Home About Us Laboratory Services Forensic Science Communications Back Issues July 2006 technical Technical Article - Forensic Science...

Technical Article - Forensic Science Communications - July 2006

Technical Article - Forensic Science Communications - July 2006

fsc_logo_top.jpg
fsc_logo_left.jpg

July 2006 - Volume 8 - Number 3

Technical Article

Video and Audio Characteristics in
VHS Overrecordings

Bruce E. Koenig
Audio/Video Forensics Expert
BEK TEK LLC
Clifton, Virginia

Douglas S. Lacey
Audio/Video Forensics Expert
BEK TEK LLC
Clifton, Virginia

Noel Herold
Video Forensics Expert
VAUDEO Forensics
Manassas, Virginia

Introduction | VHS Recording Methodology | General Overrecording Theory | Test Overrecordings | Examinations of Test Recordings | Test-Recording Results | Examples | Forensic Applications | Acknowledgments | References

Introduction

Authenticity examinations of VHS (video home system) cassettes are commonly performed in forensic laboratories and can usually determine whether a submitted recording is original, continuous, and unaltered. One of the important determinations of this analysis is identifying any portions that have been recorded over. An overrrecording occurs by placing a cassette in a VCR (video cassette recorder) or a camcorder (camera-recorder) and recording over previously written information. This paper reviews the appropriate VHS recording principles, explains overrecording theory, details the procedures and results of test overrecordings, provides examples, and discusses forensic authenticity applications.

This paper is concerned with the normal overrecording process, not insert or other editing functions available on some VCRs and camcorders. All of the video recorders tested for this research article employed the National Television System (or Standards) Committee (NTSC) standard, which is used in the United States, Canada, Korea, Japan, and some other countries; however, the methodology can be easily adapted to other related video standards, such as PAL (Phase-Alternation Line Standard) and SECAM (Séquentiel Couleur avec Mémoire, which is French for “Sequential Color with Memory”).

VHS Recording Methodology

When a VHS cassette is inserted into a standard VCR or camcorder, which is then placed in the record mode, the following mechanical, electronic, and magnetic processes occur (Epstein 2003; Goodman 1996; Grob and Herndon 1999; Luther and Inglis 1999; Trundle 1996):

  1. The tape is pulled from its cassette housing by threading guides and then wound over a series of rollers, spindles, guides, stationary erase and record heads, a pinch roller/capstan mechanism, and a helical-scan head drum. Figures 1a (without a VHS cassette) and 1b (with a cassette) show the inside of a typical consumer VCR with the major components labeled.

  2. When the record function is activated, the erase heads start erasing any previously recorded material on the VHS tape.

  3. Simultaneously with the erase function, the helical-scan video heads start recording any new video being provided to the recorder; additionally, units with helical-scan high-fidelity (hi-fi) audio heads start recording any new audio.

  4. Simultaneously with the erase, video record, and hi-fi audio record functions, the linear audio head starts recording any new audio being provided to the recorder.

  5. Simultaneously with the erase, video record, and audio record functions, the control-track record head starts recording 29.97-hertz (Hz) pulses.


Figure 1a:
Inside of a typical consumer VCR with no VHS cassette inserted. This figure is in Adobe Portable Document Format.

Figure 1b: Inside of a typical consumer VCR with a VHS cassette inserted and the front cover removed. This figure is in Adobe Portable Document Format.

VHS Format

All VHS recorders have helical-scan video, and many have hi-fi audio heads, which record information at a shallow angle across the 12.65-millimeter (mm)-wide tape surface. Simultaneously, linear control-track and audio heads record longitudinally on the tape edges. As seen in Figure 2, which is a photograph of a recorded segment of a VHS tape at SP (Standard Play) speed (see below), the narrow linear audio track is at the top, the wider helical-scan video and hi-fi audio tracks are in the middle, and the narrow control track is at the bottom. The tracks in Figure 2 were made visible by coating the tape with a ferrofluid (that is, magnetically developing), which contains small iron particles in the 1- to 3-micron range, which are suspended in Freon CFC-113, or chemically, 1,1,2-Trichloro-1,2,2-trifluoroethane (Koenig 1990). Ferrofluids normally should not be applied to evidential VHS tape surfaces because even with proper cleaning, they can cause playback problems, including tracking errors; degraded audio and video; and clogging of the video, audio, erase, and control-track heads. However, if magnetic development is required, application of a ferrofluid should be the final step in the examination process. Table 1 (International Electrotechnical Commission [IEC] 1993; IEC 1994; IEC 1999) sets forth many of the standards involved in recording and playing back the video and audio information onto VHS cassettes, including the helical-scan angles and track widths.

Figure 2: A recorded segment of a VHS tape at SP speed that has been magnetically developed with a ferrofluid. The angle and direction of the helical-scan lines and the length of a single line of one field are represented. This figure is in Adobe Portable Document Format.

Table 1: NTSC VHS tape and track standards (all dimensions in mm, except as noted)

FEATURESPEP/SLP
Videotape Width 12.65 ± 0.01
Linear Tape Speed (mm/s) 33.35 11.12

Video Tracks:
   Total Vertical Height

10.60

   Effective Vertical Height 10.07
   Width 0.058 0.019
   Helical-Scan Angle (Stationary) 5°56′07.4″
   Helical-Scan Angle (Moving) 5°58′09.9″ 5°56′48.1″
   Azimuth Angles

+6° ± 10′ (field 1)
–6° ± 10′ (field 2)

   Linear Width (Stationary) 96.86
   Linear Width (Moving) 96.30 96.67

Hi-Fi Audio Tracks:
   Width

0.010 → 0.029
   Azimuth Angles

–30° ± 30′ (channel 1)
+30° ± 30′ (channel 2)

Control-Track Width 0.75 ± 0.10

Linear Audio Track Widths:
   Monaural

1.00 ± 0.10
   Each Stereo 0.35 ± 0.05
   Stereo Guard Band 0.30 ± 0.05

Linear Audio/Control-Track
Distance from End of Field 2

79.244 79.253

Sources: Beeching 2001; International Electrotechnical Commission 1993, 1994, and 1999; Ryan 1992

Based on the NTSC nomenclature, each video picture (“frame”) has a total of 525 scan lines of information, with approximately 485 lines containing visible information (usually lines 32.5 through 517.5). Each video frame is composed of two interlaced “fields,” with the first field containing the odd-numbered scan lines and the second field, the even-numbered lines. The playback and recording rates are always 29.97 frames, or 59.94 fields per second. There are only two standardized linear record/playback speeds, SP (33.35 mm/second) and EP (Extended Play, at 11.12 mm/second). EP is often referred to as SLP (Super Long Play), and many VCRs have an interim speed of LP (Long Play, at 22.24 mm/second), which, however, has not been standardized (Grob and Herndon 1999; Luther 1999; Weise and Weynand 2004).

