Post-production is the term used to refer to the various preparations and processes which create a finished program or segment from raw footage. Post-production is very often the most expensive phase of bringing a project from planning to completion (as opposed to pre-production and production), but this is not necessarily the case. However, the decisions made regarding post either during the planning or shooting stages, or during the earliest stages of post will have a definite effect on both budget and schedule.

The first consideration is, "what is the final destination of this product?" This will help determine how much money needs to be spent. If your final product will be viewed from a VHS tape, it makes little sense to pay for D1, D2 or D3 (see videotape formats) compositing or editing. However, since VHS is a lower resolution format, it is important that the best signal possible be laid onto the master tape which will be used to make the VHS copies.

Another consideration is the amount of time you want to spend on your rough cut, (which is exactly as its name implies--a rough version of what will be your finished product). If the program is very straightforward--talking heads with a little bit of b-roll to cover any jump cuts (see the glossaryfor these definitions), a rough cut alone might be sufficient preparation before taking the piece into final editing. Any changes from your rough cut can be relayed from the producer to the editor in the course of editing your master tape. Rough cuts are often thrown together purely for the purpose of determining the approximate length of the finished piece, but pieces with more detailed and creative editing of picture and sound will require much more work. Rough cuts in this case tend to mean the various revisions of a piece as it approaches its final form. A fine cut is used to describe a version that is offered as a final version--that is, someone watching the fine cut watches it as if it were the finished piece, and assuming there will be no further changes, can give the ok for the online editing to begin.

In the past, there was a much greater difference between offline and online. While an "online edit" refers to the final mastering of a piece for broadcast or distribution, the traditional online suite--including color correction, or title camera, audio sweetening, etc. (seethe post-production suite is no longer the only place to prepare a master tape. "Offline" can refer to any pre-mastered tape, from a VHS with burned-in (a visible readout) time code to the output of a digital non-linear edit system. But with the advent of digital editing, many projects can be mastered without ever entering an online suite. Because online typically refers to a finished product, as long as the technical specifications of any edit system can meet the requirements of a given project, the output of that system can rightfully be considered an online master. Certain digital edit systems are capable of outputting a signal equal to broadcast quality-- with titles, special effects, numerous audio channels and the capability of equalizing and mixing that audio down to one, two or as many channels are desired. However, digital systems are far more expensive to rent than less sophisticated offline systems, and need adequate drives to store all the material necessary for editing. On the other hand, changes on a non-linear system are far easier to make than on an analog system, as on an analog system, everything from the point of the change onward needs to be re-edited. Digital systems need the raw footage to be loaded into the systems before the editing can actually begin; this is time-consuming, especially if there are many hours worth of field cassettes. All these considerations must be weighed when determining what course post-production will take.

Almost all edit systems are capable of generating an EDL, or edit decision list. The exceptions are very simple control track (as opposed to timecode) based systems. And whether the piece will ultimately end up in an online room or be finished with a fine cut by the offline system, it is a good idea to generate this list. An EDL notes every tape used, the exact in and out points of every cut, dissolve, or effect in the piece (in timecode--hours:minutes:seconds:frames), and the in and out point on the record machine. It is thus a blueprint of your edited master, and invaluable should you ever want to change something in that master, or should you need to track down the source of a particular shot. EDL's are fed into the edit system computer of an online room and are then capable of auto-assembling the piece--that is, cueing the editor as to which tapes are needed, selecting which tracks will be edited onto, cueing up both the source and the record tapes, and executing the edits. Always generate an EDL, even if you think you won't need one. You can format the EDL several different ways: "A mode" generates a list based on the in-point of your record tape, so it is a sequential list. "B-mode" instructs the system to perform all edits from one particular playback source before requesting another source. B-mode editing is sometimes called "checkerboarding", as the record tape will have areas of material adjacent to areas that have not yet been edited. B-mode editing is less time-consuming than A-mode, but, should any problems occur they may not be discovered until late into the edit, and will be potentially more complicated and time-consuming to fix. "C-mode" is similar to B-mode in that it also auto-assembles from one playback source before moving onto another, but C-mode progresses sequentially through the playback source rather than the record tape. A "D-mode" EDL is basically an A-mode list, except that dissolves (and any other non-cut transitions) come at the end of the EDL. Transitions also come at the end of an "E-mode" list, which in other respects is the same as a C-mode EDL. A and B mode are the most common forms of auto-assembly. When you generate an EDL from an offline system you must specify on what type of system--CMX, Grass Valley or Sony for example, your online will take place, as each system is configured differently.

