Video Capture Cards for PCs

Video data captured into computer system are organised into consecutive frames of still images. In PAL TV systems that are used in Australia and many other countries, 25 frames of still images need to be captured in one second of time, so that when these frames are played back in the same 25 frames per second speed, movement contained in the video remain smooth for human’s eyes. In a full-sized PAL video frame, there are 768X576 coloured dots or pixels arranged in 576 horizontal lines and 768 vertical columns. In full coloured mode(24-bit colour), each pixel requires 3 bytes of data to represent all possible colour information, therefore 768X576X3X25 = 33177600 bytes or 31Mega Bytes per second of data need to be captured. This is an enormous amount of data stream, and enormous space will be required for storing the captured video data: approximately 111GigaByte per hour of video. Using very fast PCs and huge amount of disk spaces, processing this un-compressed video capture is possible, and some high-end video capture cards do capture video in such an un-compressed mode. However, economy and efficiency in designing and using video capture cards call for alternative ways to capture and store video data in a less demanding fashion: reducing the video data volume before sending them to the PC for storing. Reducing data volume means compressing the data, thus a certain method need to be applied to compress the video data in such a way that significant amount of information are thrown away at the video capture stage, while satisfactory video quality can still be maintained at the video playback stage. Practice has proved that using smart compression methods, reducing video data by 5~10 times at capture stage can still achieve very high quality playback that are almost identical to human eyes as the original video. Apart from those high-end video capture cards, most video capture devices on the market use some kind of data compression mechanism, and very often than not, how they compress video decide how they are used in various application fields.

2. Host-PC Software Based Compression (top)

A series of video capture cards, normally low-end and low-video-quality capture cards, use the host-PC’s software to compress video being captured. These video capture cards normally have very simple analogue video signal to digital data conversion circuitry. These include some early model video capture cards like Intel’s Smart-Video, some current graphics display cards with analogue video input sockets, and the recent streaming video capture cards. Host PC software based video compression requires a video compression software running simultaneously with the video capture card’s capture process, this data compression software normally is called a Codec (CompressionDecompression) software, that has been installed into the PC’s operating system before the video capture process ever started. Modern PCs’ operating systems like MS Windows and Apple Macintosh’s operating systems have several built-in software Codecs(Cinepak, Indeo, Sorenson, etc), plus many other third-party Codecs available freely or as part of installing some application software. The common features of software Codec based video capture cards include low-quality video, low data volume capture, small video frame size, no simultaneous audio input facility built-in, no realtime video output ports, etc. The video files captured normally playback in slower than realtime speed, with poor colour and pixelated images, etc. The main reason for low-quality video is the low-quality analogue to digital signal conversion hardware circuitry built-into these video capture cards: increasing host-PC’s processing power here can not help enhancing the video quality. Even when software Codec used to compress incoming video signal is chosen as None, meaning that no data compression is applied onto the captured video signal, the resulting video quality is still much lower than the video quality of even the basic hardware encoding based video capture cards such as low-end M-JPEG cards. As a result, software compression based video capture cards are mainly used for creating videos for Internet transmission, e-mail, low-quality surveillance, etc, where high-quality video output to external video tapes or video disks are not required. Decent quality video editing normally can not be carried out on video files captured by software Codec based video capture cards. The final verdict is, if converting these kind of video to some form of output format to be displayed on an external TV monitor, the video quality is significantly lower than the original input video quality.

Because of the lack of on-board audio capture and play-back hardware(except for the recent Streaming Video Capture Cards), high-quality audio capture and synchronisation with video is also difficult to achieve: a sound card is required to record audio simultaneously as the video is being captured and compressed, this separate process not only introduced compatibility problem between the video and audio capture cards, but also can easily cause audio-video out-of-sync problem: audio playback lags behind the video playback, in particular when the video playback is long.

Almost without exception, host-PC software compression based video capture cards use Composite(RCA) and/or SVideo(4-pin Mini DIN) connection. Except those dedicated streaming video capture cards, audio input is usually not available on host-PC software compression based video capture cards.

As the name suggested, the Codec software used to compress the incoming video data is also used to decompress the video when video playback is carried out, that is, when playing back software Codec based video on PCs, the original video capture card does not need to be present, as long as the correct Codec software is installed on the PC. This hardware-independent playback capability makes video captured in this method useful to be passed away to other environment for either editing or playback. In recent Internet video streaming applications, all video captured use host-PC based software compression.

3. Motion-JPEG Compression Video Capture Cards (top)

Motion-JPEG is the most widely used on-board hardware compression method to capture Analogue Video Source through Composite, S-Video or Component video. “On-board” here means the video data compression is carried out on the video capture card. The name M-JPEG comes from JPEG, Joint Photographic Expert Group, a cross organization of CCITT and ISO. While JPEG deals with still images, M-JPEG deals with video as a series of consecutively arranged still images.

M-JPEG is an Intra-Frame only compression method: each still image(frame) inside a video clip is only compressed using data within this frame, no consideration is taken from the previous or subsequent frames. This makes every frame to be de-compressed and displayed entirely by extracting data from its own compressed data storage, therefore is quick and accurate each time. Intra-Frame only compressed video are essential for non-linear editing process because the video editing software needs to randomly jumping forward and backward along the video files, cutting and pasting any frames at any point with accuracy and exact result if operation is performed and then cancelled then performed again. These kind of frame-accurate random accessing and editing is handled well by video editing software in association with M-JPEG video capture cards.

M-JPEG video capture cards use on-board dedicated hardware chipsets to handle video compression/de-compression, therefore the video quality is not dependent on how fast the PC is being used, although certain performance of the host PC is required to sustain the required data transfer rate. The data rate for M-JPEG capture cards vary from a few hundreds kilobytes per second to over 20 mega bytes per second, that is, at their highest data rate, M-JPEG video capture can achieve almost the same video data rate as un-compressed video capture. Contrary to the low-end software-Codec based video capture cards, M-JPEG video capture cards normally have much better quality analogue to digital video signal conversion circuitry built-in, and a special software burnt into non-volatile memory(firmware) is also constantly running on the capture card, controlling the video capture and data compression processes. Although varied among quality and prices, M-JPEG capture cards share many important common features:

(1) High-Quality video capture and playback, even basic M-JPEG video capture cards like Pinnacle’s DC10+ can achieve near the original tape’s quality when recording

captured video back to tapes.

(2) Easy Quality Control at video capture stage, including

instant adjustment for colour components, brightness, contrast, saturation, data rate, frame rate, frame size, etc.

(3) Encode/Decode video using on-board hardware,

therefore requires no separate decoding hardware when outputting video to analogue TV/VCRs. M-JPEG video capture cards can convert any video or graphics files, even DV video files into analogue video output in high quality(in contrast, many DV-only video capture cards cannot convert M-JPEG video into DV video without jittering result). Hardware video decoding plus hardware video overlay (though not all M-JPEG capture cards have hardware video overlay) combined can produce perfect full-screen live video display on PC’s. For example, Pinnacle DC30+ can display full screen full speed video through Microsoft MediaPlayer even on PentiumMMX 200 PCs with 4Mbyte video card, with a better video quality than software-based DV video play-back on PentiumIII 1GHz PCs with 64MB AGP display card.

(4) Can capture noisy video signal properly, in this aspect M-JPEG capture cards beat many DV/Analogue dual-mode capture cards like DV500/DVNowAV/DVStorm/RT2500 that convert analogue signal into DV format at the capturing stage. Typical example will be when capturing video from a poor quality old VHS tape, DV500/RT2500/DVStorm etc will capture jittering or garbled frames or complain on dropping frames, while DC30+ captures quite normal video.

(5) Clear still image capturing, in particular still images from fast-moving objects: this is superior to DV video capture cards.

(6) Widely accepted input video device, almost all video output device including DV cameras (through their analogue output sockets) can be used as input for M-JPEG capture cards, and the device types/models rarely cause any compatibility problem.

(7) Connection wires are low in price and easy to make, in particular the RCA connectors. Also the wires can sustain long distance like 5~8 meters without significant quality loss.

(8) Difficult to have accurate device control and no

universally accepted device control standard. Device Control means using PCs’ software to mechanically control the tapes movement inside camera/VCR in terms of play/stop/pause etc operations. Although several video camera and VCR manufacturers have created their own device control mechanism in their own brand of camera/VCRs, including JVC’s JLIP, Panasonic’ 5-Pin, Sony’s LANC, etc, implementing a generic interface on the PCs to control these mutually in-compatible devices is very difficult, and earlier products like Pinnacle’s Studio 200/400 failed miserably in terms of usability.

Typical M-JPEG video capture cards are PCI cards, they include DC10+, DC30+, DC50 and Reeltime from Pinnacle, DigiSuite from Matrox, Fuse and Ignitor from Aurora, etc. They all use their own proprietary software for device drivers and video capture utilities, although to some extent, some of them can work with generic video capture software from ULEAD or Microsoft.

Because of the proprietary device driver and hardware-assisted decoding, video files captured by M-JPEG capture cards are also in proprietary formats: only PCs with the same capture cards installed can read, edit and play these files. This feature makes M-JPEG videos un-suitable for archiving and transmission applications, while editing remains the main application field for M-JPEG video capture.

The differences among different M-JPEG video capture cards include captured video quality, input/output connection, audio capture capability, supported computer systems, supported operating system software and video editing software. The following table lists the differences between two typical M-JPEG video capture cards: low-end Pinnacle DC10+ and mid-range Pinnacle DC30+:

Lower end M-JPEG capture cards like DC10+ are primarily designed for entry-level users: they are easy to use, but normally cannot sustain heavy-duty usage like multi-hour video editing, while middle-ranged capture cards like DC30+ have been extensively used for editing 1~3 hour long video projects. Most low-end M-JPEG video capture cards have no audio capture/playback hardware built-in, relying on PC’s sound card to capture and play back audio, therefore might cause audio/video synchronisation problem in long video clips.