Video Heads

To record the high frequencies required for VHS video, VCRs normally use a spinning 62.0-mm-diameter helical-scan drum with a tape wrap of slightly more than 180 degrees. The drum contains two embedded video heads, 180 degrees apart on the drum, with each head recording one of the two fields in a frame. The drum spins at 1,798.2 revolutions per minute (rpm) in the same direction as the linear movement of the videotape, so that each recorded field lasts 0.01668 second (1 ÷ 59.94). Many camcorders use a smaller helical-scan mechanism with a 41.3-mm-diameter drum, a 270-degree wrap, and a 2,250-rpm spin speed (Trundle 1996). The helical-scan drums are placed at a slight angle to the horizontal alignment of the tape, so that each track starts near the lower edge of the tape and finishes near the upper edge in a right-to-left configuration. This produces a series of parallel diagonal tracks, each containing a recorded field. Each video field comprises 262.5 scan lines of information, recorded along the length of the diagonal track (see Figure 2). Although the standards reflect that the track widths are 0.058 mm for the SP speed and 0.019 mm for EP, in reality, on most VCRs, the actual widths are usually narrower for SP and wider for EP. To avoid interference between adjacent tracks, the two field tracks in each frame are recorded at different azimuth angles, as listed in Table 1 (Beeching 2001; IEC 1994; Ryan 1992).

For illustration purposes, Figure 3 is a drawing of the curved VHS tape path represented in a straight-line, linear mode, for a tape at the SP speed. In this figure, the magnetic side of the tape is shown moving right to left from a viewpoint behind the erase, video, audio, and control-track heads. The vertical dimensions have been enlarged four times compared to the horizontal to provide increased detail. The video heads travel diagonally across the tape at a shallow angle (see Figures 2 and 3), which is directly affected by the linear tape speed. As noted in Table 1, the heads in the helical-scan drum are at a “stationary” angle of 5°56′07.4″ (5 degrees, 56 minutes, and 7.4 seconds of arc); however, when the tape is transported during recording or playback, the effective angles are changed slightly, to 5°58′09.9″ for SP speed and 5°56′48.1″ for EP. These changes in angle affect the length of each field track on the videotape, including its linear width, as noted in Table 1. The linear width of the tracks recorded on the tape change from 96.86 mm for the stationary drum to 96.30 mm for SP speed (see Figure 3) and 96.67 mm for EP.

Figure 3: A linear representation of the magnetic side of a segment of VHS tape, aligned with the video, audio, and erase heads, in a typical VCR at SP speed. The actual tape path is not straight, but curved, and the helical-scan drum is circular, not flat, as shown in Figures 1a and 1b. The vertical dimensions have been enlarged four times compared to the horizontal to allow for increased detail. This figure is in Adobe Portable Document Format.

The helical-scan drums often contain additional heads for better video recording and playback at different VCR speeds, for video special effects, for hi-fi audio, and for “flying erase” capability.

Audio Heads

Most modern VHS recorders and camcorders have two types of audio heads: linear and hi‑fi. The stationary linear audio head is located exactly 79.244 mm from the very end of the second field track of a frame at SP speed (79.253 mm at EP) and continuously records the incoming audio information. It is usually in the same head block as the control-track head, as shown in Figure 3. The quality of the linear audio is directly dependent upon the linear record speed of the recorder; thus information at the SP speed will be of higher fidelity than at EP. The linear audio track is usually monaural on newer VHS units, but some older recorders can have a stereo track configuration (IEC 1993; IEC 1994; Luther 1999).

The hi-fi, or frequency-modulation (FM), audio stereo heads are located on the helical-scan drum, often 60 degrees out of phase with the video heads. The hi-fi audio is recorded during the 0.0334-second time period just prior to, but in the same tape location as, the corresponding video information. To allow differentiation of the audio and video signals and to avoid complete erasure of the audio track by the subsequent overlying video track, the signals have different azimuth angles and track widths and record at different tape depths. The video head partially erases the hi-fi audio, usually dropping its amplitude about 12 decibels (dB). The two channels of stereo audio are recorded with different FM carrier frequencies for better record and playback characteristics. Because of the recording methods, the signal-to-noise ratio, frequency response, and other specifications are always better for hi-fi than linear audio. Hi-fi audio quality is virtually the same at both SP and EP speeds; however, not all VCRs have hi-fi audio capability (Beeching 2001; IEC 1999; Trundle 1999).

Control-Track Head

The stationary control-track record head is located exactly 79.244 mm from the very end of the second field track of a frame at SP speed (79.253 mm at EP) and is usually located in the same head block as the linear audio head, as shown in Figure 3. The control-track head records and reads the 29.97-Hz pulses placed on the track at the bottom edge of the VHS tape, designating the beginning of the first field in the two-field frame. Each synchronization pulse is composed of a rectangular signal consisting of:

  1. A full-amplitude positive impulse lasting less than 200 microseconds (µs).

  2. A direct-current (DC) segment lasting about 0.0222 second.

  3. A full-amplitude negative impulse lasting less than 200 µs.

  4. A DC segment lasting until the start of the next synchronization pulse.

Each positive pulse is physically recorded on the tape at the very end of each frame (see the bottom of Figure 2), where the positive and negative impulses are recorded as a series of short vertical lines. Only the positive pulses are used by the playback VCR, where the magnetically developed control track is aligned with the rectangular pulse signal, as reflected in Figure 4. The control track is used to determine playback speed, to maintain proper playback speed even with tape stretching or shrinkage, to ensure that the playback heads properly read the recorded video tracks, and to update the elapsed time on VCR real-time counters (IEC 1993; IEC 1994; McComb 1995).

Figure 4: A portion of the control track from Figure 2, which has been aligned with the rectangular signal that produced it. This figure is in Adobe Portable Document Format.