Because it is possible for the editor (the human editor) to override instructions given by the EDL, or redo an edit, or in a variety of ways change the actual edited master to something not reflected in the EDL, it is important to make sure that the EDL is updated and revised to reflect these changes. This is known as list cleaning and list management. A "clean" list will reflect what is actually on your edited master, a "dirty" list will in some ways be accurate but in other ways not. List management is the way in which the editor ensures that the EDL is clean.

Scopes and Timebase Correctors

If your camera was properly white balanced and the lens was set for the proper exposure (see the camera), your recorded video signal should adhere to technical specifications. When it is time to play that signal back, it is necessary to recreate the conditions on the playback machine that existed at the time of the recording--in other words, you will set up your tape to playback levels that were recorded. This is accomplished by setting up the tape to a signal called "color bars", which is also referred to as "bars and tone". Tone is a standard reference signal which plays a 0dB reference audio signal (so you can adjust playback to 0dB). Bars allow you to properly set the levels of various parts of your video signal--video (white), pedestal (black), chroma (color saturation) and hue (color tint). You adjust these levels on something called a "timebase corrector" or "TBC", while monitoring the changing levels by looking at a piece of equipment called a "scope", short for oscilloscope. There two kinds of scopes used--waveform monitors, and vectorscopes.

TBCs stabilize and adjust the video signal. During recording, as each field of video is scanned synchronizing pulses are also laid down. The scanning of each field lasts 1/60th of a second, and any slight deviations from this precise timing can accumulate and result in signal abnormalities. When the picture is played back, if the playback machine does not travel at exactly the same speed as the record machine did, or if the speed varies at all, the result can be an unstable signal. The recreation of the video image according to the timing of the recorded sync pulses is known as the timebase of the video signal, and any error in the timbase can cause the picture to jitter or be otherwise unstable. Timebase correctors bypass this potential problem by taking control of the signal and scanning (recreating) each field at the precisely correct instant, making sure that the timing of the tape on the playback machine is in "genlock"--that is, in sync and in phase with any other equipment used in the system. Timebase correctors take control of other aspects of the signal as well--the correct levels of video (white), pedestal (setup, or black), chroma (color saturation), and color phase (hue)--are all set using controls on the TBC.

Setting levels is critical not just to ensure that your picture will look the way it is supposed to, but because in a broadcast signal, out of spec video can interfere with or cause other disruptions in the signal. Too high chroma results in "tearing" or "ringing"--the color gets raggedly at the edges and areas of the picture will seem to fall apart. Too high video level in titles can interfere with the audio portion of the signal, causing buzzing on some televisions whenever a title appears on the screen. Reference black, which is set at 7.5 IRE on a waveform should not go lower, even though the picture might look better this way--because it will interfere with sync and other signal information.

Analog Editing

Modern editing falls into two categories: linear and non-linear. While some people often believe the terms analog and digital are synonomous with linear and non-linear, they are not really the same. Analog and digital refer to the signals themselves. An analog signal is a continuously changing form of energy--in the case of videotape, electromagnetic energy. A digital signal is a representation of energy (voltage, electromagnetism, acoustic energy) that is accomplished by repeated sampling, or measuring of it. The values obtained by the samples are then translated into the 0s and 1s of computers. Analog signals deteriorate when they are copied, so an analog recording that has been dubbed down several times from its source will display an image or sound inferior to the original. Digital signals do not deteriorate no matter how often they are "copied", because the information is created anew from the 0s and 1s each time.