4. DV and Firewire Video Capture Cards (top)

4.1 DV and Firewire: Relations and Differences

There are instances where any video stored on PCs are called “digital video”, even the promotional materials published by many manufacturers claim their M-JPEG or MPEG video capture cards “convert analogue video to digital”. To some extent these are not wrong because any data including video files stored on PC’s hard disk are binary data thus are in digital format (against analogue format stored on analogue video tapes etc). However, in a narrower definition, “Digital Video” refers to video originated from Digital Video Cameras: these cameras store video signal collected from their optical lens directly as binary digital data similar as data stored on PCs. There are basically two types of “Digital Video Cameras”: higher-end Digital BetaCam cameras that communicate with PCs through Serial Digital Interfaces(SDI), and the relatively lower-end “DV” cameras that usually communicate with PCs through “Firewire/iLink/IEEE1394” interfaces. In this article, we limit ourselves to discuss only the video signal created by the secondary type of DV cameras, that is, our “Digital Video”, or “DV” refers to video recorded by those cameras/VCRs in the format of MiniDV, Digital8, DVCam or DVCPro manufactured by Sony/Panasonic/JVC/Canon etc. These cameras/VCRs record video signal in a special way called DV format. DV cameras and VCRs compress video directly before storing it onto DV tapes in a digital format like computers’ data backup tapes.

Sony owns the patent of DV technology, as well as the Computer-to-Camera/VCR connection called iLink. However, iLink came from Apple Computer’s patented data communication protocol Firewire which has become an IEEE standard called IEEE1394. Although often called interchangeably by many people, DV and Firewire are actually two totally different technologies: DV is a video recording and compression technology, while Firewire (iLink/IEEE1394) is a data transmission technology. DV does not have to use Firewire(there are DV cameras that do not have Firewire built-in), and Firewire can be used without DV camera/VCR involved(such as Firewire Hard Disk, Firewire Scanner etc). The transmission of DV signal over Firewire does not use the full bandwidth of Firewire, and Firewire’s application covers much wider areas than transmitting DV signal. DV video has a bandwidth of 3.6Mbytes/sec., while the current IEEE1394a protocol has a 50Mbytes/sec. (400MBPS) maximum data rate. The upcoming IEEE1394b will boost this maximum data rate to over 100Mbytes/sec. (around 1GBPS).

Not only Firewire Video Capture cards are not limited to be used for DV capturing, they do not really compress Video Signal as other types of video capture cards do. The video signal which is digital already inside DV cameras/VCRs are merely transmitted as binary data over the Firewire cable and received by the computers as it is, just like the binary data transmitted between two computers over Ethernet network cable. The video compression process is entirely handled by DV cameras/VCRs, and is very similar to that of M-JPEG video capture cards, except the compression ratio is fixed at 5:1, and the frame size is fixed at 720X576 for PAL, and 720X480 for NTSC video. These parameters combined make the data rate for DV compression at 3.6Mbytes per second for PAL or 3.1MBytes/s for NTSC. Unlike M-JPEG video capture cards, typical Firewire cards do not have video compression hardware built-in, therefore relying on the DV camera and host PC software to compress/de-compress video. This could become an issue if video needs to be recorded back to an analogue VCR while a suitable DV camera is not available: without a suitable DV camera or DV VCR connected, most Firewire video capture cards cannot output DV video from PCs to anywhere! Here “suitable” means the DV camera or DV VCR must have DV-Input capability, i.e., DV signal has to be able to be sent into the camera or VCR --- you might be surprised to know that not all DV cameras can do this!

There are TRUE DV Video Capture Cards that not only have Firewire sockets built-in to transmit DV(or other) signals, but also have DV Hardware Encoding/Decoding chipset built-in to convert non-DV video signal to DV signal, as well as convert DV signal to non-DV video signal without relying on an external DV camera/VCR. These cards include Canopus DVRex, DVStorm, Pinnacle DV500/Pro-One, Matrox RT2000/2500, DVICO FirebirdXE, Dazzle DVNowAV etc. The differences between a Firewire-Only card and a DV Video capture card which includes Firewire ports are summarised in the following table:

From these comparisons, we might say that a DV video capture card is always a Firewire card, while a Firewire card might or might not be a DV video capture card. There are some special Firewire cards like Canopus DVRaptorRT, Pinnacle StudioDV Plus etc, that have some of the DV video capture cards’ features like outputting video to analogue TV directly, without relying on DV cameras.

When outputting DV video back to analogue TV/VCRs, if the Firewire card has no hardware encoding/decoding chipset, the video signal has to be fed into an external DV camera/VCR first, then transmitted(at the same time, no recording to the DV camera is necessary) to the analogue TV/VCR connected to the analogue output socket of that DV camera/VCR. While DV VCRs can handle this transcoding properly, quite a few DV cameras can not: they do not have DV signal input capability! Some DV cameras made a few years ago, and DV cameras sold in Europe have such a nature. For example, Sony’s VX1000 with serial number prior to a certain value all have no DV input capability. These Output Only DV cameras caused such an inconvenience for Video Editing community that a company in Germany even manufactures a device that it claims can enable many DV cameras’ DV Input. Many individuals have long known that certain DV cameras can be re-programmed to enable the DV-Input, by writing some simple commands into the DV camera’s eeprom memory chipset. Instructions and even eeprom burning tools floating around the Internet teaching people the detailed process of tinkering crippled DV cameras.

4.2 Common Features of DV Capture Cards (top)

Although having some important differences, Firewire only cards and DV video capture cards all share many common features that make them different from other types of video capture cards like M-JPEG and MPEG cards:

(1) No parameter adjustment allowed during video

capture and playback, which are really binary data transmission processes. Unlike the M-JPEG or MPEG cards, if you want to make a DV video brighter you cannot do this before and during DV video is played into your PC. This is true even when you are using those TRUE DV video capture cards like DVRex or DV500 that have DV encoding/decoding hardware built-in. There are some exceptions in this aspect: some DV capture cards like Pinnacle StudioDV use pure software mechanism to provide low-resolution video capture from DV cameras, and some recent DV-to-MPEG software tried to convert DV video from generic Firewire cards into MPEG video file in realtime. So far the real usability and success of these exceptions are not obvious. Some DV-Analogue encoding chipsets built by companies like DIVIO do allow lower resolution/smaller-frame size DV-video capture, but video capture cards taking advantages of these features are not available on the market, possibly due to no strong user demands.

(2) Audio is captured and played back together with video signal. When playing back DV video through Firewire cards, audio goes back to DV camera with video and might not go to PC’s sound card, although some capture cards like DVICO Firebird and Dazzle DVNowAV do simultaneously send audio through PCI bus to the PC’s sound card. When audio goes only through the Firewire cable, to monitor DV audio you either have to use the DV cameras’ speakers or re-wire the DV camera’s audio output to PC’s sound card audio input socket (Line-in socket). Audio capture also can choose from multiple audio channels on the DV camera/VCR. At 48-KHz sampling DV camera/VCR have two stereo channels, while at 32KHz sampling they will have four stereo channels. Which sampling rate and which audio channel was used for recording audio was decided at the camera/VCR’s recording stage, DV video capture utility software normally provides selection boxes for choosing from these audio channels.

(3)Device Control is always available for software on the PCs. All DV camera/VCRs use the function control protocol (FCP) defined by IEC 61883, Digital Interface for Consumer Electronic Audio/Video Equipment. This makes the device control from PC’s software relatively easy to implement. In addition to device control signals like Play, Record, Pause, ReWind, FastForward, FastBackward. Firewire Cards can also display the various status of the connected DV camera or VCR: the device is a camera or VCR, the tape is being played or recorded, if there is a tape and if the tape is write-protect, if the camera/VCR is PAL or NTSC standard, etc.

(5)Easy Batch-Capture.

Batch-capture means the user specify several pairs of start and stop timing points within a video tape, then ask the software to capture all the video contents between every pair of points in one go, the video between each pair of points will be captured into a separate video file(as most batch-capture software do), or into one single file(some software like DocuCap for FirebirdXE can do this).

Batch-capture is designed against Manual Capture, where the user hits Record or Capture button of the software to start capturing video at the time point where the tape is playing or paused at. Batch-capture can save disk space by capturing only the portion of video from the tape the user is interested, but it also requires the DV video tape to be properly time-coded: that is, the time-code recorded onto the video tape is continuous from zero at the start of the tape to smoothly progress to the end of the tape. In recording practice the DV camera might be used in such a way that multiple zero points of time code are recorded: this usually happens when people stop recording, rewind or forward the tape(for viewing purpose probably), then re-start recording at a point not exactly following the point where the previous recording stopped. When multiple zero points exist, batch-capture will not work properly: the captured video usually will not be from between the start/stop points you specified, or the batch operation will just fail totally. Sometimes multi-zero point time code can even cause manual capture to abruptly stop. To avoid these situation, it is a good idea for each new DV tape you perform a continuous blind recording(press Record Button while leaving the camera lid on) from the start to the end of the tape, so that a continuous time code is burnt onto the entire tape. After such a blind recording, any subsequent video recording will keep using the smoothly recorded time code therefore no trouble will be caused during video capture operation.

Speaking of Manual Capture, several DV capture cards, including DVICO FirebirdDV/XE and Matrox RT2000/2500, have a nice feature that allows them to pause the DV camera/VCR at a specific time point, then hit the Record/Capture button of the software, so that the video from the exact time code where the tape was paused at will be captured. This is against the usually awkward manual capture process, where the DV tape has to be played first, then the user has to press the Record or Capture button of the software to start capturing video.

(6) Super-Realtime Recording and Playback:

With Firewire’s 50Mbytes data transmission capability(even higher in the new IEEE1394b protocol), and only 3.6Mbytes/sec. (3.1Bytes for NTSC) required for DV video’s normal transmission speed, it is possible to capture video from and record video to DV camera/VCR at faster than real time speed: through application software, transmitting higher than 3.6 Mbytes/sec. DV data and order the DV camera/VCR to spin the tape faster than normal play or record speed will make this happen. This will be very useful to quickly dump long-hours of video from DV camera to PC or vice versa.

(7)Firewire cable is a single 4-wire or 6-wire copper cable with 2 pairs of data transmission wires. In the 6-wire situation, extra 2-wire are used for DC power supply. A single Firewire cable carries all the video signal, audio signal and device control signal to/from the DV camera/VCR. Firewire cables use male connectors, leaving the DV camera/VCR and Firewire cards using the female sockets. Almost all DV cameras use female 4-pin Firewire sockets. Four-wire and six-wire combinations make it possible to produce 3 kinds of Firewire cables: 4-to-4 pin, 4-to-6 pin, and 6-to-6 pin. The length of these cable cannot be more than 4.5 metres. Although the Firewire cables are light and clean, they are not as easy to make as RCA composite video cables, and the female 4-pin sockets on DV camera/VCRs are relatively fragile so extreme care should be taken when plugging Firewire cable into and off these sockets.