Erase Heads

VHS recorders can contain up to four separate erase heads: full-track, flying, linear audio, and control-track. The full-track erase head precedes all of the record, playback, and other erase heads; it erases all previously recorded information on the tape, including the control, linear audio, hi-fi audio, and video tracks. Some VCR units and many camcorders also contain a flying erase head (sometimes two), located on the helical-scan drum. The flying erase head allows the start of a video recording over previously recorded information with little or no distortion or degradation of the ensuing video images and hi-fi audio. A properly functioning flying erase head allows for a recording with an uncorrupted image of the last frame of the underlying recording, followed by an uncorrupted frame of the new recording. The linear audio and control-track erase heads normally are located in the same housing and erase their respective tracks when the record mode is activated. Most units do not have a separate control-track erase head and instead use the erasing effect of the record head and the full-track erase head to delete any underlying information (Epstein 2003; Luther 1999; Luther and Inglis 1999; McComb 1995).

General Overrecording Theory

An overrecording occurs when new information is written over a previous VHS recording, erasing a segment of the existing video, audio, and control-track information, while replacing it with new video, audio, and control-track data.

Video Changes

When a VHS video recorder is placed in the record mode over an existing recording, the full-track erase head starts erasing all of the information on the tape, including the video tracks. However, at the start of the video overrecording, in the “pre-video erased area,” the full-track erase head cannot erase the information physically located between itself and the writing video heads (see Figure 3). If the recorder has a flying erase head, this underlying video information will be erased just ahead of the video-recording process, and the new information will be recorded properly. However, without a flying erase head, the existing video information normally will not be completely erased by the video record head in this portion.

When the overrecording ends and is long enough for the video record heads to have reached the fully erased portion of the VHS tape (at the end of the pre-video recorded area), then a portion of completely erased tape will follow the end of the newly written video. At the end of the completely erased portion, the underlying video information returns as very short partial-field tracks, which progressively increase in length until the entire track is present. This portion of partial tracks is equal to the linear width of the helical-scan tracks because the full-track erase head produces a vertical, 90-degree erasure on the videotape across the slanted (about 6 degrees) field tracks. Therefore, at SP speed, the segment is 96.30 mm in length (96.67 mm at EP), which is based on solving the sine function using the 10.07-mm effective vertical height of the tracks and the angle of 5º58′09.9″ (5º56′48.1″ at EP), which results in a playback time of 2.89 seconds (8.69 seconds at EP). If the overrecording is not long enough for the video heads to reach the fully erased portion, then a segment of the underlying video will be present between the end of the overrecording and the completely erased segment.

Audio Changes

When a VHS video recorder is placed in the record mode over an existing recording, the linear audio record, the linear audio erase, and the full-track erase heads are activated simultaneously. The linear audio erase head deletes the underlying audio recording, except for the short distance between the audio erase and record heads, where the new signal will mix with the old. The full-track erase head starts erasing all of the information on the tape, including the linear audio track(s). At SP speed, the linear audio information is physically located 175.544 mm (175.923 mm at EP) from the beginning of its matching video field track because the linear length of the helical-scan tracks is 96.30 mm (96.67 mm at EP) and the distance from their end to the linear audio track is 79.244 mm (79.253 mm at EP), as shown in Figure 3. When an overrecording ends and its physical length is at least equal to the pre-video erased area, the linear audio does not return until 175.544 mm (175.923 mm at EP) after the end of the erased portion. If the physical distance of the overrecording is less than the pre-video erased area, then a segment of the underlying audio will be present between the end of the overrecording and the erased portion.

When the recorder has a flying erase head, the underlying hi-fi audio information will be erased just ahead of the hi-fi audio-recording process and the new information will be recorded properly. However, even without a flying erase head, the FM audio information normally is completely erased by the hi-fi audio-writing head. When the overrecording ends and its physical length is at least equal to the distance between the hi-fi heads and the full-track erase head, then a portion of completely erased tape will follow the end of the newly written audio. After the end of the completely erased portion, the underlying hi-fi audio information is present on the videotape, but as partially erased tracks. These partial tracks will not play back on most VCRs because their audio-output circuitry automatically switches to the linear audio track whenever the hi-fi signal is not present, is mistracking, or, as in this case, has complete dropouts (Trundle 1999). If the overrecording is not long enough for the audio head to reach the fully erased portion, then a segment of the underlying audio will follow the end of the newly written hi-fi audio.

Control-Track Changes

When a VHS recorder is placed in the record mode over an existing recording, the underlying control-track information is erased by both the full-track erase head and the control-track record head. If the overrecording is not long enough for the control-track record head to have reached the fully erased portion, then a segment of the underlying control-track information will follow the end of the newly written control track.

Test Overrecordings

For this experiment, a series of test recordings was prepared on four VCRs, ranging from consumer-level to professional-quality units, with the following parameters:

  1. The video test signals were interlaced, composite, and at a 1.0-volt peak-to-peak amplitude. The test audio signals were sine waves adjusted to peak at 0 dB on the hi-fi input meter of each unit. On the VCRs that did not have an audio meter, the test signals were supplied at an amplitude of -10 dB volt peak.

  2. For the underlying recordings, continuous audio/video recordings were prepared at both SP and EP (SLP) speeds, with an all-green raster and a 400-Hz audio signal.

  3. Overrecordings consisting of a white crosshatch grid over a black raster video signal and a 1-kHz audio signal were then produced. Overrecordings were made with lengths ranging from 2 to 30 seconds, at both SP and EP (SLP) speeds.

  4. Additional test recordings were made with a variety of other audio and video samples to reflect more real-world scenarios.

Examinations of Test Recordings

The test recordings were examined as follows:

  1. Professional-quality laboratory equipment was used, and all connections were made using either S-video (separated video) or BNC (Bayonet Neil-Concelman) cabling, as appropriate.

  2. The recordings were played back on a VCR, viewed on a high-resolution monitor (usually in the “underscan” mode), and listened to with high-fidelity headphones in both linear audio and hi-fi audio modes. (Note: the VCR automatically switched to the linear mode when either no hi-fi audio was present or the track information was partially erased.) Optimized manual tracking, instead of automatic tracking, was used for the underlying and overrecording segments, as appropriate, for improved playback.

  3. The video was cabled through a time-base corrector (TBC) with digital storage capability, which allowed the capture and review of both individual frames and fields.

  4. The recordings were also played back on the same VCR with the linear audio output selected, cabled through the TBC, and connected to an external computer video-capture/audio-capture device. The capture device was connected to a laboratory computer via a high-speed cable and set to the Moving Picture Experts Group 2 (MPEG-2) format at 15 million bits per second (Mbps), 720- x 480-pixel resolution, and 48-kHz stereo linear pulse-code modulation (LPCM). The recordings were then saved as MPEG (“.mpg”) files.