Linear and non-linear refer to the process--of editing, or even more basically, of thinking. Linear thinking proceeds in an orderly fashion, non-linear thinking jumps all over the place. Though analog editing is most often done linearly, and digital editing benefits from the freedom of being non-linear, it is possible to edit linearly on a digital system and sometimes possible to edit non-linearly on an analog system. The most important difference is this: linear/non-linear refers to the ability of the editing system to make changes which affect the overall length of the segment in question without having to re-edit from the point of that change on. A non-linear system is able to do this, a linear system is not. If I remove 5 seconds from the middle of a 10 minute linear-edited sequence, in order to watch the sequence from beginning to end I must re-edit it from the point of the deletion until the end of the piece, re-recording onto the tape. In a digital system however, my sequence exists only as a series of 0s and 1s, and each time I order the system to play my sequence, or make a change in it, the system simply recomputes the sequence to accomodate my instructions and plays it accordingly. Until I record the output of the computer onto a tape, my sequence resides solely within the computer and can be manipulated and viewed in its entirety. There is no need of control track, nor is there any distinction between assemble or insert editing in a digital system, because the information exists as a digital file. For this reason I can move sections around, extract portions of the piece or put new material into it with a few clicks of the mouse.

On an analog system I can edit non-linearly by auto-assembling a sequence from an EDL using B-mode, or checkerboard assembly. The edits will not be laid down linearly, in the sense that they will not go from the first shot of the piece to the second, then on to to the third, etc., but will be laid down according to which playback source is loaded. I can also edit in a linear--that is, sequential--fashion on a non-linear system, working from the beginning to the end of the piece. But if I later decide that the last 6 minutes of an hour story works better in the middle, I can make this change almost instantaneously on a digital system. On an analog system I would have to re-edit from the new position of the six minute segment until the end of the show.

Analog editing can be either assemble or insert mode. An assemble edit, like an assemble record, lays down new control track as the material is recorded onto the tape. When you do an assemble edit, you will always record video and all tracks of audio. When you insert edit, you can select video only, or audio only, or video and one track of audio, etc., so you have alot more control over your editing. But in order to insert edit you need a tape on which you have pre-recorded control track.

When you perform an edit, the new record signal being laid down locks onto a a specific point in the pulse of control track, resulting in a clean edit. You always need control track at the edit point for a true "edit" to occur. If you make an edit and when you play it back you see noise (also called snow or hash), or if your picture loses stability at the edit point, or if it is constantly rolling, it is probably because you performed an insert mode edit onto a tape that had no previously recorded control track. Tapes without continuous prerecorded control track can only perform edits in assemble mode, and then only when there is control track present at the point of the edit. The surefire way to tell if a tape has control track is to put it in the machine and hit fast forward or rewind. If the counter numbers (either timecode or counter time) on the deck or edit controller change, you have control track. If they don't, even when you fast forward or rewind quite a distance, you've got a blank tape. Another way to tell if you have control track is to look at something called the "servo" light on your deck. If you have control track, the light will remain solidly on while the tape is playing. If you don't have it the light will be off, and if you have a hit in your control track the light will flicker until the machine has relocked to the signal. Yet another way to tell if you have control track is to look at the "Video/Tracking" meter on the deck. When the needle is all the way to the left, registering no movement, you have no control track, and if it wavers momentarily from a stable read, you've got a hit. Remember that you always need control track at the exact point of your edit, regardless of whether you are in assemble or in insert mode. The control track must continue for the duration of your edit in order to continue editing in insert mode. In assemble mode, you can run out of control track a second after your edit point, but your edit and subsequent recording will still be stable, as the new material you lay down includes new control track.