(8)Multiple Firewire devices can be connected together through the same Firewire card: upto 16 devices can be daisy-chained from one single Firewire port, upto 63 Firewire devices can be connected from the same Firewire card with multiple Firewire sockets. Many Firewire hard disks and scanners have at least two Firewire ports, and many Firewire cards have 2 or 3 Firewire sockets. Since Firewire uses no ID to differentiate devices, each device plugged onto the Firewire bus will self-initiate and de-initiate when plugged in or off the bus: this provides easy hot-pluggable capability meaning no power shut-down is needed when connecting a Firewire device onto a working Firewire bus, at least in theory. But in practice, many Firewire cards do not work very well without the Firewire devices connected before the whole PC system is powered up. Frequent connecting or disconnecting DV cameras while PC is running also cause trouble on the Firewire card to recognise DV cameras properly. While multiple Firewire hard disks can be connected/daisy-chained together without too much trouble, connecting more than one DV camera/VCRs to the same Firewire card will make none of them to work properly. The reason probably lies more in the PC’s software: single PC’s video capture utility rarely deals with multiple DV cameras/VCRs connected to the same Firewire card simultaneously, therefore DV camera identification is not provided, and device control software is always assuming that the same camera/VCR is connected. This assumption makes it difficult to share several DV cameras among multiple PC users over a network.

Device drivers also play crucial role in making or breaking the Non-DV camera/VCR Firewire device connection: under MS Windows, only those Firewire cards that use Microsoft OHCI compliant IEEE1394 device drivers work well with Non-DV Camera Firewire devices like Firewire hard disks and scanners, Firewire cards using proprietary device drivers such as Pinnacle DV500, Canopus DVRaptor, currently have no support for connecting to non-DV camera/VCR Firewire devices.

(9)Smooth live video overlay on PC’s screen is difficult for Firewire cards without DV encoding/decoding hardware, since the display of DV video is very much relying on PC’s performance. Most PCs today are not powerful enough to play back DV video smoothly in large window size. Depending on the capture card’s hardware design, even some hardware DV decoding cards like Pinnacle DV500 and Matrox RT2000 cannot display large window DV video on PC smoothly. Exceptions in this aspect include Canopus DVRaptor that feeds analogue video signal from the DV camera onto PCs then overlay this video signal live on a window of the PC’s screen, resulting in very smooth realtime display of incoming DV video during the whole process of video preview, capture and output. Another exception would be the DVNow Firewire card from Dazzle, that uses hardware assisted chipset to overlay live DV video onto PC’s graphics card, much like the functions of some M-JPEG cards such as DC30+, giving a smooth and realtime video overlay even on slow PCs. Apple's Mac G4 PC also plays DV video smoothly in full screen without the help of DV decoding hardware. However, rapid performance increases in PCs, and in particular in the video display cards, and manufacturing costs make these special hardware acceleration on DV video display less favourable, with many DV capture cards manufacturers rely entirely on the host PC’s software and hardware to play back DV video.

(10) Moderate Still Image Capture

Still images captured by Firewire capture cards are of fixed size(720X576 for PAL, 720X480 for NTSC) and their image quality are as good as the DV camera can offer: good with still or low-speed moving objects, blurry or distorted with high-speed moving objects. With high-speed moving objects, DV still images are of lower quality than video frames captured from M-JPEG video capture cards. DV cameras that have separate still image storage media like Memory Stick, Flash-Compact Memory cards etc, which might store higher quality still images, normally do not allow still images stored in those media to be transferred over Firewire port to PCs.

(11) Abnormal Video Capture Stop

Un-like capturing video from Analogue video tapes, DV video capture from DV tapes relies heavily on the time-code transmitted from the DV video device. While this might be advantageous in many circumstances, it also introduced numerous problems. Typically, a manual capture from DV video tapes can suddenly stop/abort while the DV camera/VCR is still running with contents showing on their view-finder/LCD. The reason given by the capture software for such an ugly stop are usually more confusing than any help. The fact behind this abnormal stop normally lies in the time-code recording mechanism on different DV video tapes. Sometimes some DV capture card/software will stop whenever a zero time-code appears, or sometimes they stop because a tape is playing back in a DV camera different from the original recording DV camera, or sometimes just for no reason a particular DV capture card will just abort capture on one tape but not on another. In this aspect DV capture cards are poor performer compared with M-JPEG analogue capture cards. As we will see later, MPEG video capture cards will also have some similar problems.

4.3 Capture DV Video into Non-DV Formats (top)

Some Firewire Video Capture Cards have special software to capture DV video directly into Non-DV formats. The video file format conversion can be handled either by the on-board hardware, or by software running on the host PC. Typical examples include Z-Fire from Omni Technology(Realtime MPEG1 encoding),

DC2000 from Pinnacle(Realtime MPEG2 capture), Osprey500 from ViewCast (Realtime Windows Media Technology Video Encoding), Pinnacle Studio DV(Realtime proprietary video format), etc. As overall performance on PC increases, more effort will be spent on utilising the low-priced pure Firewire cards to realtime encode DV video into different formats from PC based software.

4.4 Connection Type to Host PC

While DV Video Capture Cards with hardware encoding/decoding capabilities are always PCI devices, pure Firewire cards(no DV hardware encoder) have many PCMCIA models for Laptop PCs connections, as well as numerous PCI devices. Other connection types have not been seen on the market.

4.5 Software for Firewire/DV Video Capture Cards (top)

Most of the pure Firewire cards have no firmware running on board, and utilise PCs’ operating system support for device drivers: this is true for IBM compatible PCs using Microsoft Windows as well as Apple Computer’s Macintosh PCs, which have had Firewire built-in as standard interface. MS Windows 98 SE, Millennium, 2000 and XP editions all have built-in device drivers for Firewire devices conforming to OHCI format. Standard application software development interface is also supplied in both Windows and Macintosh operating systems, making customised application software development possible by third party software developers without relying on proprietary hardware interfaces. Good examples would be software to realtime encode incoming DV video into RealVideo/MS .wmv formats, thus negating the necessity to have dedicated Streaming video capture cards. Apple’s software development support is called Firewire API, while Microsoft’s support is built into its latest Windows SDK to allow any DirectDraw application software to call some standard interface functions to control and communicate with the Firewire devices connected to OHCI-compliant Firewire cards.

Contrary to Firewire socket only cards, True DV Video Capture Cards with hardware encoding/decoding capabilities have much less common support from the operating systems. Apart from their on-board proprietary firmware, these video capture cards’ device drivers and/or video capture/playback utilities are supplied as proprietary software from the cards’ manufacturers. DVICO’s FirebirdDV/XE and Matrox’s RT2000/2500 might be an exception because they use Microsoft’s built-in device driver for its TI Firewire chipset(therefore they can connect to Firewire hard disks etc).

Another important software for DV/Firewire video capture cards is the DV Codec, the software used to encode and decode DV format video files on the PC. Microsoft Windows supplies some DV Codecs that are used by many OHCI compliant Firewire cards, but many medium to high end DV video capture cards use their own proprietary DV Codec software. DV Codec plays crucial role in the quality of the DV video outputting back to external TV monitor or cameras, they also decide how good a non-DV video file can be converted to a DV format video file. For example, the Windows’ built-in DV codecs can hardly convert any non-DV video file into a perfect DV video file to be played back to external DV camera smoothly, while some proprietary DV software Codec like Digital Origin’s Codec can accomplish this task relatively well.

4.6 Compatibility with DV Camera/VCRs (top)

Being a device conforming to a standard serial communication protocol, Firewire and DV video capture cards have a compatibility issue to solve: not all DV/Firewire video capture cards work with all DV Camera/VCRs, and every DV capture card has some problems working with some DV cameras. Problems can range from system crash to no device control or no video display at all. Contacting either capture card manufacturer or camera manufacturer normally result in wasted time and effort. Each well-known Firewire/DV capture card manufacturer has some DV Camera Compatibility Listing on their Web site or Product packaging boxes, but the fast emerging camera models and different availability of cameras in different countries make these listing difficult to follow: if my DV camera is not on this Firewire card’s compatibility list, is it compatible or not compatible? The answer is unknown until your camera is tried on that particular Firewire card installed in your PC.

Firewire devices connected to Firewire cards communicate with PC's Firewire device driver software all the times, including at system boot up stage, this is especially true with Microsoft Windows built-in Firewire device driver. While this tight integration with the operating system makes connecting Firewire devices easy, it is also easy to cause hanging and crashing of the operating system software when there is a problem with the Firewire device. Typical problem will be connecting DV camera/VCR to an incompatible Firewire card, or simply connecting a faulty DV camera to a healthy Firewire card: PC can crash and hang at booting up or malfunction. In one instance three brand new Panasonic DS15 miniDV cameras caused 5 to 6 PCs with different Firewire cards to either hang or crash or not recognize the cameras. In another instance one Sony TRV17 miniDV camera caused 5 to 6 DV/Firewire cards to malfunction. This problem is normally more serious with Firewire cards using Microsoft Windows built-in Firewire device driver.

There are some non-DV type video devices that also have Firewire interface built-in, such as stand-alone DVD-Recorders, Sony MicroMV MPEG camcorders, etc. These devices normally cannot interface with Firewire video capture cards properly, regardless what their manufacturers claim. Most of the times even their bundled software do not work on PCs, let alone the vast number of third party software and numerous brand of Firewire cards.

4.7 Video Output from DV Capture Cards (top)

It will be ideal if edited DV video can be output to external DV or analogue camera/TV all the time without any extra effort, plus smooth display on the PC’s screen simultaneously. In practice different DV video capture cards implement different video output mechanism: the best is like the implementation of DVStorom and DVNowAV cards: edited video goes out to all possible output devices simultaneously: external DV camera, analogue TV, and on PC screen display. The vast majority of Firewire port only cards output DV video to DV camera only, analogue TV/VCR can only receive video signal through a DV camera’s analogue output connector. Some of the Firewire capture cards need manual switch between outputting video to DV or analogue output ports(such as Pinnacle DV500), others need manual switch between displaying video on PC’s screen or on external DV camera(DVICO Firebird, Pinnacle DV200). Some Firewire Port only cards can independently output video to analogue port without going through a DV camera(Canopus DVRaptorRT, Pinnacle StudioDV Plus).