  5. An additional set of MPEG files was produced, using the same procedures as number 4 above, except that the hi-fi audio output was selected on the VCR.

  6. The MPEG computer files were then reviewed using software that simultaneously displayed the video (with frame numbers added) and the audio information.

  7. The MPEG computer files and the original recording for each test were then reviewed to determine the following

    1. The general visual characteristics of the overrecorded and underlying video information.

    2. The timing of the changes in the video information, including the beginning and end of the overrecording, the fully erased segment, and the partial-track portion.

    3. The timing of the changes in the linear audio information, including the beginning and end of the overrecording, the fully erased segment, and the partial-track portion.

    4. The timing of the changes in the hi-fi audio information, including the beginning and end of the overrecording, the fully erased segment, and the partial-track portion.

    5. The visual and time differences for overrecordings and underlying recordings produced at different record speeds.

    6. The visual and time differences for overrecordings of different lengths.

Test-Recording Results

A review of the test recordings revealed two general classes of results, broken down by the length of the overrecordings. The longer test overrecordings were of sufficient length to completely erase the portion of the videotape between the full-track erase head and the beginning of the helical-scan heads (the pre-video erased area) (see Figure 3); whereas the short overrecordings did not completely erase that segment.

Longer Overrecordings—Video Characteristics

When the longer class of overrecordings was played back, it produced a sequential series of separate video segments, as follows (see Figures 5a, 5b, 6a, and 6b):

U → O or (OU + OP + O) → E → UNP → UN → U

where

U =  Underlying video recording (green angled lines)


O =  Video overrecording (red angled lines)


OU =  Video overrecording with remnants of underlying video recording—only present on overrecording VCRs without a flying erase head—(red angled lines with green shading)


OP =  Video overrecording with a pull-down effect—only present on overrecording VCRs without a flying erase head—(red angled lines with blue shading)


E =  Completely erased portion (gray area)


UNP =  Underlying video recording with a pull-down effect and no control-track sync (blue angled lines)


UN =  Underlying video recording with no control-track sync (light-green angled lines)


Figure 5a: Color-coded representation of a longer overrecording on a VCR at SP speed, with the vertical dimension enlarged 35 times. This VCR has a flying erase head. This figure is in Adobe Portable Document Format.

Figure 5b: Color-coded representation of a longer overrecording on a VCR at SP speed, with the vertical dimension enlarged 35 times. This VCR does not have a flying erase head. This figure is in Adobe Portable Document Format.

Figure 6a: Example test video clip of a longer overrecording made on a VCR at SP speed. This VCR has a flying erase head. This figure is an XviD-encoded MPEG-4 video file with an .avi file suffix.

Figure 6b: Example test video clip of a longer overrecording made on a VCR at SP speed. This VCR does not have a flying erase head. This figure is an XviD-encoded MPEG-4 video file.

This sequence is illustrated in Figures 5a and 5b as recorded on the videotapes and in Figures 6a and 6b as video clips from test overrecordings. Figures 5a and 5b have a vertical dimension that has been enlarged 35 times compared to the horizontal dimension. For example, a 10-mm length in the horizontal direction would be represented as 350 mm in the vertical direction. This nonlinear representation is necessary to allow the complete video sequence to be displayed over a relatively short length. These figures show the narrow linear audio track at the top, the wide helical-scan video and hi-fi audio tracks in the middle, and the narrow control track at the bottom (only the positive sync pulses are displayed). The white spaces between the tracks are the guard bands. The angled, vertical lines in the video track represent individual fields, which are recorded and played back at an angle from the bottom right to the top left (because of scaling, the field tracks are at an angle of about 75 degrees in the figures, instead of the standard of about 6 degrees, as shown in Table 1). The video line widths are not to scale. The color coding in Figures 5a and 5b for the video, linear audio, and control information is consistent throughout both drawings. For example, the red vertical lines in the linear audio track are the physical location of the linear audio for the red angled tracks in the O video. This color coding also reflects a lack of corresponding linear audio or control information, for example, in the blue UNP portion, which has neither control-track sync nor linear audio. Figures 5a and 5b represent the recorded information at SP speed. For EP and other linear speeds, some of the values will be slightly different (see Table 1).

When the overrecording VCR had a flying erase head, the beginning of the overrecorded portion, O, usually produced an uncorrupted frame transition from the underlying recorded portion, U, if both were recorded at the same speed (segments OU and OP are not present on units with a flying erase head). When the overrecording VCR did not contain a flying erase head, remnants of the underlying recording combined with the new video information, producing a pull-down effect as illustrated in the OU and OP portions of Figure 5b and in the Figure 6b video clip. During the OU segment, complete field tracks played back with a mix of video from the underlying recording, U, and the new overrecorded video. These complete OU tracks continued until the physical distance between the full-track erase head and the beginning of the video record heads, which is located at the beginning of OP on Figure 5b, was reached. The OP segment started with the playback of very short track segments of only the overrecorded video information, followed by the mixed information in the OU portion. As playback continued, the percentage of the video track containing only the overrecorded video information progressively increased, while the mixed portion proportionally decreased, until a full field of only the overrecorded video was present at the beginning of the O segment. This progression produced the so-called pull-down effect, which displayed the pure overrecorded video as a horizontal band starting at the top of the frame (beginning of the field tracks) and progressing downward to completely replace the mixed video at the bottom. This pull-down effect lasted a distance of 96.30 mm on the videotape at SP speed (see Figure 3) and 96.67 mm at EP. Therefore, on playback this segment lasted 2.89 seconds (about 87 frames) at SP (96.30 ÷ 33.35) and 8.69 seconds (about 260 frames) at EP (96.67 ÷ 11.12). During the visual review of OP, because only about 485 of the 525 video lines were visible because of vertical blanking, the pull-down effect was visible for no more than 2.67 seconds at SP and 8.00 seconds at EP, even using the underscan feature on professional monitors. When the overrecording was at a different recording speed than the underlying information, the beginning of O (for the units with a flying erase head) or OU (no flying erase head) usually had a short series of distorted frames containing information from both the underlying video and the overrecording.

The completely erased video portion, E, followed the end of the images in the overrecording, O. The length of this erased area was exactly equal to the physical distance between the full-track erase-head gap and the beginning of the video record heads on the drum of the overrecording VCR (see Figures 1a, 1b, and 3). This erased portion was displayed only as noise, usually in a random, herringbone-like pattern, as seen in the Figure 6a and 6b video clips.

