Windows Video Capture Cards
`
`By Bob Doyle
`
`C users who need to cap-
`ture video for corporate
`presentations, CD-ROM
`games, kiosks or educational appli-
`cations used to be at a decided dis-
`advantage compared to their Mac
`peers. Just one year ago we tested
`14 Video for Windows capture
`cards, and only half could even
`achieve 8-bit color,
`160-by-120-pixelcapture at 15 frames per sec-
`ond. (Intel's Smart Video Recorder
`was a notable exception.) Further-
`more, no editing software was then
`available for the PC. Adobe's Pre-
`miere was at version 2.0 on the
`Mac, but only just about to appear
`for Windows.
`Today, there are more than 40
`PC capture cards that do better
`than last year's standard, and at
`least 20 include hardware compres-
`sion chips enabling full-color (24-
`bit) capture, many at full -motion
`(defined as 30fps) and some at full-
`screen (640 by 480). Several new
`editing packages are also available,
`including Premiere 4.0 for Win-
`dows, the undisputed champ.
`For this review we looked for
`boards that were supposed to be
`able to capture and compress full-
`color (24-bit), full-motion (30fps),
`quarter-screen (320 by 240) video
`in real time on Level 2 multimedia
`PCs. This size is appropriate for
`computer screen presentations,
`and the data rates are small enough
`to play back from CD-ROMs and
`over video server networks. Thirty
`frames per second can also elimi-
`nate lip-sync problems in the "talk-
`ing head" shots so common in
`multimedia presentations. While
`
`we started out with 22 cards, only
`11 met this standard and nine
`are top-rated. Eighteen are
`reviewed here.
`Full-motion video capture
`should not be confused with still-
`frame capture. All these cards can
`capture very high-quality 640-by-
`480 still frames for use in image
`processing.
`Prices for the top-rated
`boards range from $500 to
`$3,000. We do not include two
`boards from Matrox and Truevi-
`sion in the $6,000 range that are
`capable of 60-field, full-screen
`capture, the standard for broad-
`
`cast-quality nonlinear video edit-
`ing. (We reviewed these cards in
`the August issue.)
`Surprisingly, many of the
`more expensive cards do not pro-
`vide better performance. These are
`generally cards of older design,
`some taking up two full-length ISA
`slots and originally intended for
`high-quality JPEG still-frame
`capture and video-in-a-window
`overlay. M-JPEG video capture
`capability was added to these cards
`to take advantage of the emerging
`desktop video market, and capture
`performance is generally poor.
`Some older cards also use the
`
`Chips & Technologies PC Video
`chip, which requires less than
`16MB of RAM, a severe limitation
`for work in Windows. Newer
`cards are usually 3/4- or 1/2-length
`ISA and PCI cards that exploit
`compression chips from LSI
`Logic and Zoran optimized for
`320-by-240 capture or integrate the
`C-Cube CL-550 chip onto much
`smaller cards or daughtercard
`options. Despite their small size
`and lower prices, many newer
`cards provide video overlay func-
`tions, excellent 640-by-480 still-
`frame capture and even NTSC
`video encoding.
`
`Testing Platform: To test Windows video capture
`cards we used a state-of-the-art Dell 90MHz Pentium
`with PCI and ISA buses as our principal test machine
`along with two similarly configured clones. We
`chose the P90 because it has enough horsepower to
`play back software-only codec files like Indeo and
`CinePak at their top rates. Each machine had 16MB
`of RAM, an internal 440MB IDE hard
`drive, an Adaptec AHA 1542CF SCSI
`adapter/controller and a 1.7GB Microp-
`olis Microdisk AV-ready removable
`hard drive. Creative Labs Sound Blaster
`16 cards were installed with double-
`speed CD-ROM players. We also used
`a 486DX2/66 with 8MB of RAM to test
`the cards with PC Video chips, which
`require less than 16MB of memory to
`be present.
`Our standard software included DOS 6.2.1,
`Windows for Workgroups 3.11, Video for Windows
`1.1a, Adobe Premiere 1.1, Adaptec EZ-SCSI, Sound
`Blaster 16, graphics display card drivers, Norton
`
`Utilities and a number of hard disk speed test and
`video playback test utilities. We loaded standard
`config.sys and other boot files with the baseline sys-
`tem. The Adaptec BIOS settings were set to the
`same shadow RAM and DMA transfer rates, and the
`motherboards' BIOS used the same shadow RAM for
`graphics display. We restored the entire boot disk
`
`before installing each new card for testingandSCI.