During and after video editing process, it would be most convenient to display the video on DV camera or analogue TV right from the spot where video is being edited(timeline or story board), but not all of the Firewire/DV video capture cards, or not all of the video editing software can do this: Firewire cards running under ULEAD VideoStudio or MediaStudio need to switch to some output mode and load in a video file before video output can be started, while the Pinnacle Studio DV and StudioDV Plus need to goto a MakeMovie process to play DV video back to DV cameras. The recently released Canopus DVRaptorRT can only output video to analogue monitor from the timeline of its video editing software(Adobe Premiere 6). Not only having these confusing and in-efficient varieties of video output methods, DV video output often is not instant: between user pressing output button and the actual video signal reaches the external DV camera, there is a time delay, that is, after PC’s software has started playing the DV video clip, it is after awhile that the playing video signal shows up on the external DV camera: this often leads to missing the first few seconds of video to be recorded onto the DV camera. To avoid this, many editing software like Adobe Premiere have a dedicated Output to Tape or Print to Tape mode, where a specific time code on the DV camera/VCR can be specified to be the starting point of recording, then the software will pause the camera slightly before the recording point before starting the video play on PC and sending a Record command to the DV camera, for the purpose of recording the entire video clip on the PC precisely at the recording point of the DV camera.

4.8 Realtime Effects Video Editing (top)

During the video editing process, outputting titles, transitions, filters etc, that are applied on video clips immediately in realtime speed to external monitor is one of the most desirable features for video editors. With the appearance of relatively low-priced(below $3000) realtime effects video editing cards like Pinnacle DV500/Pro-One, Matrox RT2000/RT2500, Canopus DVRaptor RT and DVStorm, some of the effects and transitions can be applied without going through the so-called “rendering” process: the software compilation of the effects/titles into the original video clips. The simplest example of realtime effects is superimposing a title over a video: after arranging the title somewhere on the video and clicking the Play button, on the monitoring TV screen, the title shows up instantly in front of the video playing underneath in realtime speed. Here “on the TV” and “playing in realtime speed” have to be emphasised: although long before these realtime effects video capture cards, some video editing software already allow instant preview of applied effects on PC’s screen, but the effects do not output to external video device such as TV or camera. Yet another type of video editing cards and software, such as those low-end OCHI-compliant Firewire capture cards running under Adobe Premiere 6 using MS device driver and MS DV Codec, can instantly preview any effects on the external TV or camera, by manually drag video play indicator over the video and transition slowly with Alt Key pressed simultaneously, but the video’s play speed is not realtime(not 25 frames per second for PAL). Only video capture cards that display effects instantly on an external TV or camera in realtime speed can be counted realtime effects video capture card. Some of the features of these entry-level realtime-effects video capture cards can be listed as follows:

(1) Almost all of them capture video in DV format, even when the actual input video is from analogue source such as S-Video or Composite video: the captured video files are still in DV format because the analogue video signal is instantly converted into DV video by the video capture card.

(2) Video files that can be used for realtime effects have to be captured by the same video capture card, and they have to be in exactly the same format(if the same video capture card offers different video capture format).

(3) Realtime effects can certainly be output to analogue socket such as S-Video and Composite video socket, but only Canopus DVStorm and DVRex can output realtime effects to DV video camera/VCR through Firewire, the other cards(Pinnacle DV500/Pro-One, Matrox RT2000/RT2500, Canopus DVRaptorRT) have to render every effects when the video output is through DV cameras or DV VCRs.

(4) All realtime effects are video capture card dependent: only those effects that are supplied or modified by the video capture card manufacturers can be output in realtime, otherwise any effects/transitions/filters/titles will have to be rendered. The types and numbers of realtime effects offered by each video capture cards vary, and there is no such a video capture card that offers all realtime effects that all other capture cards can offer: there is always some realtime effects offered by one capture card that others do not offer. Third-party application development toolkits do not exist, so no third-party realtime effect is possible.

4.9 Video Quality Issues for DV Capture Cards (top)

Even in Firewire/DV video capture, occasionally corrupted frames --- video with noise or totally non-related contents or coloured-image turned into black and white etc--- can be seen in the captured video, this is especially true when the input video footage has high-level noise, such as video recorded into DV camera from analogue video sources. Corrupted video frames could be due to bad data transmission result between DV camera and the Firewire/DV capture card, which might not always be picked up by the error correction procedure of the Firewire communication protocol, or could be a bad data file creation process that stored corruted binary data. Another symptom is that when using DV video tapes recorded in one DV camera/VCR, then play back in a different DV camera/VCR, corrupted or jittering video frames will appear in the captured DV video files. Contrary to the widely distributed media claims that all DV capture produce exactly the same quality as the original video tapes, almost all DV/Firewire capture cards on the market can capture corruted video frames at some stage, these include the mid-range products from Pinnacle/Matrox/Canopus/Dazzle/DVICO and all other low-end Firewire-only capture cards from numerous manufacturers.

There has been a common (mis-)understanding that DV video captured from different Firewire cards all have the same quality, provided no frame is dropped during capture and playback. However, this understanding does not take into account the fact that, different format of DV video files inside PC, have to be decoded differently by different DV decoding software(Codec) during the video output process. More over, when these DV video clips are being edited, they will be encoded differently by those different DV encoding software or hardware(Codec). These different decoding and encoding processes will make even perfectly(remember they are not always perfect!) captured DV video files to output different quality video. Different hardware design and data file storage methods also result in different quality DV video captured & output by different Firewire cards.

In practice many low-end Firewire cards using Microsoft Windows’ built-in software Codec(ADS Pyro, Pinnacle StudioDV etc) often output DV video with occasional jittering, while several proprietary software DV codec based Firewire cards (DVICO Firebird, Canopus DVRaptor etc) can output their video perfectly smooth, although both kind of cards have no hardware DV Codec chipset built-in.

Most DV video capture is for editing. Editing involves changing the content of the video by adding titles, graphics, filters, transitions and special effects. These modified portion of the video has to be seamlessly incorporated into the raw DV video captured: every frame of the modified video has to be re-built pixel-by-pixel mixing the changed content with the original raw video, in the format strictly conforming to the requirement of the DV video capture card. This complicated video re-creation can be handled either by software or hardware: by software will invoke software DV encoder associated with the video editing software and DV video capture card, by hardware will invoke some on-board DV encoder built onto DV video capture card. Since the quality and efficiency of DV encoders vary, the result of edited DV video sent back to DV cameras can not always keep the high quality as the captured DV video. This is especially true when converting Non DV video format into DV format video. Using low-end Firewire port only capture cards, converting non-DV format video into DV video will almost always show jerky and jittering video on the DV camera’s screen. Another easy example is adding swift transitions like short period page turns or band-wipes into captured DV video. Quite often low-end Firewire cards will generate jittering around the added transitions during DV playback. Pure Firewire socket cards(no DV hardware encoder built-in) need software encoding for edited video contents, normally the software DV encoding is realised during the video rendering or exporting process from within the video editing software: the results depend on the Compression Method(Codec) selected, the Firewire card and the device driver for the Firewire card. Using the same editing software Adobe Premiere 6.0 for Canopus DVRaptor and ADS Pyro, both are Firewire Socket Only cards, with complicated transitions/titles rendered, generally DVRaptor produces better quality output video than Pyro: the play back video around the added transition area are cleaner, smoother, and with no obvious artificial jittering. This clearly indicates the differences of software compression methods and device drivers used by the two cards: DVRaptor has its own proprietary codec and device driver, while Pyro uses generic DV codec and Firewire device driver provided by MS Windows. Interestingly, these quality-lowering problems can even happen on DV video generated by some hardware-based DV Codec: capture cards like Pinnacle’s DV 500 can occasionally generate jittering video output around transitions area when displaying video to DV camera.

Regardless software or hardware Codec involved, lower-quality DV video output due to encoding/decoding defects should be differentiated from problems due to poor PC system performance: the latter problems can cause much more severe symptoms like frame skipping, video play stopping and total video blackout etc. Some DV capture cards like Matrox RT2000 have warning messages displayed when PC system performance can not meet the requirement of DV video playback.

Being a video digitising device, DV cameras themselves can introduce un-wanted result, such as mosaic blocky frames that can be seen from tape playback when the shooting camera was swirled quickly from one scene to another totally different scene.

4.10 Capture Analogue Video into Firewire Card (top)

With the proliferation of the low-cost Firewire Only cards, how to capture analogue video into these cards have prompted the appearance of Analogue-to-DV conversion devices. These external boxes convert incoming composite, S-Video or Component analogue video signal plus audio signal into combined DV signal and send the DV signal out through a Firewire socket, making an appearance to the Firewire card that a DV camera or DV VCR is sending signal to it. After video has been captured into PC’s hard disk, a reversed process can convert the processed video back to analogue TV/VCR connected to the analogue video/audio output sockets of the same converter, without even re-wiring the Firewire cable. This will make connecting DV Camera unnecessary to output video to analogue TV/VCRs from Firewire PCs.

Combining some low-end Analogue-to-DV converter with a low-cost Firewire Only card(or Firewire ports that have been built into PC’s main board) will create a make-shift DV Video Capture Card with hardware DV encoding/decoding capabilities. However, the performance and stability of such a combination is far lower than any of the True DV Video Capture Cards with built-in DV encoding/decoding hardware chipset.

Since the Analogue-to-DV converter are built independent of the Firewire card or port, it is not un-usual the two devices have various problems in working together. Typical problems are related to DV device control: since the Analogue-to-DV converter has no Device Control, which kind of DV device should the Firewire card treat it as? If it is a DV Camera, the DV signal should always pass through since the “DV Camera” is turned on: but the analogue signal might not be fed into the AV-DV Converter all of the times. If it is a DV VCR, then how can you play and rewind the tape? In fact, all of the AV-DV Converters tried to play the role of a DV VCR, this can be observed from the status displayed by some Firewire DV capture card like Firebird, and also from the fact that the Play button is always enabled from the video capture software. Video capture can be achieved on some Firewire cards by clicking Play then Capture button while others just do not show any video signal coming through. Video playing back to Analogue TV/VCR can also be achieved on some Firewire ports only cards. The following table lists the working status of some of the Firewire cards when connected to Dazzler’s Hollywood DV Bridge Analogue-to-DV Converter:

For PC Firewire cards it looks like that those using Microsoft Windows’ device driver and DV Codec have a better chance to work well with AV-DV Converters than those using proprietary software: although in theory, these AV-DV converters should have no relationship with the software used by the Firewire cards.

When the AV-DV converter does work with a particular Firewire card, other problems might also appear, typically there are delays in outputting video signal after pressing Play button, the output video quality is lower than quality Firewire cards, wrongly recognized TV formats(PAL and NTSC mixed up), etc. From practice it has been found that booting up PCs with live analogue video signal continuously fed into the AV-DV converters often help the PC recognize the converted incoming analogue video signal as DV video.