The next segment on the test recordings was UNP, which is the return of the underlying video but with a pull-down effect and no control-track sync. As reflected in Figures 5a and 5b, at the beginning of UNP, the playback of very short track segments of video information was followed by the erased information in the E portion. As playback continued, the video track lengths progressively increased, while the erased portion decreased, until a full field of video information was present at the beginning of the UN portion. Usually, at the beginning of the UNP portion, the underlying information was not in color and was very distorted, but it improved in quality as the sequence continued. This progression produced the visual pull-down effect, which displayed the underlying video as a horizontal band starting at the top of the frame (beginning of the field tracks) and progressed downward to fill the frame with the underlying video. This sequence lasted a distance of 96.30 mm at SP speed (96.67 mm at EP) on the videotape (see Figure 3); therefore, on playback this segment lasted 2.89 seconds (about 87 frames) at SP and 8.69 seconds (about 260 frames) at EP. During the visual review of UNP, because only about 485 of the 525 video lines were visible because of vertical blanking, the pull-down effect was visible for no more than 2.67 seconds at SP and 8.00 seconds at EP, even using the underscan feature on professional monitors. There was no control-track sync in this portion because the control-track information for the beginning of each field track in UNP would be located 175.544 mm “earlier” at SP speed (175.923 at EP) on the videotape, which had already been erased by the full-track head, as noted in Figures 5a and 5b. At SP speed, the 175.544 mm was based on the standardized 79.244-mm distance from the end of each field track plus the 96.30-mm linear distance of each track (see Figure 3). At EP speed, the 175.923 mm was based on the standardized 79.253-mm distance from the end of each field track plus the 96.67-mm linear distance of each track (see Figure 3). The most obvious effect of this loss of sync was that the information for individual frames was often combined with data from either previous or following frames, producing images that were not a true reflection of the originally recorded information. An additional effect of the loss of sync was that the underlying video in this segment played back at the same speed as the overrecording; that is, SP overrecordings with EP underlying recordings played back the underlying recording at SP instead of EP speed.

The last segment of the overrecordings, UN, had all of the video information from the underlying recording but had no control-track sync because it had been erased by the full-track erase head. This portion was exactly 79.244 mm in length at SP speed (79.253 mm at EP), based on the standardized distance between the end of the video heads and the linear audio record head (see Figure 3). This segment lasted 2.38 seconds (about 71 frames) at SP playback and 7.13 seconds (about 214 frames) at EP. As the UP portion had, this portion played back at the overrecording speed, regardless of the underlying recording speed, and had sync problems.

Longer Overrecordings—Audio Characteristics

The hi-fi audio information played back only in portions where complete video fields were present on the videotape and not, for instance, in the UP pull-down segments. Therefore, hi-fi audio was present throughout the overrecording (the O, OU, and OP segments) but then ended and did not return until the beginning of UN. However, when the overrecording was at a different speed than the underlying information, the hi-fi audio did not play back until after UN. Unlike the video information recorded by VCRs without a flying erase head, there was no observed mix of the underlying and overrecorded hi-fi audio information in the OU segment. The start and stop times in these segments were very close to those of the video information, except when the overrecording and underlying recordings were recorded at different speeds. In those cases, a delay usually occurred at the beginning because of the speed change.

The linear audio was present during the overrecorded video segments O, OU, and OP, except when the underlying information and the overrecording were recorded at different speeds. In those instances, a delay usually occurred at the onset of the recorded audio. After the end of the overrecording, there was no high-level recorded linear audio information until the end of UN; however, low-level audio (30 to 45 dB below the original amplitude) from the underlying recording was often present in this portion.

Longer Overrecordings—Control-Track Characteristics

The control-track information matched the high-level information on the linear audio track. Therefore, whenever the control-track and linear audio track recordings were not present, the video information lacked control-track synchronization and linear audio, as noted in Figures 5a and 5b.

Short Overrecordings—Video Characteristics

When the short overrecordings were played back, they produced a sequential series of separate video segments that were mostly different from the longer overrecordings, as follows (see Figures 7a, 7b, 8a, and 8b):

U → O or OU → U → UP → E → UP → U → UN → U

where

UP =  Underlying video recording with an erased area pull-down (light-blue angled lines)


This sequence is illustrated in Figures 7a and 7b as recorded on the test videotapes and in Figures 8a and 8b as video clips from test overrecordings. Figures 7 and 8 have the same scaling and other characteristics as Figures 5 and 6, respectively.

Figure 7a: Color-coded representation of a short overrecording on a VCR at SP speed with the vertical dimension enlarged 35 times. This VCR has a flying erase head. This figure is in Adobe Portable Document Format.

Figure 7b: Color-coded representation of a short overrecording on a VCR at SP speed with the vertical dimension enlarged 35 times. This VCR does not have a flying erase head. This figure is in Adobe Portable Document Format.

Figure 8a: Example test video clip of a short overrecording made on a VCR at SP speed. This VCR has a flying erase head. This figure is an XviD-encoded MPEG-4 video file.

Figure 8b: Example test video clip of a short overrecording made on a VCR at SP speed. This VCR does not have a flying erase head. This figure is an XviD-encoded MPEG-4 video file.

The O and OU segments had the same general characteristics as the longer overrecordings, except they were shorter in length and usually had no pull-down area, OP, because the short overrecordings were shorter than the pre-video erased area.

Between the end of the overrecorded portion and the erased area, there was a short segment of the original underlying recording, U, that was not present on the longer overrecordings.