`permanent 50MB Windows swap file on the boot
`drive. We then installed each card according to the
`manufacturer's directions. All cards had installation
`(continued on page 78)
`
`77 NEWMEDIA NOVEMBER 1994
`
`HTC Exhibit 1009
`
`

`

`Feature'.
`Choosing the right capture card has
`become a lot more complicated
`than last year. Major decisions
`include:
`>Which compression algo-
`rithm to use—Motion-JPEG,
`Indeo, DVI, Auravision, MPEG or
`software-only;
`>Which slot and system bus
`type—ISA, EISA, VLB, PCI;
`>Connection to your VGA
`card for video overlay—VESA Fea-
`ture Connector, VESA Advanced
`Feature Connector, VESA Media
`Channel or external passthrough
`connector (see page 82);
`>Whether you want movie
`playback acceleration and scaling
`features—these can reside on your
`capture/overlay card or on your
`graphics display card;
`>Whether you want the card
`to double as a VGA graphics card;
`>Whether you want NTSC
`encoding (output) for videotape
`recording; and
`>Whether you want audio
`capture on the video capture card
`or on a separate sound card.
`Other options include TV
`tuners for video-in-a-window,
`MPEG playback and A/B switchers
`with wipes and dissolves between
`two video input sources.
`
`Quality
`As you might expect, average video
`image quality is much better than last
`year, and the range of quality is nar-
`rower. There is less color variation
`from card to card, and quarter-screen
`images (320 by 240) are the same
`quality as the user interfaces of cur-
`rent high-end nonlinear editing sys-
`tems such as the Avid Media
`Composers. This quality is fine for
`"offline" video editing in which you
`capture with time code, edit in Pre-
`miere and later take the batch list of
`your Premiere project to a higher-
`end system for broadcast-quality 60-
`field capture. (Premiere is the only
`program offering time code support
`and batch recapture on the PC.)
`For multimedia distribution
`on CD-ROMs or on video server
`networks where data rates are lim-
`
`NewMedia Lab Test Procedures
`
`(continued from page 77)
`woes (see "Installing and Tuning Windows Capture
`Cards," page 92).
`Capture Procedures: To provide identical source
`media for each card, and a challenge for their video
`compression chips, we created a four-minute digital
`media diagnostic test on Betacam SP tape with the
`Matrox Studio 2300 nonlinear editing system. Our
`tape sources were top-quality (i.e., low-noise) Beta
`SP materials.
`Each section of the tape was designed to high-
`
`light typical compression problems. One section
`emphasized motion to test interframe compressors
`like Indeo, another section had little motion but com-
`plex scenes and another had still frames. White
`frames and beep tones at one-minute intervals
`tested audio-video synchronization. We recorded
`standard video test signals generated by a Truevision
`VIDI/0 Workbench, and five seconds of a resolution
`test chart were shot with an Ikegami studio camera.
`We also generated random video noise alternating
`with black-and-white frames and a number of wipes
`and DVE moves of checkerboard black/white/-
`color/noise patterns, using Inscriber CG software and
`the Matrox Studio's DVE capability.
`We captured this test tape with each board in
`
`NewMedia Lab's Digital
`Media Test Tape is
`designed to evaluate the
`i mage and motion quality
`achieved by video com-
`pression products. A vari-
`ety of text, animation and
`complex motion video
`sequences are included
`on the tape.
`
`several combinations of compression quality, win-
`dow size and color depth, and determined the maxi-
`mum frame rate that could be obtained with each
`one. So that we could quickly detect dropped frames
`upon playback of captured video, our test tape had
`an animated time code display with a sweeping bar.
`The Micropolis AV drives were defragmented
`with Norton Speed Disk before each capture to
`ensure optimum performance. We then used the
`Video for Windows VidCap utility as the capture soft-
`ware whenever possible. Contiguous sectors could be
`pre-allocated for the video capture file in VidCap or in
`the card's own capture utility. We color-calibrated
`our three 17-inch monitors with a standard SMPTE
`color-bar pattern stored as a disk file and made video
`adjustments for each capture card's video input with
`the SMPTE color bars on our test tape.
`Analysis: Our test
`results varied widely
`in image quality and
`frame rates achieved.