4.11 Dual-Mode DV AND Analogue Video Capture Cards (top)

As DV video format getting popularity among cameras, VCRs and other video recording/displaying devices, while in the same time many other devices will still remain in the analogue video format for a long long time to come, many video capture cards manufacturers started building both DV and Analogue video Input/Output interfaces into their products, thus currently on the market quite a few video capture cards boast “DV AND Analogue I/O” interfaces. Typical products such as PinnacleDV500/MatroxRT2500/CanopusDVStorm/DVICOFirebirdXE have been on the market for a while, which normally are video editing capture cards. Recently even some low-end realtime MPEG encoding capture cards have added Firewire port along with their traditional RCA and S-Video connectors, costing the manufacturers roughly extra US$10 per capture card. Although having the similar Firewire plus Analogue input connectors, the video capture mechanism among these dual-mode capture cards can be very different, which normally result in a very different application of them.

Dual-Mode I/O Video Editing cards(FirebirdXE/DV500/StudioDeluxe/RTX.100 etc) on one side, normally use on-board hardware to convert incoming analogue video signal to DV video signal, resulting in that all captured video files, regardless their original video sources are DV or analogue, are of the same format and data rate/frame size etc, therefore can be edited in the same way. Low-end MPEG encoder cards on the other side, use Firewire port differently from their analogue video ports: video captured from Firewire port does not get converted into analogue video by hardware, rather they are processed by some host-PC based software encoder to be compressed into MPEG video files. This software based DV to MPEG encoding is very different from their on-board hardware based analogue-to-MPEG encoding, resulting in different quality and flavour of MPEG video files. Further more, adding simple Firewire port does not transform realtime MPEG encoding cards into a video editing card, in particular the analogue video file captured is still in the un-editable MPEG format, while their DV video capture capability, to say the best, would be like a low-end $50 Firewire PCI card can offer.

For those dual-Mode I/O Video Editing cards that use on-board hardware to convert analogue signal to DV video, such as DV500 or RTX.100, they can also convert edited video stored on PC’s hard disk directly to analogue output devices such as TV or VCRs, without relying on any connected DV camera/VCR, making a live video monitoring and outputting during editing process easy and convenient. However, for these same capture cards, capturing noisy or degraded incoming analogue video signal to DV format video causes extra difficulties: sometimes a Time-Based-Converter(TBC) is needed to correct mis-aligned sync. signals from the analogue source before feeding the resulting video to the video capture cards, or distortions will appear in the captured video files. This makes them inferior comparing to those (old-styled) M-JPEG based analogue only video editing cards such as DC30+, which can handle similarly noisy or degraded analogue video signal much better.

4.12 DV/Firewire Summary (top)

DV or Firewire video capture cards offer relatively high-quality, moderate data rate and easy device control, thus making them a strong competitor over the traditional low-mid range M-JPEG video capture cards. Currently DV/Firewire video capture cards are doing well in serving the low-to-mid level quality video editing applications, plus experimental involvement in some new application fields traditional M-JPEG capture cards face difficulties, such as video streaming, realtime MPEG conversion, even video conferencing over Internet.

However, the limitations of DV capture are also obvious: inability to adjust capture parameters, difficulty to capture analogue video, reliance on DV camera for analogue video output and many many compatibility problems with different video cameras and VCRs.

5. MPEG Video Capture Cards (top)

Unlike all the video capture cards discussed above, MPEG Video Cards are not built for video editing. They capture video primarily for distribution or archiving. The reasons why M-JPEG and DV video files are not suitable for distribution or archiving can be easily seen in their captured video files: the files are huge in volume and proprietary in formats: only PCs with the same capture cards can read and playback M-JPEG and DV video files properly. On the other hand, MPEG video files are relatively light in data volume, and the decoding software has been built-in to many PCs’ operating system or even non-computer devices like VCD or DVD players. In particular, the MPEG1 video decoding capability is almost universally available on many computers’ operating system software, including Windows, Mac, Linux, Unix, etc. As the DVD getting more and more support from consumers and manufacturers, sooner or later MPEG2 decoding capability will be built into many computers’ operating systems: Windows Me and Windows 2000 already come with DVD playback utilities(though relatively primitive), newer versions of Mac and Windows will be almost certain to support more MPEG2 playback capability. Even without operating system’s support, current MPEG/DVD playback software are widespread and in very low prices.

MPEG video files can be of many different formats. In this article we confine MPEG video files to those that can be decoded entirely by using MPEG decoding software, excluding video files in video capture card dependent formats, although some of them were produced using some sort of the MPEG encoding algorithms. We also confine “MPEG video capture cards” to those that can produce MPEG video files(as indicated above) in realtime, without the need to further converting the video files after they have been captured onto PC’s hard disks. Video capture cards like Pinnacle DV500, DVStorm etc therefore are not considered as MPEG video capture cards, because on PCs without these video card cards or without some special decoding software supplied by the capture card manufacturer, their captured video files can not be played back by software MPEG decoders. Video capture cards like Dazzle DVCII, DVICO FusionMPEG, etc produce only video files that are playable by any MPEG software decoders therefore are considered MPEG video capture cards. Some video capture cards like DataTranslation Broadway Pro, Pinnacle DC2000 can capture video in both hardware-dependent and software-dependent formats, therefore are also considered realtime MPEG video capture cards.

The MPEG name comes from the Moving Picture Experts Group, an ISO committee that defines compressed video and audio data within a certain data rates. The highest data rate defined for MPEG video+audio is 100Mbits/sec. (12.5Mbytes/sec.), while no lowest data rate was defined. In practice, MPEG in particular MPEG1 capture cards can achieve very low data rate suitable for Internet/Web transmission. While higher data rates from 1.1Mbps above are suitable for CD ROM / DVD ROM video encoding and higher-quality video transmission such as Digital TV etc.

The MPEG standards do not specify the implementation of encoding process, therefore the quality and efficiency of MPEG Encoding/Decoding hardware and software substantially vary. Ranging from absolutely free software downloadable via Internet to half-million US dollar commercial hardware encoders, MPEG encoding devices are the most widely available in different forms and channels.

Unlike M-JPEG and DV video capture, MPEG capture cards use combined Intra-Frame and Inter-Frame compression methods, resulting in much higher compression ratio. The Inter-Frame compression method uses video content differences between previous and subsequent video frames to save considerable amount of data by recording only these differences, not an entire video frame content.

MPEG capture cards can produce excellent quality video at a data rate considered as low by M-JPEG capture cards. For example, the SVCD template in DVICO’s FusionMPEG MPEG2 capture card uses 2.5Mbps(290 Kbytes/sec.) variable data rate to achieve video quality far-better than the video quality M-JPEG card Miro DC30+ can achieve at 4.8Mbps(600Kbytes/sec.). At standard DVD Movie data rate 8Mbps(1Mbyte/s) the Dazzle’s DVCII captures video from MiniDV tapes through S-Video socket to achieve visually almost identical result as the incoming DV signal which as we know is compressed at data rate of 28.8Mbps(3.6Mbytes/s).

As in previous sections, we first list the important common features and problems of MPEG video capture cards, then discuss their different implementations. Some of the features and problems discussed will be more related to the MPEG video encoding/decoding methodology than just the physical capture cards, that is, video compression method MPEG inevitably has some features/problems, such as being difficult to edit, while the implementation of a particular MPEG video capture card does not contribute to alleviate such a difficulty.

(top)

5.1 Common Features of MPEG Capture Cards

(1) Many encoding parameters can be adjusted before video capture starts, this is similar to the M-JPEG capture cards. The incoming video/audio signal filters like brightness, contrast, hue, saturation, volume etc are always available for user adjustment.

Unique parameters that can be adjusted for MPEG video capture include:

-- Video Line Masks at top or bottom portions of the video frame. These are for shifting the encoded video frames to avoid flickering noisy video display at the top or bottom lines, which actually are flip-over images from the other side of the frame(flickering lines at top came from bottom, flickering lines at bottom came from top).

-- Video Line Shifting towards left of right of the frame, to avoid too much black stripes on either side. Not all MPEG capture cards provide video line masking and shifting options, because these features are video-encoding hardware dependent. Pure software MPEG encoders (no MPEG capture card hardware involved) normally do not provide these options because they cannot have an instant feedback of the video after changing these encoding parameters.

-- Video-Audio Synchronization boxes. These are hardware features directly derived from the encoding/decoding chipset(for example, Dec 21230) that once enabled, can make the played-back video and audio much better synchronised through an external TV connected to the MPEG capture card, while at the same time disabling the on-screen video display on the PC. Video-Audio synchronisation is a more serious problem in MPEG Video capture than in M-JPEG and DV video capture.

-- Constant Data Rate(CBR) or Variable Data Rate(VBR): these are for balancing between quality and file size of the captured video. Constant Data Rate will result in a fixed file size within a certain video capture time, while variable data rate will make the capture process to dynamically adjust compression ratio: more data for complicated video segment, less data for simpler video segment. Choosing variable data rate normally results in smaller video file size, with no video quality difference from or sometimes better video quality than using Constant Data Rate. However, Variable Data Rate obviously requires more computing power, therefore when host PC is not powerful enough to guarantee no dropping frames, CBR is a better selection.

-- Frame Size and Data Rate are adjustable, but with many restrictions. Data Rate is normally restricted by the encoding hardware, while Frame Sizes follow the MPEG standard definition and cannot be set to some arbitrary values. The following table lists some of the common encoding frame size and data rate selection range used by most MPEG video capture cards:

MPEG1

MPEG2

VCD

SVCD

DVD

Frame Size

352X288(PAL)

320X240(NTSC)

720X576(PAL)

720X480(NTSC)

352X288(PAL)

320X240(NTSC)

480X576(PAL)

480X480(NTSC)

720X576(PAL)

720X480(NTSC)

Data Rate

<=4Mbps

<=15Mbps

1.15Mbps

2.5Mbps

4~8Mbps for MPEG2 Video

-- Group Of Picture(GOP) Structure

In MPEG video creation process, first a totally Intra-Frame encoded frame is created, this is called I-Frame. I-Frame is used to create P-Frames: Predicative Frames which normally record only changes predicted to happen after the I-Frame, therefore saving storage than recording an entire new I-Frame data. In between I-Frame and P-Frames, B-Frames(Bi-directional-Frames) are constructed using information from both the I-Frame before itself and P-Frame after itself, therefore saving even more storage than P-Frames. The arrangement of I, P and B frames inside MPEG video files are Group Of Pictures or GOP: normally expressed in the way of IB…PIB…P, where the number of frames from I-Frame to the last P-Frame immediately preceding the next I-Frame is the N (group length, 1<=N<=15) parameter, and the number of B-Frames plus 1 is the M(0<=M<=3) parameter. Between two I-Frames the formation of B-Frames plus P-Frame is repetitive, such as:

IBBPBBPBBPBBP(N=13,M=3), or IBPBPBPBPBP(N=11,M=2) etc. Obviously the longer is the GOP length(N is bigger), less storage space is required per fixed time of video with lower video quality, while the shorter is the GOP length(N smaller), more storage space is required with higher video quality. Many MPEG2 encoding cards give N=13, M=3 as the default DVD movie encoding values. Some MPEG capture cards also allow user selection of capturing video in I-Frame only, IP Frame only, or IBP Frames, with I-Frame only being most suitable for video editing.