The next segment contained an erased portion, E, in the middle of the underlying recording. The erased portion physically appeared on the videotape as a vertically erased area across the angled video tracks of the underlying information, as seen in Figures 7a and 7b. When played back, the erased, herringbone-like portion started appearing at the top of the frame, progressively replacing the partial tracks of the first underlying recording segment, UP. At the end of the erased portion, E, the partial tracks of the second UP video started appearing at the top of the frame, with the erased information appearing as a band below it and the first UPP portion disappearing off the bottom first, followed by the erased portion, and finally, only the second UP segment is visible at the beginning of the next U segment. The length and time of the erased area were exactly equal to the length of the short overrecording, either O or OU. One method of determining this distance and its timing is to count the total number of video lines in the erased band and then use the following formulas: information beneath it, as seen in Figures 8a and 8b. These three bands progressed downward in the frame, with the first U

Length (in mm) =  (linear length of video track) x (scan lines in erased band) ÷ (total lines in frame)

Time (in seconds) = length ÷ playback speed

As an example, at SP speed with 350 scan lines in the erased band, the length and time would be:

Length = 96.30 mm x 350 ÷ 525 = 64.20 mm

Time = 64.20 mm ÷ 33.35 mm/sec = 1.92 seconds

The second UP portion occurred across the full linear width of the scanning video heads, a distance of 96.30 mm at SP speed (96.67 mm at EP). On playback this segment lasted 2.89 seconds (about 87 frames) in SP and 8.69 seconds (about 260 frames) in EP. During the visual review of the second UP portion, because only about 485 of the 525 video lines were visible because of blanking, the pull-down effects were visible for no more than 2.67 seconds at SP and 8.00 seconds at EP, even using the underscan feature on professional monitors. This area played back at the correct speed, even if the underlying recordings and overrecordings were at different speeds, because the control-track information was present. If the overrecording measured more than 79.244 mm in length at SP speed (79.523 mm at EP), then the end of the second UP portion lacked synchronization.

The second underlying recording segment, U, appeared after the end of the second UP portion and contained synchronization and full tracks. However, if the overrecording measured more than 79.244 mm in length at SP speed (79.523 mm at EP), then this U segment was not present.

The last segment of this overrecording sequence was UN, which had all of the video information from the underlying recording but no control-track sync. The loss of sync was caused by the erasure of the control-track information in segment E, which was located 175.544 mm earlier at SP speed (175.923 mm at EP). The combined length of UN and the preceding U segment was exactly 79.244 mm at SP speed (79.253 mm at EP), based on the standardized distance between the end of the video heads and the linear audio record and control-track heads. The length of UN was the same as the erased/overrecorded portion whenever UN did not exceed 79.244 mm in length at SP speed (79.523 mm at EP).

Short Overrecordings—Audio Characteristics

The hi-fi audio information was present throughout the overrecording and the first U segment (after O or OU) and then returned at the beginning of the second U segment before UN. These results reflect that the hi-fi audio was present only in the portions that contained complete video fields. If the overrecording and underlying recordings were recorded at different speeds, then a delay usually occurred at the beginning because of the speed change.

The linear audio normally was present throughout the entire overrecording sequence, except for the UN portion; however, if the overrecording was more than 79.244 mm in length at SP speed (79.523 mm at EP), then the linear audio was not present at the end of the second UP portion (the second U segment was not present). Because of the erasure of the linear audio track in the E segment, the audio was not present 175.544 mm ahead of the entire erased portion at SP speed (175.923 mm at EP); therefore, the entire UN segment lacked linear audio. If the overrecording and underlying recordings were recorded at different speeds, then a delay usually occurred at the beginning because of the speed change.

Short Overrecordings—Control-Track Characteristics

The control-track information matched the high-level information on the linear audio track.

Examples

The following three examples of overrecording configurations are commonly encountered in forensic applications and are based on the previously described video overrecording theory and test results. All three examples use 80.00 mm for the pre-video erased area, and the frame numbers have been rounded to the nearest whole number. Both the underlying recording and overrecording VCRs have hi-fi audio, and some of the Video Sequence totals in Tables 2 through 4 have slight variances because of rounding errors.

The first example is a 10.00-second overrecording with the following characteristics: (1) the underlying recording is at SP speed, (2) the overrecording is at SP speed, and (3) the underlying recording and overrecording VCRs have flying erase and hi-fi audio heads. As listed in Table 2 and generically illustrated in Figure 5a, this overrecording configuration is divided into four segments:

  1. Segment O is the overrecording, lasting 10.00 seconds, which is recorded on the videotape over a total of 333.50 mm (10.00 x 33.35) and consists of 300 frames (10.00 x 29.97). The control track is present during this portion, and so are the linear audio and hi-fi audio.
  2. The 80.00-mm erased segment, E, which is the pre-video erased area, has no video, hi-fi audio, linear audio, or control-track information because they were erased by the full-track erase head. This portion lasts 2.40 seconds (80.00 ÷ 33.35).
  3. Segment UNP is a pull-down portion of partial tracks, without synchronization, of the underlying recording. It is 96.30 mm in length, lasts 2.89 seconds (96.30 ÷ 33.35), and consists of 87 frames (2.89 x 29.97). The 96.30-mm distance represents the linear distance of the full helical-scan video-head track at SP speed. The pull-down consists of the underlying recorded information progressively replacing the erased segment. The hi-fi audio does not play back in this area, even though it is present on the videotape, because only partial tracks are present. The linear audio and control-track information are not present because they have been erased by the full-track erase head. If the audio and control data had been present, they would have been located 175.544 mm earlier on the tape (see Figure 5a).

    The last portion of the overrecording sequence is segment UN, which contains the underlying recording without synchronization. It is 79.244 mm in length, lasts 2.38 seconds (79.244 ÷ 33.35), and consists of 71 frames (2.38 x 29.97). The 79.244-mm length represents the distance between the end of the video tracks and the recording control-track and linear audio heads. The hi-fi audio information is present in this portion, but the linear audio and control-track information have been erased by the full-track erase head. Following this UN segment, the underlying recording, U, returns with the video, audio, and control-track information intact.

Table 2: An example of the representative video, audio, and control-track times, lengths, and frames for a 10.00-second overrecording with the following characteristics: (1) the underlying recording is at SP speed, (2) the overrecording is at SP speed, and (3) the underlying recording and overrecording VCRs have flying erase and hi-fi audio heads. The distance from the full-track erase head to the video-recording head has been designated as 80.00 mm, the frame numbers have been rounded to the nearest whole number, all lengths are in mm, and all times are in seconds.

The second example is a 20.00-second overrecording with the following characteristics: (1) the underlying recording is at SP speed, (2) the overrecording is at EP speed, and (3) the underlying recording and overrecording VCRs have hi-fi audio heads, but the overrecording unit does not have a flying erase head. As listed in Table 3 and generically illustrated in Figure 5b, this overrecording configuration is divided into six segments:

  1. Segment OU is the first part of the total overrecording and contains a mixture of both the underlying recording and overrecording because there is no flying erase head. This section is 80.00 mm in length, based on the pre-video erased area between the full-track erase head and the beginning of the video-head recording. This segment lasts 7.19 seconds (80.00 ÷ 11.12) and consists of 216 frames (7.19 x 29.97). The control track is present during this portion, and so are the linear audio and hi-fi audio.