`(For image-quality
`results, see page 84.)
`As far as frame rates,
`only 11 of the 18
`cards we tested met
`our standard of 24-bit
`color, 30fps capture
`of a 320-by-240 video
`window. The other
`cards dropped lots of
`frames—if they
`could capture at this
`level at all.
`We also tested
`how well the cards
`played back a cap-
`tured video file on disk and found that few could play
`back as well as they could capture—a surprising
`result, since M-JPEG compression takes longer than
`decompression (although both tasks are completed in
`real time with appropriate chips). Most would occa-
`sionally skip frames.
`Video for Windows apparently has trouble pro-
`ducing smooth playback. This may be caused by Win-
`dows and DOS software servicing higher-priority
`interrupt handlers and not coming back in time. We
`also found that VidCap's status line was inaccurate in
`some cases. VidCap would report zero frames dropped
`during capture when there were actually missing
`frames. The capture driver, written by the card manu-
`facturer, may not be reporting accurately to VidCap.
`
`7 8 NEWMEDIA NOVEMBER 1994
`
`HTC Exhibit 1009
`
`

`

`Windows Video Capture Cards
`
`Company
`Phone
`
`Product
`
`Alpha
`Systems Lab
`(714) 252-0117
`
`ATI
`(905) 882-2600
`
`Cogent
`Electronics/ADS
`(415) 591-6617
`
`Creative Labs
`(800) 998-1000
`
`Diamond
`Multimedia
`(408) 736.2000
`
`Fast
`Electronic US
`(415) 802-0772
`
`Hauppauge
`Computer Works
`(800) 443.6284
`
`IBM/Intel
`(800) 241-1620
`
`MegaMotion
`
`Video-Itl
`
`Video Wizard
`
`WIN/TV Cinema
`
`ActionMedia II
`
`Video Blaster
`RT300
`
`VideoStar Pro Movie
`Machine Pro
`M-JPEG option
`
`M-JPEG option
`
`Tested with
`
`2MB VRAM
`
`Slot type, size
`
`Minimum system
`
`Memory limit
`
`Configuration needs
`
`VESA/Passthrough connectors
`
`ISA,
`
`full
`
`ISA, 1/2
`
`ISA,
`
`full
`
`ISA, 3/4
`
`ISA, 3/4
`
`2 slots, ISA 3/4
`
`ISA, 1/2
`
`386, 4MB RAM,
`sound card
`o
`
`I/O
`1
`2 IRQ,
`address, 2 memory
`base addresses
`*t•
`
`486/33, 8MB RAM
`
`386/33, 4MB RAM
`
`o
`
`8MB
`
`386DX/33, 4MB RAM,
`Sound Blaster
`o
`
`486SX/25,
`8MB RAM
`0
`
`386/25,
`4MB RAM
`)
`
`386
`
`0
`
`1 1/0 address
`
`o/o
`
`I/O
`I
`IRQ, 2 DMA,
`I
`address,
`I memory address,
`I buffer address
`bib
`
`1
`
`I/O address,
`
`I
`
`IRQ
`
`oh
`
`2 IRQ, 2 DMA,
`2 I/O addresses,
`I memory address
`e/•
`
`I/O address,
`I
`IRQ,
`I
`I memory address
`
`• /•
`
`Capture option
`
`ISA (also MC)
`
`386/25,
`6-8MB RAM
`0
`
`I/O,
`1
`IRQ,
`1
`I memory
`address
`°t•
`
`VIDEO CAPTURE
`Standards
`
`Connectors
`
`Sampling
`
`NTSC, PAL
`
`1 S-video
`
`YUV 4:2:2
`
`Selectable
`
`YUV 4:2:2
`
`Compression chip
`
`C-Cube CL-55O
`
`Intel 8275OPD