To record control track onto your record tape, you can use as a source a tape that has a pre-recorded signal on it (called a striped tape, a blacked tape, or a crystal, which is written "XTAL"). You can make one of these tapes with something called a black burst generator which is a device that generates the color burst reference signal encoded in a video signal. A black burst generator can be a separate piece of equipment, but professional record decks have an internal black reference signal generator. You can also use any synchronous signal--meaning any signal that is in "genlock", or locked to the system you are working with. Most editors prefer to use a blacked tape, but that is because black introduces no visual interference when viewing edits. As long as the tape has continuous control track insert edits can be laid onto it, and you can as easily use color bars or a tape that has a previously recorded program as black. Convention however, dictates that edit masters are made using brand new tape (which will have a mimimum of dropouts--see dropout) that has been blacked.

The steps involved in making an actual edit depend upon the degree of sophistication of the edit controller you are using. Beta and Beta SP decks are capable of editing with or without a separate edit controller (as are MII machines). One deck (the source) can be remotely controlled by the other (the record). Buttons allow for the selection of which machine is currently being controlled and which tracks (audio and/or video) will be recorded onto (assuming your record tape has pre-recorded control track--if not, you will only be able to perform assemble edits, which record video and all tracks of audio) On a pre-blacked tape, an in-point (and/or out-point) can be set on either machine by cueing the machine up to the proper frame and hitting "IN" (or "OUT" and then, while holding this button down, hitting "ENTER". For a illustration of this and other functions, see the interactive Beta SP deck. Trimming--that is, adjusting the in or out-point--is done by adding or subtracting frames from the originally set point. This is accomplished by holding down the IN point, for example. The counter (either timecode or control track) will reflect the exact frame of the current in-point. By holding down the "IN" button and at the same time pressing the "+" or"-" keys for each frame you wish to add or subtract, you will adjust your in-point accordingly. This is very basic editing, but on a professional deck such as a Beta or Beta SP, it is quick and frame accurate. Many news organiztions edit in such a fashion.

Of course, this type of editing can execute cuts only. In order to do any transition such as a dissolve, you need a more sophisticated edit controller, one that allows you to program two machines to cue up and play simultaneously, and take first one and then the other as a source. These types of edit controllers range from the basic--which will let you do the afore-mentioned effect, and will also let you type in a specific timecode (hours:minutes:seconds:frames) as your in or out-point--to the very complex, such as a CMX system, which can be programmed to trigger countless effects on several different pieces of equipment. Trimming on these keyboard based edit controllers always involves typing in the differential, as opposed to moving the in or out-point frame by frame. These edit systems will also let you type in in and out points, or mark them on-the-fly.

Digital Editing

Digital editing is an exhaustively large subject that can only be described here in general terms. Random access digital editing systems are the current state-of-the-art, with Avid Media Composer being probably the best known and most frequently used in this country. (Avid's Film Composer is similar to Media Composer, but with functions specific to editing film.) Professional digital editing systems are remarkably complex, with proprietary formulas for executing the various functions involved. In general, the interface of these systems provides the editor with one or more monitors for previewing clips (individual pieces of video and/or audio) and viewing the segment during the course of editing, and a timeline which displays a graphical representation of the segment. Clips are organized into "bins" (a reference to film editing--cut pieces of film were kept in bins to keep them from dragging on a dusty floor), and databases allow the editor to sort the bins according to a variety of criteria, or to sift through any bin in the search for a particular shot or types of shots. The editor is also able to custom-design the work space with the selection of several user-designated buttons which perform a variety of functions: setting in and out-points, splicing (inserting video and/or audio at a designated point, thereby pushing everything that comes after that point down the timeline an amount of time equal to the duration of the inserted section, resulting in an increase in the overall length of the sequence), overlaying (covering existing footage with new audio and/or video, without changing the duration of the sequence), extracting (taking video and/or audio out of the sequence and closing up the hole left by it, thereby shortening the sequence), lifting (removing footage without altering the overall duration), creating a variety of effects or titles, etc. Basically any function that an editor would wish to perform on video can be mapped onto the workspace or exists as a function in a pulldown menu within the application.