  2. Segment OP is the pull-down segment and the second part of the total overrecording. It starts at the beginning of the full-track erasure and ends 96.67 mm later. The 96.67-mm distance represents the linear distance of the full helical-scan video-head track at EP speed. The pull-down consists of the pure overrecording information progressively replacing the mixture of the underlying recording and the overrecording. This segment lasts 8.69 seconds (96.67 ÷ 11.12) and consists of 260 frames (8.69 x 29.97). The linear audio, hi-fi audio, and control-track information are present during this segment.

  3. Segment O is the pure overrecording segment and the third part of the total overrecording. It lasts 4.12 seconds (20.00 - 7.19 - 8.69), covers a length of 45.81 mm, and consists of 123 frames (4.12 x 29.97). The linear audio, hi-fi audio, and control-track information are present during this segment.

  4. The 80.00-mm erased segment, E, which is the pre-video erased area, has no video, hi-fi audio, linear audio, or control-track information because they were all erased by the full-track erase head. This portion lasts 7.19 seconds (80.00 ÷ 11.12) and has an equivalent of 216 frames (although no video information is actually present).

  5. Segment UNP is a pull-down portion of partial tracks, without synchronization, of the underlying recording. It is 96.67 mm in length, lasts 8.69 seconds (96.67 ÷ 11.12), and consists of 260 frames (8.69 x 29.97). The hi-fi audio does not play back in this area, even though it is present on the videotape, because only partial tracks are present and the playback speed is incorrect (EP versus SP). The linear audio and control-track information are not present because they have been erased by the full-track erase head. If the audio and control data had been present, they would have been located 175.923 mm (96.67 + 79.253) earlier on the tape.

  6. The last portion of the overrecording sequence is segment UN, which contains the underlying recording without synchronization. It is 79.253 mm in length, lasts 7.13 seconds (79.253 ÷ 11.12), and consists of 214 frames (7.13 x 29.97). The hi-fi audio information is not present in this portion because the playback speed is incorrect (EP versus SP). The linear audio and control-track information are also not present because they have been erased by the full-track erase head. Following this UN segment, the underlying recording, U, returns with the video, audio, and control-track information intact.

Table 3: An example of the representative video, audio, and control-track times, lengths, and frames for a 20.00-second overrecording with the following characteristics: (1) the underlying recording is at SP speed (2) the overrecording is at EP speed, and (3) the underlying recording and overrecording VCRs have hi-fi audio heads, but the overrecording unit does not have a flying erase head. The distance from the full-track erase head to the video-recording head has been designated as 80.00 mm, the frame numbers have been rounded to the nearest whole number, all lengths are in mm, and all times are in seconds.

The third example is a 2.00-second overrecording with the following characteristics: (1) the underlying recording is at SP speed, (2) the overrecording is at SP speed, and (3) the underlying recording and overrecording VCRs have flying erase and hi-fi audio heads. As listed in Table 4 and generically illustrated in Figure 7a, this overrecording configuration is divided into six segments:

  1. Segment O is the overrecording, lasting 2.00 seconds, which is recorded on the videotape over a total of 66.70 mm (2.00 x 33.35) and consists of 60 frames (2.00 x 29.97). The control track is present during this portion, and so are the linear audio and hi-fi audio.

  2. The first U segment is the return of the underlying recording for 13.30 mm, or 0.40 second (13.30 ÷ 33.35), with the control track, the linear audio, and the hi-fi audio present. On short overrecordings, the total of the O and first U segments equals the pre-video erased area length.

  3. The E–UP pull-down segment has the same length and timing as the overrecording, or 66.70 mm, 2.00 seconds, and 60 frames. This pull-down consists of the underlying recorded information being progressively replaced with the erased segment, E, until the erased portion is at the top and the underlying recording is at the bottom of the frame. The linear audio and control-track information are present; however, the hi-fi audio is not present because there are only partial video tracks.

  4. The UP–E pull-down segment is 96.30 mm in length, lasts 2.89 seconds (96.30 ÷ 33.35), and consists of 87 frames (2.89 x 29.97). The 96.30-mm distance represents the linear distance of the full helical-scan video-head track at SP speed. The partial tracks of the second UP video start appearing at the top of the frame, with the erased information appearing as a band below it and the first UP information beneath it. These three bands progress downward in the frame, with the first UPP segment is visible at the beginning of the second U segment. The hi-fi audio does not play back in this area, even though it is present on the videotape, because only partial tracks are present. The linear audio and control-track information are present.

    portion disappearing off the bottom first, followed by the erased portion, and finally, only the second U
  5. The second U segment is the return of the underlying recording for 12.54 mm, 0.38 second (12.54 ÷ 33.35), and 11 frames. The control track, the linear audio, and the hi-fi audio are present. On short overrecordings, the total length of the second U and the UN segments equals the 79.244-mm distance between the end of the video tracks and the recording control track and linear audio heads.

  6. The last portion of the overrecording sequence is segment UN, which contains the underlying recording without synchronization. It is the same length and time as the overrecording, or 66.70 mm, 2.00 seconds, and 60 frames. The linear audio and control-track information are not present because they were erased by the full-track erase head; however, the hi-fi audio is present. Following this UN segment, the underlying recording, U, returns with the video, audio, and control-track information intact.

Table 4: An example of the representative video, audio, and control-track times, lengths, and frames for a 2.00-second overrecording with the following characteristics: (1) the underlying recording is at SP speed, (2) the overrecording is at SP speed, and (3) the underlying and overrecording VCRs have flying erase and hi-fi audio heads. The distance from the full-track erase head to the video-recording head has been designated as 80.00 mm, the frame numbers have been rounded to the nearest whole number, all lengths are in mm, and all times are in seconds.

Forensic Applications

The analyses of the test overrecordings and the underlying video-recording principles reflect a number of audio and video parameters that can be measured and considered by the authenticity examiner. The following are recommended examination procedures for analyzing VHS cassettes for suspected overrecordings and a list of observations that can be important to the forensic examiner.

Examination Procedures

The following are the generally recommended examination procedures for reviewing an evidential VHS videocassette in the laboratory. However, other analysis steps or modification of some of the recommendations may be necessary for a specific recording.