-- Separate or Combined Video and Audio Files

MPEG encoding can create combined video and audio files as *.mpg, *.mpe or *.m2p files, or create separate video and audio files. In MPEG1 encoding, combined file is called System stream, in MPEG2 encoding, they are called Program stream. When encoded in separate video and audio files, audio can be of *.mpa extension, video can be in *.mpv, *.m2v, etc. Some VCD, or DVD authoring software require or prefer input MPEG files to be in separate video and audio formats, while combined video+audio MPEG files are easy to handle for MPEG playback software such as Microsoft MediaPlayer.

-- Audio Format(PCM/MPEG/AC)

Audio encoding in MPEG can be in PCM, MPEG or AC format, with the MPEG1 Layer2 being the most common default format.

(2) High Ratio of Video Quality over Data Rate.

As discussed early in this section, at the same data rate MPEG video can achieve far better quality video than other compression methods like M-

JPEG and DV.

(3) Different Device Connection to Host PC

MPEG capture cards have been implemented in different connection types to the host PC, including PCI, USB, and even Parallel port! In fact many early MPEG1 capture cards were Parallel port (printer port) devices, such as the Python, Snazzi Parallel Port, and Pinnacle Studio MP10. Parallel port MPEG1 capture cards tend to have many audio-video synchronisation and video playback quality problems, therefore have been virtually abandoned recently by manufacturers. USB(USB1) interface are successfully used for MPEG1 and low-mid data rate MPEG2 encoding, but its limited bandwidth(max. 12Mbps and in general can sustain far lower than that) is not suitable for high quality MPEG2 video capture. In theory PCMCIA Bus interface is cable of MPEG video capture(for both MPEG1 and MPEG2) but no actual product exist implementing a PCMCIA MPEG capture card. The new and faster USB2 standard might introduce new MPEG capture devices using that interface.

(4) Various Video and Audio Input/output Connections

MPEG video capture cards can take video from all sorts of input sockets, including Composite Video, S-Video, Component Video, Firewire, SDI, etc, with Composite and S-Video being the most commonly implemented.

Audio capture is built into most of the MPEG capture cards, but some of the MPEG1 video capture cards are built without audio capture hardware, such as the BroadwayPro from Data Translation and StudioVCD from Pinnacle: they rely on a co-operating sound card to capture audio simultaneously when video is being captured. No on-board audio hardware tends to introduce compatibility problem with audio card, and have more chance to cause audio/video out-of-synchronisation problem.

Not all MPEG Video Capture Cards have video/audio output connectors built-in. Unlike M-JPEG video capture cards, MPEG capture is not primarily for editing the captured video, therefore sending captured video back to video tape is not absolutely necessary. Although it will be nice and convenient to have video/audio output sockets on board, many MPEG capture cards including some expensive cards like Darim MPEGAtor have no such connectors. Outputting video/audio involve de-coding MPEG data in realtime using hardware, manufacturing costs might be a consideration as well as the capability of the MPEG encoding hardware chipset: not all MPEG encoding chipsets have de-coding capability.

When a MPEG capture does have video/audio output connectors, live video/audio decoding can happen during video capturing process: TV can be connected to the output port, therefore providing a useful way to preview the video/audio before or during they are captured.

MPEG video output is also useful for playing back any MPEG files to TV or VCR, no matter how these video files were created: by capture card, by software encoder, by Satellite receiving facility, etc.

(5) On-Board Hardware or On-Host-PC Software Encoding

MPEG video/audio files can be created by software running on the host PC taking incoming video signal from some form of input ports, this prompted some MPEG capture cards being built only supplying video signal input but using a PC-based software to finish the MPEG encoding process. This is in contrast with the on-board hardware encoding MPEG video capture cards: where the encoding is entirely finished by the hardware before the MPEG data stream is transmitted to PC to store on the hard disk. With hardware encoding, the video quality is fixed: quality mainly depends on the MPEG encoder chipset and the design of the video capture card. With the software encoding, many factors affect the video quality and video-audio synchronization: the speed of PC, the algorithms used by the software, the status of the PC’s operation, etc. There is a theory that as PC’s hardware speed enhances, encoding can be handled by the software on the host PC with a better and better encoded video quality. While PC’s performance does improve dramatically every year, too many un-predictable conditions on an operating PC have made it difficult to see even MPEG1 video being reliably encoded realtime in high quality by host-PC software-based captured card. Host-PC software-encoder based MPEG capture cards like Pinnacle Studio VCD, Darim VideoOffice, Vitec DCM all suffer stability, quality and video-audio synchronisation problems. As a realtime data collection and compression operation, a fully dedicated hardware chipset(which in essence is software burnt-into firmware running in an un-interruptible status) seems to easily beat some sophisticated software running on the host PC which normally is not a realtime operating environment.

(6) Widely Available Software Encoder and Decoder

As discussed above, some MPEG video capture cards use software running on the host PC to encode incoming video signal into MPEG format files, this process can be called Realtime Software MPEG Encoding. MPEG encoding software can also be used in a non-realtime fashion, that is, by using video files(not in MPEG format) already stored on hard disk as input, software MPEG encoders can output MPEG video files, that is to convert these non-MPEG video files into MPEG video format. Encoding MPEG video files in this way will not have the time constraints imposed by the realtime video capture cards, therefore sometimes can create very good quality MPEG video(depending on which software is used). Also software MPEG encoding process can offer many flexibility in parameter settings. For example, normal realtime MPEG video capture cards have strict limits on the frame sizes and data rates, but software encoding in non-realtime mode normally can set very different frame size like 720X576 for MPEG1 or very high data rate for MPEG2 like 15Mbps, therefore improving the encoded video quality.

As overall PC's performance increases, software MPEG encoding in non-realtime mode also becomes faster. Encoding DV video files captured by Pinnacle DV500 into MPEG1 files will take about 1.3 times of the video files video length(1Minute video takes 1Min. 18sec. To encode), if a PentiumIII1000MHz CPU, 1024Mbyte RAM, 7200RPM IDE Hard Disk, ASUS Gforce 64MB AGP display card and MS Windows 2000 are used.

Many software MPEG Decoders also exist, they can be in the form of a stand-alone MPEG/VCD/DVD playback application software, such as PowerDVD, WinDVD, but can also be installed as part of the Operating Systems’ multimedia playback Codec, such as Mediatrics DVDExpress. Many MPEG software encoders and decoders also exist as freeware.

MPEG video files are the most widely accepted video files across multiple platforms. For example, MPEG1 video files created on MS Windows environment can be directly played back by Apple Macintosh Computers.

It is interesting to note that various versions of Microsoft Windows since Windows 95 natively support MPEG1 playback capability(have MPEG1 software decoder built-in), but have never had any support of MPEG1 Encoder. Only the Windows Millennium (WindowsMe) has a built-in MPEG4 Codec.

(7) Dedicated Hardware Decoder Products

MPEG video needs decoding software to playback on computers, it also needs decoding hardware to playback on external Television screens, because normal PCs do not always have a TV output socket, also because video displayed on TV gives a much better feeling to human eyes than video displayed on PC’s monitor: colour is more vivid, brighter, smoother movement, etc. MPEG1 video decoder cards have been very popular when PC’s CPU speed was around 100MHz and PC’s display cards were 1~2 Mbytes VGA. MPEG2 video decoder cards are still useful for PCs with around 500~1000 MHz CPUs and 32Mbytes AGP display cards. Probably when HDTV quality video capture becomes available on PCs, high definition MPEG2 video decoder card will be needed for decoding those videos because 1 or 2 GHz PCs won’t be fast enough to smoothly display 1960X1280 pixel live video in full colour at 24 Frames/sec.

Generic video display cards(AGP cards) with TV output sockets exist, such as Matrox G450 and many TNT2 and Gforce chipset based AGP cards: they can output full PC’s screen to external TV, as well as output a specific logical window’s content onto the external TV. But in general their output of decoded MPEG video quality would not be as good as a dedicated MPEG decoder card, in particular when PC’s performance is not high. Dedicated MPEG2 decoder cards(they also decode MPEG1 video) like Hollywood Plus from SigmaDesign can play back excellent quality DVD video on even PentiumII 233 MHz CPU PCs with 64Mbyte memory.

(8)Closely Related to Video Movie Disk Production

All video movie disks use MPEG video as raw video clips: VCD’s video comes from MPEG1 encoding, SVCD and DVD’s video come from MPEG2 encoding. These movie video disks are primarily for entertainment purposes to be played on set-top player boxes not directly related to PCs, but their production process is mostly through PC systems. Using MPEG video capture cards or encoding software, some appropriate authoring software and CD writer or DVD writer, desktop PCs can be used to produce VCD/SVCD/DVD movie disks to play on VCD/DVD players.

(top)

5.2 Problems with MPEG Video Capture

(1) Difficult to Edit

The inter-frame encoding(video frames’ content relate to their previous or subsequent frames) nature of MPEG encoding makes their video files difficult to edit: cutting and pasting arbitrarily at random points will disturb the originally defined GOP structure, and incorporating titles/graphics and audios into MPEG video files is very slow because total re-encoding of the whole video file is required. Even jumping from one point to another within the video file is slow, making the editing process less Non-Linear than many other video capture cards designed for non-linear editing. MPEG video editing software do exist: some are stand-alone, some are embedded into generic video editing software like Adobe Premiere and ULEAD VideoStudio and MediaStudio. Some MPEG encoding software even claim to be able to do frame-accurate editing, but in reality, arbitrary random editing will inevitably destroy the original MPEG video file’s structure therefore making the un-edited part of the video in the resulting video file different from the original video file. This is mainly because edited MPEG video needs to be re-encoded at the finishing stage, but this re-encoding process is using a different source video: the MPEG video that has been “edited”. The result of MPEG encoding on a totally different video source will definitely generate a differently structured MPEG video file, even using the same MPEG encoding software, let alone to mention that many MPEG files being edited have been created by totally different MPEG encoding software or hardware.