Before editing can begin on a digital system, you must digitize the material.--that is, converted from an analog signal into the 0s and 1s understood by computers, and stored on drives which will then provide access to the material during the editing and viewing processes. (A digital signal, which does not need to go through an analog to digital conversion, can be brought directly into the computer by something called a "firewire" connection.) Analog to digital conversion is achieved through sampling, which is the repeated measuring of an analog signal at regular intervals, in order to represent that signal. The "sampling theorem" states that any signal must be sampled at at least double the rate that the signal can change, or else an inadequate representation will result. Subsampling is the term used to describe a rate less than what is required by the sampling theorem, and also refers to the technique used to reduce the number of samples taken to measure a particular signal. (see compression) Too low a sampling rate results in an unacceptable loss of information. Too high a sampling rate results in large file sizes.

Each frame of the analog video signal is comprised of two fields which are then interlaced together in order to be displayed. (see also interlacing) Computers, however, do not use interlaced displays, and work from only one "field" of information. So an analog video signal (2 field) must be converted to a digital one (1 field) to bring it into the computer for processing, and this process must then be reversed in order to record the output of the computer-edited sequence. This difference between the way computers scan and the way televisions scan leads to there being a "roll bar" in the picture if you record video from a computer screen-- unless you adjust the computer monitor refresh (scan) rate to be a multiple of 30 (because a TV scans 60 fields per second) you will be able to see the progress of the electron beam as it travels down the screen.

Each clip of digitized material can be assigned any name the user wishes. Each clip's vital information--such as tape number, the timecode at the beginning and end of the clip, etc.--is stored by the computer and can be displayed at the editor's discretion. It is important to remember that a clip is not the actual digitized material--it is only a pointer to that material, which is stored on (most frequently) external drives, facilitating rapid access. Video uses an enormous amount of storage. One second of NTSC video uses approximately 30 megabytes of storage. (HDTV requires almost 4MB per frame, and a whopping 119.2MB per second!) Considering that the raw material for one segment may consist of anything from a few to hundreds of tapes, adequate digital storage is a major consideration. In order to maximize storage capacity, video is often digitized selectively--that is, a decision will be made as to which shots will be used in the edit prior to digitizing, and those shots alone will be digitized. An alternative way to maximize storage, and one that does not compromise the amount of material that can be brought into the system, is compression--filtering out some information during the analog to digital conversion--that will make the resulting files smaller.

Compression

ASCII is a computer language that uses binary (meaning based on the power of 2) representations of letters and numbers. The number 0 is ASCII is 0, the number 1 is 1, but the number 2 is 10. The number 3 is 11 in ASCII, and so on. Since computers only understand 0s and 1s, once it has been translated into ASCII, any information can be brought into a computer or sent from one computer to another. Each representation of 0 or 1 is called a "bit", which stands for binary digit. Each letter of the alphabet is represented in ASCII by an eight-bit value, a combination of eight 0s and 1s--the letter L, for example is 01001100.

An anlog to digital converter samples the voltage of the analog signal. A binary numerical value is then assigned to each piece of information from the sample (thereby translating that information into ASCII). Each pixel of the signal is anazlyzed for color information, for luminance, and for sync, and each separate piece of information is encoded in an 8-bit (called a byte) ASCII value, ready for input into the computer and decoding by it. Because so much data needs to be transferred, digital systems must find a way to limit the transmission of data without losing any valuable information, and they do this with the help of compression algorithms, or formulas. Compression, whether it is accomplished with the use of hardware or software or a combination of the two, has the goal of making the resultant file size smaller while preserving the content of the source.

Compression techniques are divided into two types--lossy and lossless. A lossless compression method takes the original signal, compresses it according to a particular algorithm, and sends it. Once it is received, it is decompressed. There is no difference from the signal that was encoded for transmission and the signal that was decoded upon reception. Nothing was lost. A lossly compression system on the other hand, throws out certain information during the encoding process and once it is discarded, it is irretrievable. So it is imperative that the information that is discarded for the purpose of compression of the file is not information crucial to the recreation of the meaning (whether written, visual or aural) of that file.