  1. The VHS videocassette should be played back on a professional-quality VCR and visually reviewed on a professional monitor with underscan capability and/or digitized and reviewed using an appropriate high-resolution computer monitor (with appropriate hardware and software that enables the user to view individual fields). The separate optimized manual-tracking setting, instead of automatic tracking, should be used for the underlying and overrecording segments because this often produces better playback, especially when there is a speed change. If audio and/or video problems are encountered on one VCR, then playback on other brand/model VCRs is advisable.

  2. The video signal from the professional-quality VCR should be routed through a TBC, which will add stability in the areas lacking a control track.

  3. The linear audio and hi-fi audio should be listened to separately on the original videocassette. Those portions containing no hi-fi audio (when the VCR automatically switches to linear audio) should be identified.

  4. The audio and video information in question should be digitized with a high-quality video-capture/audio-capture device, using an appropriate format that exceeds the quality of the VHS recording.

  5. A software program should be used that can simultaneously display both the audio and the individual video frames and/or fields, with accurate time registration and frame/field numbering.

  6. An initial determination should be made of whether the suspected event is consistent with the longer or shorter overrecording configuration.

  7. Based upon the specific characteristics of the overrecording category, the pertinent audio and video times should be measured. If no audio was recorded on an evidential tape, then record- and erase-head signals, changes in the noise floor, and other indications of segment boundaries should be identified.

  8. The parameters identified with the specific type of overrecording should be compared with the information on the videocassette in question.

Test-Recording Observations

  1. The test recordings clearly showed the importance of the linear audio information, which also reflects the presence or absence of the control track. This was useful even with videotapes containing no high-level linear audio because system artifacts were often recorded on the linear audio track at segment boundaries.

  2. It was important to identify the portions that lacked synchronization because those areas often had video images with a loss of color, added noise, and vertical instability.

  3. Pull-down portions usually did not show accurate video information and had color loss, added noise, and synchronization problems.

  4. In some cases, the linear audio was not completely erased by the full-track erase head; the remaining audio signal was about 30 to 45 dB below its original amplitude. This was probably an effect of the full-track erase head, which is designed to delete the helical-scan tracks and not the linear track information.

  5. The linear audio and the control tracks are in the same head stack, and thus, when linear audio was present, the control track was also present. Similarly, when the linear audio was erased, the control track was also erased.

  6. A thorough examination of the separate video, linear audio, hi-fi audio, and control-track information provided the best representation of the overrecording sequence.

  7. The updating/stationary counter on most real-time counters is a direct indication of the presence or absence of control-track pulses.

  8. Because super VHS (S-VHS) and compact VHS (VHS-C) use pertinent standards that are identical to VHS, the findings of this paper apply to those formats as well (IEC 1991; IEC 1993).

Acknowledgments

The authors thank the following individuals who reviewed this paper and provided important technical and grammatical improvements: Holly Breault and George Skaluba (FBI, Quantico, Virginia); Marla E. Carroll (Unilux, Ltd., Fort Lauderdale, Florida); Dean Catoggio, Bryson Shearwood, and Jason Ferridge (Victoria Police, Macleod, Victoria, Australia); Steven A. Killion (BEK TEK LLC, Clifton, Virginia); Graeme Kinraid (Australian Federal Police, Canberra, Australia); and Wayne R. Runion (U.S. Army Crime Laboratory, Fort Gillem, Forest Park, Georgia).

References

Beeching, S. Video and Camcorder Servicing and Technology. Newnes–Butterworth-Heinemann, Oxford, 2001, pp. 18–24, 150–165.

Epstein, S. Video tape recording. In: Standard Handbook of Video and Television Engineering. 4th ed., J. C. Whitaker and K. B. Benson, eds. McGraw-Hill, New York, 2003, pp. 49–60, 71–72.

Goodman, R. L. Maintaining & Repairing VCRs. 4th ed., TAB Books, New York, 1996, pp. 1–38.

Grob, B. and Herndon, C. E. Basic Television and Video Systems. 6th ed., Glencoe–McGraw-Hill, New York, 1999, pp. 38–49, 211–216, 308–311, 326–360, 366–371.

International Electrotechnical Commission. International Standard 61077 Helical-Scan Video Tape Cassette System Using 12,65 mm (0,5 in) Magnetic Tape on Type VHS—Compact VHS Video Cassette. International Electrotechnical Commission, Geneva, Switzerland, 1991.

International Electrotechnical Commission. International Standard 60774-3 Helical-Scan Video Tape Cassette System Using 12,65 mm (0,5 in) Magnetic Tape on Type VHS—Part 3: S-VHS. International Electrotechnical Commission, Geneva, Switzerland, 1993.

International Electrotechnical Commission. International Standard 60774-1 Helical-Scan Video Tape Cassette System Using 12,65 mm (0,5 in) Magnetic Tape on Type VHS—Part 1: VHS and Compact VHS Video Cassette System. International Electrotechnical Commission, Geneva, Switzerland, 1994.

International Electrotechnical Commission. International Standard 60774-2 Helical-Scan Video Tape Cassette System Using 12,65 mm (0,5 in) Magnetic Tape on Type VHS—Part 2: FM Audio Recording. International Electrotechnical Commission, Geneva, Switzerland, 1999.

Koenig, B. E. Authentication of forensic audio recordings, Journal of the Audio Engineering Society (1990) 38(1–2):9–14.

Luther, A. C. Video Recording Technology. Artech House, Boston, 1999, pp.14–17, 24–25, 75–91, 95–117, 205–208.

Luther, A. C. and Inglis, A. F. Video Engineering. 3rd ed., McGraw-Hill, New York, 1999, pp. 295–309.

McComb, G. Troubleshooting & Repairing VCRs. 3rd ed., TAB Books, New York, 1995, pp. 1–21, 33–35, 37–62, 67–73.

Ryan, D. M. Helical scanners. In: Standard Handbook of Video and Television Engineering. 2nd ed., J. C. Whitaker and K. B. Benson, eds. McGraw-Hill, New York, 1992, pp. 89–98.

Trundle, E. Guide to TV and Video Technology. 2nd ed., Newnes–Butterworth-Heinemann, Oxford, 1996, pp. 223–242, 284–292, 368–371, 374.

Trundle, E. Newnes Television and Video Engineer’s Pocket Book. 3rd ed., Newnes–Butterworth-Heinemann, Oxford, 1999, pp. 344–354.

Weise, M. and Weynand, D. How Video Works. Focal Press-Elsevier, Burlington, Massachusetts, 2004, pp. 15–38, 214–220.