session Screen Thus, for each trial, the pair of recordings would play side-by-side for 10 seconds. After which, the video screen would prompt the test subjects to input their response as well as provide them a 5 second countdown for when the next trial will begin. In total, the time required to carry out the experiment with the test subjects including time required for instructions was approximately 25 minutes. 17 Materials The materials required for this experiment can be divided according to the separate steps and setups of the aforementioned test method. Materials – Step 1: Footage Capture General: - 1 (one) Panasonic AG-HVX200 High-Definition Camera - 1 (one) Focus Enhancements Firestore FS-100 DTE DVCPRO-HD Tapeless Recorder - 1 (one) digital metronome - 1 (one) tripod dolly The Panasonic AG-HVX200 is a high-definition 3CCD video camera. It is capable of filming at 480i (standard definition), 720p, 1080i, and 1080p. The camera is also capable of filming at a wide range of frame rates, including 24fps, 30fps, and 60fps [11]. The HVX200 is a popular and widely used camera in the prosumer segment, and the ability to natively record a progressive image at 24fps makes it an ideal candidate to simulate cinematographic applications (see Figure 8 below). The Firestore FS-100 is an external recording device that connects to the HVX200 and records the captured footage onto a hard-disk (see Figure 8 below). 18 Figure 8 – Panasonic AG-HVX200 Camera (left) and Firestore FS-100 Recorder (right) (Sources: http://cache.gizmodo.com/assets/resources/2007/01/hvx_200_right500-1.jpg and http://catalog2.panasonic.com/webapp/wcs/stores/images/models/fs-100.jpg) Setup 1 (no stabilisation assistance): - No extra materials required Setup 2 (on-board stabilisation): - No extra materials required Setup 3 (SMS): - Constructed mechanical stabilisation system The SMS was constructed using the procedure available at www.steadycam.org. The system design is by Johnny Chung Lee, and is a cost-effective solution that provides equivalent stabilisation performance to more expensive, yet functionally identical, systems that are manufactured professionally. Setup 4 (System X): - System X: Configured to process the footage captured by HVX200 using Setup 1 19 The configuration of System X and the processing of the captured video was completed by Herman Lo. System parameters were kept constant when processing the video for each camera operator. Materials – Step 2: Performance Evaluation - 1 (one) Video projector - 1 (one) Projector screen - 1 (one) Final Cut Pro editing suite - Test subjects - Response forms for test subjects 20 Results The evaluation phase was carried out twice. Once without the System X processed footage and once with. The evaluation session carried out without the System X footage was done before the processed footage was made available in an attempt to gain data in the event that the System X footage was ultimately unavailable. As such, the number of test subjects used was greater than in the session with the System X footage. Thus, the results from the first evaluation phase can be used to support the results obtained from the session with the System X footage included. Session 1 Results – Without System X Footage It should be noted that since Setup 4 was omitted for this evaluation session, there are only 9 total cases that need to be evaluated. Thus, the total number of pairs that the test subjects were required to respond to was 36. There were 53 test subjects in total The raw data collected from the response forms were inputted into a chart (see Appendix A) in order to find the proportion of test subjects that preferred a certain recording over its paired counterpart. Since the test subjects are not aware of which cases they are observing, a key (see Appendix B) was used to decode the results such that they can be put in matrix form. The following matrix provides the proportions for which the test subjects preferred the case in the first column to the case in the first row. Note that the letter (A, B, or C) represents the camera operator and the number (1, 2, or 3) represents the camera-stabilisation system setup: 21 Table 3 – Proportion of Choice of Case Footage Based on Stability (Setup 4 Omitted) Then as per the analysis method outlined in Trygg Engen’s, Psychophysics: II. Scaling Methods [12], the proportions above were converted to equivalent Z-Scores. Then the mean of each row was found and then linearly transformed so that all the number are positive: Table 4 – Z-Scores (Psychological Distance) Between Case Footage (Setup 4 Omitted) Thus, the performance of each case can be placed on a single continuum to gain a sense of their performance relative to one another as follows (for a larger version, please see Appendix C): Figure 9 – Relative Psychological Distances (Setup 4 Omitted) 22 Table 5 – Case Rankings (Setup 4 Omitted) Session 2 Results – With System X Footage The same analysis procedure was carried out for the evaluation session that included the System X processed footage. There were 31 test subjects in total. The raw data is available in Appendix D and the associated key in Appendix E. The proportions for the responses are as follows: Table 6 – Proportion of Choice of Case Footage Based on Stability 23 The converted Z-scores are as follows: Table 7 – Z-Scores (Psychological Distance) Between Case Footage And thus, the relative performance of each case is as follows (for a larger version, please see Appendix F): Figure 10 – Relative Psychological Distances Table 8 – Case Rankings 24 Discussion It is important to note