Compression is not limited to the digital realm. Even analog video is routinely compressed. The image we see when we watch television is inferior to the original camera signal, and this is partly the result of analog compression. Any procedure which takes the separate signals of red, green and blue as they exit the camera and encodes them with other signals or reduces the amount of the information during processing accomplishes this by compression, and this is done whether the final product is analog or digital, Beta, BeatSP or VHS.

The analog signal is also compressed by a technique called chroma subsampling. Each pixel of a video field contains 24 bits of color information (8 bits each for red, blue and green, the components of the video signal). These 24 bits of information per pixel yield the possiblity of 16.7 million different colors for that one pixel (each 8 bits = 256 possible colors; 256 x 256 x 256 = 16.7 million possible colors). With chroma subsampling Instead of preserving the array of colors made possible by a 24 bit signal, the amount of color information is compressed to only 8 bits, or 1 byte--3 bits of red, 3 bits of green, and 2 bits of blue. Representing the color by only 8 bits yields the possibility of only 256 colors.

A common way to compress the file size of a video clip in a digital editing system is to bring the information in at only partial resolution. If each pixel of a video field contains X bits of information and we sample only half, or a third of these bits, though our signal will be visibly degraded it may still be sufficiently recognizable for certain editing purposes. If we digitize video at only one-quarter of its resolution (see more about resolution in the glossary) we can expect to be able to digitize 4 times as much as if we digitized it at full quality. So the way many long-format sequences are editied is this: a large amount of video is digitized into the computer at a low resolution. While much of this video will not end up in the final sequence, all of it is available to the editor to be used. When the piece is edited, all unused digitized video is purged from the drives, freeing up storage space. The video that actually appears in the sequence is then re-digitized at full resolution and married to the sequence (the software replaces the old, low-resolution clips with the new, full resolution video).

The top image is digitized at Avid Video Resolution 12. Note the artifacting throughout.The bottom image is digitized at Avid Video Resolution 77, which is full resolution. There is no artifacting or loss of quality.

Hardware-assisted compression allows for faster computation (and therefore faster processing of the signal) than does compression using software alone. It also allocates more time-per-frame for analysis, so information can be discarded only on an as-needed basis. JPEG, for Joint Photographic Experts Group, is a hardware-assisted, lossy compression technique based on still images which offers a sliding scale of compression. JPEG compression takes into account the functioning of the human eye, which sees luminance information as more critical than color. The RGB signal from the camera is converted to the Y,R-Y,B-Y signal (see the Video Signal). The luminance (Y) portion of the signal is left untouched in JPEG compression, but chroma subsampling is applied. (see chroma subsampling) The JPEG compression method is a "symmetrical" method, meaning the time it takes to compress the image is equal to the time it takes to decompress it. "Asymmetrical" techniques typically take more time during the compression phase than during the decompression phase of the process. CD-ROMS are an example of asymmetrical compression--it takes alot of time and compression to encode the information onto the disc, but that information can be accessed rapidly.

Some images may be extremely complex while others are relatively simple, with large areas of solid color for example. A complex frame will have more information to process during compression than another frame, but often the same amount of storage is alloted for each frame's data. When that is the case, you are using a fixed frame size implementation method--a certain amount of data is allocated per frame, and if the information in the frame exceeds that threshhold, this information will not be included. With fixed frame size techniques, if the threshhold is set too low, storage space is maximized but entire sections of a complex frame can be missing from the compressed image because the frame contained more than the allowed amount of data. If the threshhold is set too high, the integrity of the image will be maintained, but at the cost of much wasted storage. Variable frame size compression means the amount of data processed per frame increases or decreases depending on the complexity of each particular frame. Variable frame size methods utilize storage capacity more effectively, though the price is longer computation time and more complicated hardware design.