Editing MPEG video file also tends to cause video-audio synchronisation problem(or make it worse). Only at very basic home/amateur editing level or as a last resort to modify video files that can never be re-captured, MPEG editing software need to be used.

One of the important purposes for editing video is for outputting the edited video to TV or video tapes. But for MPEG1 video capture cards, even though many of them have video or video+audio output sockets, their video output quality is questionable: with maximum frame size 352X288(PAL) or 352X240(NTSC), data rate maximum 5~6Mbps, the output video quality is far lower than the low-end M-JPEG video capture cards or DV video capture cards, their loss of quality from the original video input makes MPEG1 video capture card really un-suitable for editing to output back to video tapes.

(2) Audio-Video Synchronisation

Audio not synchronised with video is a typical problem in MPEG video capture and encoding process. Normal symptom is the audio lapse behind video, as the video progress, the time gap between video and audio widens: therefore funny result might happen where one person’s spoken words will come out of another person’s mouth. The complexity of Inter-Frame encoding in realtime could be the main reason causing these out-of-sync problem. Faster and dedicated PC systems, better designed realtime MPEG capture cards, or even a better MPEG decoding hardware can alleviate this problem. Using decent MPEG software encoder on high-performing PCs to convert high-quality video files into MPEG files normally also reduces video-audio out-of-sync possibilities.

Some MPEG capture cards have an Audio Post-Processing feature that once enabled, can cause audio portion to be post-processed after video capture process finished, making the video and audio better synchronized. There are also stand-alone audio post-processing software that even re-synchronise the video-audio in a MPEG video file.

Audio-Video out-of-sync problem can also happen on an entirely software encoded MPEG video: converting non-MPEG video files to MPEG IBP-Frame video files can cause audio drifting from video! In particular when the original video file contains audio with sampling frequency different from the resulting MPEG video file's audio sampling frequency. For example, using software MPEG encoder to convert a Matrox RT2000's DV format video files that can only have 48KHz audio sampling rate into VCD format that has to be 44.1KHz audio sampling rate, often the resulting MPEG1 video file has audio problem, and the problem might be more than audio-video out of sync, it could also be no audio at all at some stage of the video playback.

(3) Pixelation Effects and Garbage Frames

When a video clip contains fast object movement or sudden scene changes, or when the video content is difficult to be compressed such as

artificially created animation or slow motion etc, the captured MPEG video files tends to have obvious pixelation effects: the frames show

multiple areas of rectangular blocks here and there which seems to be totally un-related to the video contents. Sometimes a severely pixelated frame looks like a totally garbled image with blocky coloured areas spread around many places to presenting a totally meaningless picture.

(4) Zigzagged Edges

Similar to M-JPEG and DV video but more obviously, captured MPEG video can introduce artificial distortions along naturally straight edges like eaves, telephone poles, window frames, door edges, etc. The originally straight edges will become zigzagged. These distortions are more obvious when MPEG video is displayed on PC's screen than on external analogue TV screen. Zigzag in edges are due to the very nature of the Discreet Cosine Transform, a mathematical formula used in MPEG as well as in M-JPEG and DV video compression, to convert image pixel’s random colour values into a kind of repetition frequency values.

(5) Jumpy Effects at Object Movement

When video clips contain continuous movement like sports activities, moving ships or vehicles, the objects seems to move not smoothly: people or vehicles stagger and there are multiple shadows follow their movement. This problem can happen even on pure software MPEG encoder converting already captured high-quality video files into MPEG files. For example, the Pinnacle DV500 has its own MPEG2 IBP encoder built-in as a Adobe Premiere plug-in but it tends to produce such a jumpy effects MPEG2 video whenever there is movement in the original DV video captured from DV500. This problem is also obvious -- more obvious than M-JPEG and DV capture cards -- when the original video has the camera’s lens moving across the objects like panning horizontally or zooming in or out.

(6) Noisy Top and Bottom Stripes

These are blurry video pixel lines appearing at the top or bottom of MPEG video files, they could happen on both realtime captured and pure software encoded MPEG video files. These noisy stripes are most obvious when playing back on PC’s screen. Most of the times even though they show up on PC’s screen, they will not appear when displayed back on a Television, either through a hardware MPEG decoder card, or through VCD/DVD players, due to the fact that TVs do not display some top and bottom lines of the video.

(7) Movie Disk Standard Compliance Is Not Always Achieved

When VCD/SVCD/DVD compliant MPEG video capture is required, many otherwise perfect MPEG video become in-compatible with the required movie standards. Typical symptoms include being rejected by VCD/SVCD/DVD authoring software, or created movie disks do not play on set-top VCD/DVD players, although most of them play OK on PCs. The VCD/SVCD/DVD standards require movie disks to contain a subset of MPEG video that is strictly defined according to some standard formats, while not all MPEG video capture cards and encoding software follow these standard formats properly. The DVD authoring software are much more restricted in picking up input MPEG video files than VCD and SVCD authoring software. Very often a particular MPEG2 capture card can capture video only suitable for a certain DVD authoring software. In this aspect even MPEG1 video capture cards have some similar problems. Many MPEG capture cards have pre-defined capture templates to provide fixed parameters for capturing VCD/SVCD/DVD compliant MPEG video files, but still very often their pre-defined templates do not match the requirement of some movie authoring software.

(8) Realtime PC Screen Video Overlay

Like M-JPEG and DV video capture cards, MPEG video capture cards can display incoming video signal live on PC’s screen in a window. However, in most cases, this live video display is slightly slower than the external video signal, that is, the live video displayed on PC’s screen lags slightly behind the incoming video. MPEG2 capture cards normally have longer delays than MPEG1 capture cards in this aspect. The reason for this delay is due to the fact that, video displayed on PC’s screen is decoded by the MPEG capture cards or by MPEG decoding software on PCs after they have been encoded first, while decoding MPEG video requires much more computing power than displaying M-JPEG and DV video, because MPEG video has been compressed more heavily. Most current PCs will have to spend most of their computing power to serve the encoding first, the decoding and sending decoded video to PC’s screen is handled after video encoding therefore a time delay is inevitable, in particular in slower PCs. On the other hand, for those MPEG capture cards that have video output ports such as Dazzle’s DVCII, video displayed on TV connected to the video output port does not lag behind the incoming video signal, because this video is handled by a hardware video signal pass-through, no encoding/decoding process is involved as with the video displayed on PC’s screen.

(9) Host PC Compatibility and Performance Issues

Hardware Encoding Chipset based MPEG1 video capture cards do not have too many compatibility problems with host PCs' components like mainboard and graphics cards. They do not require high-performing PCs either: the entire signal-conversion and MPEG encoding task is accomplished by the capture card. Normal PentiumMMX 200 or PentiumII233 PC will work quite happily with these MPEG1 capture cards in both realtime video capture and playback. Software Encoding based MPEG1 video capture cards will have more issues with the host PCs’ performance, while MPEG2 realtime video capture cards are much more demanding in these aspects: at least PentiumII or PentiumIII 500MHz PCs with 128MB RAM plus compatible AGP video display cards are required, and the types of mainboard and other plug-in cards in the same PC can also affect the MPEG2 encoding/decoding. AMD CPU-based mainboards often cause much more compatibility problems with realtime MPEG2 video capture cards than Intel CPU-based mainboards.

(10) Problem Solutions

The situations in which the above problems appear vary: by using the same encoding hardware/software, they could happen on some video clips but not others, they could happen on one PC but not another, or the problem could happen on one playback device(e.g. DVD player) but not the other. The un-certainty of these problems makes solution finding difficult. Faster PCs, dedicated operating environment, higher-quality encoding hardware/software might all help. Some MPEG encoding products offer solutions like multi-pass encoding, multiple-encoding templates customised for different video contents, manual insertion of I-Frames, audio post-processing, etc, but in general MPEG video encoding/decoding is not as stable and predictable as M-JPEG and DV video capture and playback.

(top)

5.3 Implementing MPEG Encoding and Decoding

(1) MPEG1 and MPEG2 Encoders

MPEG video standard has many different formats, this results in the many different implementation of MPEG video capture cards and encoding/decoding software. The first and most widely used is of course the MPEG1 capture cards and encoding/decoding software. MPEG1 video capture cards normally have composite and S-Video input, most of them also have stereo audio input. Video/audio output might not be available. Firewire, Components and SDI input are rare and available only on high-end cards. Typical maximum data rates are 3~6 Mbps, while frame sizes are limited by 176X144 and 352X288 for PAL, 176X120 and 352X240 for NTSC. Connection interfaces are mostly PCI, but USB version is also available. Software MPEG1 encoders are more flexible, their availability is even wider, including many very low-price and even freeware on the Internet. Software can come either as stand-alone version or plug-in version. Stand-alone software can take any(in theory) video files on the PC and convert them into MPEG video files. Plug-in version normally needs to be installed into a generic video editing software like Adobe Premiere, then function as one of the movie file export formats. Typical software MPEG1 encoders include software from Panasonic, Ligos, and Xing, and a freeware called TMPG.

Almost all MPEG1 encoders, software or hardware have VCD compliant templates or pre-defined parameter sets, so that when VCD-format MPEG1 video files is required, the output files are guaranteed(at least in theory) to follow VCD’s standard coding parameters, including pre-defined GOP structure, 1150Kbps data rate, 352X288(PAL) or 352X240(NTSC) frame size, and MPEG1 Layer2 44.1KHz Audio, etc. Dedicated MPEG1 hardware decoder cards are now almost non-existent, due to the increased PC processing power and the availability of MPEG2 hardware decoding cards that normally also decode MPEG1 video.

More and more realtime MPEG2 video capture cards are becoming available for the past two years. They normally come in PCI format, although occasionally one or two USB version or even software encoding MPEG2 capture card tried to show up. Typical implementation follow MPEG main-level/main-profile standard, with frame sizes <= 720X576, data rate <= 15Mbps. Some MPEG capture cards capture MPEG1 format video as well. DVD compliant parameters setting are often claimed, but in reality many of them have problems with their encoded video not even being accepted by DVD authoring software. SVCD compliant encoding are available only in a few MPEG2 video capture cards. All MPEG2 video capture cards have audio capture built-in together with video, but video/audio output might not be always available. Video input interface include all possible types, from composite, Svideo, to Firewire, Components and SDI for digital BetaCam cameras. Audio can be encoded in MPEG1, PCM or Dolby AC3, AC5 and audio frequencies include 48KHz and 44.1KHz.