that the scores for each case do not represent a measure of absolute performance, but rather a relative psychological distance between the perceived performance of each case [12]. For both evaluation sessions, the highest and lowest ranking cases remained the same. Namely, the experienced camera operator using the SMS setup ranked the best, whereas the novice camera operator using just the bare camera ranked the worst. Furthermore, setups 1 and 2 rank fairly close to one another for all camera operators, which may suggest that the on-board OIS system does not mitigate the camera vibrations enough to create a very noticeable difference for this application. Across all camera operators, the SMS setup was received much better than the others. However, it is of interest to note that the System X software performed quite well. Notably, case A4 outperformed case C3. This is indicative that the System X software certainly has potential for application for on-board camera stabilisation applications. From a qualitative perspective, the camera vibrations in the video processed by System X do seem to be much “slower”. Which may suggest that the temporal nature of the footage is being modified. Thus, for a true cinematographic application, future builds of the System X software must ensure that objects moving independently of the frame (such as people in motion) are not affected by this temporal ‘warping’. Clearly, for frame compositions involving no independently moving objects, the algorithms in place work quite effectively. Future research may include investigating how System X performs with a dynamic frame composition. 25 Generally, the results agree with what is expected. Setup 1 with no stabilisation assistance consistently performs the worst, with Setup 2 (OIS) performing only marginally better. Then there is a sizeable gap leading up to Setup 3 and 4. Furthermore, the ranking of setups 1, 2 and 3 are in general agreement between the two evaluation sessions, thus fortifying the results. Conclusion Using a paired comparison to evaluate the relative performance between various camera stabilisation methods proved to yield interesting results. Interestingly, the on-board stabilisation system of the Panasonic HVX200 did not provide significant vibration reduction that can be noticed by the test subjects. Furthermore, the results fortify that simple mechanical systems continue to provide much better vibration reduction performance than the alternatives, which suggests that among the systems tested, simple mechanical systems may still be the best choice for amateur film-makers and serious videographers. Finally, the strong results of System X indicate that the potential for its cinematographic applications is significant. Future research regarding System X’s performance with dynamic frame compositions will provide additional information as to how future builds of the software should be approached in order to deal with any potential temporal warping of the footage. Thus, future builds of System X can very well be considered for application in both consumer and prosumer video cameras. 26 References 1. Harris, Tom. November 22, 2001. How Steadicams Work. Internet. <http://entertainment.howstuffworks.com/steadicam.htm> . October 2, 2008. 2. Unknown. Unknown. Panasonic MegaOIS Explained. Internet. <http://www2.panasonic.com/webapp/wcs/stores/servlet/ MegaOISExplained?storeId=15001> . October 2, 2008. 3. Encyclopedia Britannica. 2008. Motion-picture camera. Internet. <http://www.britannica.com/EBchecked/topic/394179/motion-picture-camera>. November 9, 2008. 4. Wikipedia, The Free Encyclopedia. November 9, 2008. High-definition video. Internet. <http://en.wikipedia.org/w/index.php?title=High-definition_video&oldid=250609389>. November 10, 2008 5. Encyclopedia Britannica. 2008. Television. Internet. <http://www.britannica.com/EBchecked/topic/1262241/television-technology>. November 11, 2008. 6. Wikipedia, The Free Encyclopedia. October 24, 2008. Frame rate. Internet. <http://en.wikipedia.org/w/index.php?title=Frame_rate&oldid=247467059>. November 12, 2008. 7. Knowledge@Wharton. February 27, 2006. Garrett Brown: Inventing the Future – And a Few Handy Gadgets. Internet. <http://knowledge.wharton.upenn.edu/article.cfm?articleid=1352&specialid=48>. November 13, 2008 8. Brown, Garrett. 1983. Steadicam Plates for Star Wars: Return of the Jedi. American Cinematographer. June 1983 9. Brown, Garrett. 1980. The Steadicam and “The Shining”. American Cinematographer. August 1980. 10. Borys Golik. October 2006. Development of a Test Method for Image Stabilizing Systems. Cologne University of Applies Sciences: Department of Imaging Sciences and Media Technology. 11. Unknown. Unknown. Panasonic – DVCPRO P2 – <http://www.panasonic.ca/english/broadcast/P2/AGHVX200.asp>. November 9, 2008. AG-HVX200A. Internet. 12. Trygg Engen, "Psychophysics: II. Scaling Methods”. Chapter 3 in J.W. Kling & L.A. Riggs (ed's), Woodworth & Schlosberg's Experimental Psychology, Third Edition, Holt, Rinehart & Winston, 1971, pp. 47-86. 27 Appendix A – Data for Evaluation Session 1 Each “1” indicates an instance when the test subject indicated the footage on the left side as more stable. 28 Appendix B – Key for Evaluation Session 1 29 Appendix C – Relative Psychological Distances (Large) (Setup 4 Omitted) 30 Appendix D – Data for Evaluation Session 2 Each “1” indicates an instance when the test subject indicated the footage on the left side as more stable. 31 Appendix E – Key for Evaluation Session 2 32 Appendix F – Relative Psychological Distances (Large)
© Copyright 2026 Paperzz