JPEG compression uses intraframe coding, which means that each and every frame that is compressed has all the information it needs to display the frame in its entirety upon decompression. Interframe coding on the other hand, does not compress each frame independently, but interpolates what information belongs in particular frames based on information from previous frames measured against frames that come later in the moving sequence. MPEG, for Moving Picture Experts Group, uses interframe coding. Interframe coding works like this: there are three different kinds of frames--"I" frames, which are intraframe coded, "P" frames, which are predicted from a previous I frame or from another P frame, and "B" frames, or bi-directional frames. B frames take information from the I and P frames preceeding and following them, and interpolate that information to create the B frames. Usually I frames occur every half second, but in instances where there is much movement in the frame they will occur with greater frequency. Though interframe coding conserves storage space, it presents a problem during digital editing, as each frame does not display all of the information. This can severly hamper the editing process, and would be of use only in situations where cuts are being made based on sound (which is brought in at full quality), not visuals. So intraframe coding is still the method of choice in non-linear digital editing systems.

The Post-Production Suite
While the post-production suite evolves along with new advances in technology, its purpose remains the same--to prepare the final master of a segment or program. Today's post-production suite typically includes the following equipment: up to several analog or digital playback sources of a variety of formats--cassette, 1", D1, D2 or D3; a state-of-the-art edit controller; a multi-level switcher; a digital video effects generator capable of creating complex effects; a character generator; several monitors for showing the output of each piece of equipment in the room; an audio console. Some edit suites also have a soundproof booth for recording narration or voice-overs. Complex manipulation and correction of any audio signal is possible with current software and equipment, and often the final sound preparation, called "sweetening", is done in a separate studio devoted entirely to audio. Some post-production suites may still contain a title camera for shooting titles, but for the most part these have been replaced by character generators, which can store numerous titles in memory. These titles can be programmed to animate onto the screen with any number of effects, including banners and transparencies. Post-production suites can be very luxurious, as clients--who pay for the production--like to be present for the final mastering of their commercial or program.

If there is one piece of equipment that is responsible for the post-production "look", it is the video switcher. With a very basic switcher one is able to create a polished, professional edit master, complete with dissolves, wipes, and keys. Most industry switchers now are digital, and have several "cascading" banks, enabling them to perform wildly complicated effects, but one can accomplish alot by creating effects on a simple switcher, recording them, and using that recording as the source for another layer of effects.

Switchers take many different sources and mix them to generate a single signal, which is recorded onto tape. Switchers are capable of generating background colors, or filling letters with color, adding borders around them, softening the edges of the borders, and wiping the whole effect across the screen. They can hide portions of the frame, and some switchers are capable of electronically altering the basic shape of the selected wipe by modulating the signal, causing it to pulse or undulate.

The three basic functions of switchers (aside from cuts) are dissolves, wipes and keys. A dissolve is the gradual mixing of one source of video with another. Though most often one source will eventually dissolve into the other, or into black, it is not mandatory to complete the dissolve in this way. Switchers can superimpose one source over another for any duration, and can vary the intensity of one image over the other. Some switchers have a "spotlight" option which allows you to lower the signal output of most of the image but retain full brightness on a particular, spotlit area.

A wipe uses a pattern to transition from one image to another. The pattern is selected from a number of stored patterns within the switcher. It can be reversed (and move from right to left rather than left to right, bottom to top rather than top to bottom, etc.). Wipes can be used to conceal a portion of the video frame, which may be necessary to hide something, or for artistic purpose.

Keys are holes electronically cut into an image, and then filled with video from one of a variety of sources. There are four different types of keys: luminance keys, which use the white (or in some cases the black) portion of the video signal to cut the hole; chroma keys, which use a specific color to cut the hole; external keys, which use an external source such as a character generator; and matte keys, which cut the hole using luminance, chroma, or external key sources, and fill it using color generated within the switcher itself.