Realtime MPEG2 video capture cards have more compatibility and performance issues than MPEG1 capture cards. Certain mainboards, CPUs, graphics cards will not work with certain MPEG2 capture cards, and the compatibility can only be confirmed with a particular card tested on a particular PC.

MPEG2 software encoders are also showing up in various forms, even generic video editing software like ULEAD VideoStudio/MediaStudio have some version of Ligos MPEG2 encoders built-in. other famous ones include MediaCleaner from Media100, LSX-MPEG from Ligos, etc.

(2) MPEG4 Encoding and Decoding

A new MPEG encoding method called MPEG4 is appearing on a few different platforms, mainly in software encoding/decoding, video streaming and surveillance video capture cards. MPEG4 was initially designed for very low bit rate video encoding application fields, but later in implementation has been focused on so-called object-oriented encoding, meaning that the different objects inside the video frames are to be treated separately using different encoding methods, so that more compression efficiency and interactivity can be achieved. Real MPEG4 encoders available now are mainly free software encoders and some video surveillance capture cards.

Free software MPEG4 encoders come from Microsoft MPEG4 codec(Windows Me has some bundled in), and DIVX codec which originally was adapted from Microsoft’s codec but has since been enhanced greatly by some individuals and venture capital backed start-up companies. The DIVX MPEG4 codec are widely used by people to re-compress high-quality MPEG2 or other video into a much more compact format to fit long hour videos onto single CD disk or make them downloadable from Internet, while keeping a very high quality-over-data rate(or data volume) ratio. While there are very useful technical effort achieved, recent commercial release of low-price DVD recording hardware like Pioneer DVR A03 and corresponding DVD recording media have cast doubts on the economy of using these (to be matured) MPEG4 codecs to replace expensive MPEG2 hardware. Commercial DIVX hardware encoder/decoder cards are not available, partly because the MPEG4 encoding is extremely computation-intensive so realtime capture cards is either too expensive or just not possible to make with the current hardware technology: to re-encode one hour of DVD video onto a decent quality one CD-ROM based video(<= 700 Mbytes) normally takes 10 hours on PentiumIII1GHz PCs.

5.4 Applications of MPEG Video

The most widely found application of MPEG video are in the forms of Video Movie Disk creations like VCD, SVCD and DVDs, distributed for the purposes of entertaining, education, commercial promotion, long-time preservation and archiving, etc. But MPEG videos are also used for Satellite video broadcasting, High-Definition Television Broadcasting and Reception, Internet/Intranet video streaming, multimedia CD creation, surveillance video transmission, and recently some video camera recording. The most unlikely application for MPEG video is video editing.

6. Common Problems for Video Capture Operation (top)

More than any other complicated data processing operation, video capture operation realised on video capture cards inevitably can cause many problems that affect the proper creation of the video data. These problems are due to the different hardware or software designing methods of the video capture cards.

6.1 Un-necessary stop of video capture when incoming video signal fluctuates

During the video capture process, the incoming video signal can often become un-available or accompanied by strong noise, such as when a VHS tape contains signal gap, or a MiniDV tape contains multiple zero timecodes. In such a situation, an ideal operation for the capture card to take would be to temporarily pause the capture process, and wait until the incoming signal to reappear so that it can resume the capture operation. Other ideal operations include allowing the user to set-up different actions before the capture starts, or keep recording regardless the stopping of the incoming signal. However, many video capture cards actually cannot handle this situation very well. MiroDC30+, for example, will start reporting frame dropping once the incoming signal is stopped, and will record corrupted data if the capture is not manually aborted soon. Pinnacle DV500, on the other hand, will report “recording error” when no analogue video signal is present. For the multiple zero timecodes on DV tapes, Canopus EZDV will abruptly stop the capture without giving any warning or choice. For MPEG recording cards, reactions to absence of incoming signal also vary, many of the MPEG capture cards will simply stop capture.

6.2 Delays at the start of video capture process

For capturing analogue video, the moment the “capture” command is issued by the user does not necessarily mean the moment the video is started being captured into PC’s hard disk, a time delay normally exists. Some capture cards such as MiroDC30+ or Vitec RT6 have very little time delay, others such as RT2500 will cause a few seconds delay, therefore unable to capture a few seconds of video immediately after the pressing of “capture” button. For capturing DV video from Firewire interface, some cards like RT2500 or FirebirdDV/XE allow the user to manually pause at a particular time-code frame, then manually start capturing process precisely from that frame, without missing anything, while many other DV capture cards cannot handle such a frame-accurate capture, forcing users to first manually pre-roll the DV camera/VCR then start pressing “capture” button on PC slightly before the desired frame, or to use batch-capture if continuous time-code exists.

6.3 Difficulty to pause and resume video capture

During manual video capture process(user manually clicks “Capture/Record” button from PC software to start video capture), most video capture cards do not allow the user to randomly pause the capture then resume the capture by clicking a button on the software interface. The reasons people want to have such a pause/resume operation include realtime skip-over unwanted segments of the incoming video, changing video tapes inside VCR/camera, switching video sources, etc. Only some capture cards, such as Canopus DVRaptor/EZDV/DVStorm, DVStudio ApolloExpert, Darim MPEGAtor, Array VideoOne etc do allow such a dynamic function. Because this is a hardware/software combined operation, therefore it is not possible for all the video capture cards manufacturers to implement it even if they want to do so --- when they do not implement pause/resume for their capture cards, what you get is once the video capture process starts, all you can do is either to stop the current capture by clicking somewhere on the proprietary software interface, or to wait until the preset capture timer expires (if the capture software allows you to set-up a timer).

6.4 Long-time un-limited video capture

“Un-limited Video Capture” has been touted by many video capture cards makers/sellers loudly almost everywhere, mainly because too many of the cards cannot realise this simple function easily. The reasons a capture card cannot capture video unlimitedly, that is, continuously capture video while there are still free spaces on the PC’s hard disks, involve mainly software implementation limits. The ubiquitous Microsoft operating systems, from the oldest beloved MS DOS to the latest greatest Windows XP, all have some file systems – the way the hard disk is organised – that limit any single data file’s size to be no bigger than 2GigaByte or 4GigaByte. These file systems are called FAT16 and FAT32. The newer file system used by MS Windows NT and above and named NTFS, has the single file size limit extended to many many thousands gigaybytes, therefore exceeded any single physical hard disk’s capacity. Under the FAT16 or FAT32 file systems, naturally many video capture cards can only capture video while the destination file size is still under 2Gbyte or 4Gbytes. Once that limit is reached, the capture process stops. This is why when you are capture videos using all of those low-end cards, or even expensive ones like DC30+, DV500, etc, under Windows 98/Me, or under Windows NT/2000/XP with a hard disk formatted as FAT32, you suddenly see the video capture stops at 9 or 18 or 5 minutes depending on the pre-set data capture rate. However, there are capture cards such as Matrox RT2000/2500, DVICO FirebirdDV/XE/FusionMPEG, Canopus DVStorm/DVRaptorRT, that implemented smart workaround of this dreadful 2G/4B per file size limit: they transparently create multiple 2G or 4G files when video capture process is long enough to fill up a file with the 2G/4G bytes data. When the video capture process ends, these capture cards also present a header file for you to use under various video processing software such as MediaPlayer or Adobe Premiere, so that you get a feeling that you are handling a long video file. This is a smart work around but sometimes it has pitfalls: sometimes the transparently strung-together long-time video files could cause the video processing software to mal-function.

Once you are using the more recent NTFS file system under Windows NT/2000/XP, most of the video capture cards will be able to capture video continuously – some still cannot, such as DC30+ or DC10+(new software released by Pinnacle extended the capture length but not unlimited). However, they normally cannot automatically capture across hard disk boundaries, that is, if one hard disk is full they cannot automatically use another hard disk that has empty space to hold newly captured video data.

Under Windows NTFS file system, ideally “unlimited capture” should be able to capture however long video into one single video file, so the subsequent editing or conversion will be simple. But in fact some video capture cards like Canopus series cards(DVSTorm/DVRaptor/EZDV) cannot do this: they still insist the “1 header file + 1 or more data file scheme” used on the FAT32 file system, that causes unnecessary overhead and sometimes instability of the captured video files.

Not all failure to capture unlimitedly is due to MS Windows’ file size limit. Some capture cards, such as Dazzle’s DVCII, stops video capture process at some time point (55 min. for Dazzle DVCII) without any reason due to their internal bugs, regardless which file system the hard disk is formatted as. Another example is DC30+/DC50 etc Miro cards: they cannot capture unlimitedly(limited by 4 or 2 Gbyte per file) even on NTFS file systems due to the design of their special video capture software module.

6.5 Difficulty to Display High-Quality Video on Host PC Screen

6.6 Compatibility with Host-Computer Hardware

6.7 Compatibility with Video Editing Software

7. Audio Capture

Closing Words (top)

There are other video compression methods such as Wavelet and Fractal encodings, that are still in use although in smaller scales. The reason that we discussed Video Capture Cards based on the major data compression methods they used is, with few exceptions, the application fields and main features of a video capture cards are largely dependent on how the video is compressed (or not compressed) by the card. This can also explain why there will not be so-called all in one video capture card that can do everything well: video editing, encoding, transmission, surveillance, etc, since most hardware-Codec based video capture cards can be compressed only in one format. Even when multiple compression formats hardware chipsets were used, implementation difficulties and engineering resource allocations normally constraint a video capture card to be only good for its major application fields.

Video capture cards can certainly be discussed from many different perspectives. For example, based on how a video capture card is used in its major application fields, we might also group video capture cards into Video Editing Cards such as M-JPEG and DV capture cards, Video Encoding Cards such as MPEG cards, and Video Transmission Cards such as video streaming cards and video surveillance cards. These different discussion and analysis might well constitute another article about video capture cards. And further more, individual application fields for video capture cards, such as details to create VCD/DVD video disks starting from using video capture cards might also form the basis of yet another article. Finally, the detailed comparison and analysis of typical video capture cards can also be organised into some future articles. Unlike many other peripherals for PCs, successful Video Capture Cards enjoy relatively longer life-span on the market, in terms of years instead of months or weeks as the other parts of the PCs. This certainly produces more chances for objective and serious testing and analysis on video capture cards. As digital video processing and generic computing merge and mix further on personal computers, discussion and analysis on video capture process – the first step to bring over this merging and mixing will certainly generate more interests among end-users, resellers, and manufacturers.

(top)

Bibliography and Links

Send e-mail to Inventa Australia