`
`C-Cube CL-550
`
`Compression scheme
`
`Dynamic•Q control
`
`Video controls
`
`M-JPEG
`
`o
`
`Index
`o
`
`M-JPEG
`
`o
`
`2I/0 addresses,
`2 IRQ, 1 memory
`address
`• i•
`
`NTSC, PAL
`
`NTSC, PAL
`
`NTSC, PAL
`
`NTSC, PAL
`
`NTH, PAL
`
`NTSC, PAL
`
`2 RCA,
`
`1 S-video
`
`3 RCA, 3 audio
`
`3 RCA,
`
`1 S-video
`
`2 RCA,
`
`1 S-video
`
`2 RCA in,
`
`I out
`
`2 RCA
`
`NTSC, PAL
`
`Splitter cable
`
`YUV
`
`Intel
`
`i750
`
`DVI
`
`YUV 4:2:2
`
`Intel
`
`i750
`
`lndeo
`
`o
`
`YUV 4:1:1
`
`YUV 4:1:1
`
`Zoran/Auravision
`
`M-JPEG
`
`3
`
`loran
`
`M-/PEG
`
`'D
`
`YUV 2:1:1
`
`Auravision
`
`Auravision
`
`o
`
`Brightness, Contrast, Hue,
`Saturation
`
`Brightness, Contrast,
`Hue, Saturation
`
`Brightness, Contrast,
`Color, Tint
`
`Brightness, Hue, Red, Green,
`Blue
`
`Brightness, Contrast,
`Tint, Saturation
`
`Brightness, Contrast,
`Hue, Saturation
`
`Brightness, Contrast,
`Hue, Saturation
`
`Capture to AVI file
`
`VIDEO PLAYBACK
`Live overlay
`
`Onboard VGA
`
`Analog output
`
`Video acceleration
`
`Audio capture
`
`Other options
`
`o
`
`•
`o
`
`•
`
`o
`
`•
`o
`
`•
`
`0
`
`o
`
`0
`
`o
`
`•
`
`o
`
`Optional
`•
`
`0
`
`o
`
`o
`
`Premiere
`
`•
`
`o
`
`o
`
`o
`
`o
`
`0
`
`o
`
`•
`
`•
`o
`
`Optional
`
`•
`o
`
`o
`
`•
`
`•
`
`o
`
`•
`
`0
`o
`
`•
`
`•
`o
`
`0
`
`•
`0
`
`TV tuner
`
`TV tuner
`
`o
`
`Brightness,
`Contrast,
`Saturation, Tint
`
`DVI only
`
`•
`
`0
`
`o
`
`o
`o
`
`o
`
`BUNDLED SOFTWARE
`Editing
`
`Video for Windows
`
`•
`
`MediaMerge
`
`•
`
`Other
`
`ASL Utilities
`
`Media Capture
`
`Uninstall feature
`
`Warranty (years)
`
`Price (w/test options)
`
`Reader Service No.
`
`• = Yes, o = No or none.
`
`o
`
`2
`
`$1,095
`
`580
`
`•
`
`5
`
`$499
`
`599
`
`o
`
`0
`
`0
`
`I
`
`$1,950
`
`582
`
`8 0 NEWMEDIA NOVEMBER 1994
`
`Premiere
`
`•
`
`Compel PE
`
`.D
`
`1
`
`$499.95
`
`583
`
`Premiere
`
`•
`
`0
`
`D
`
`5
`
`$529
`
`584
`
`•
`
`Premiere option $SS
`
`o
`
`1
`
`$940
`
`585
`
`0
`
`o
`
`o
`
`o
`
`I
`
`$349
`
`586
`
`3
`
`a
`
`o
`
`I
`
`$995
`
`S87
`
`HTC Exhibit 1009
`
`

`

`Intel
`(800) 538-3373
`
`Matrox
`(800) 810-2550
`
`Miro Computer
`Products
`(800) 249-6476
`
`Omnicomp
`Graphics
`(713) 464.2990
`
`Orchid
`Technology
`(800) 767-2443
`
`Triumph Logistic
`Computers
`(818) 858-S700
`
`VIC Hi-Tech
`(310) 643-S193
`
`VideoLogic
`(800)S78.5644
`
`Xing
`Technology
`(800) 294-6448
`
`In-Motion
`echnologies
`415) 968-6363
`
`icture
`erfect Pro
`igital
`:R option
`
`A, 2 slots
`
`Smart Video
`Recorder Pro
`
`ISA
`
`Marvel II
`
`Studio %Press
`M-JPEG card
`
`PCI
`
`miroVIDEO
`DC1 tv
`
`M&M Pro
`
`11 -JPEG option
`
`Vidiola
`Premium Pro/D
`M-JPEG option
`
`Triumphony
`Visual Forge
`V-20 8 V-30 options
`
`ISA,
`
`full
`
`ISA,
`
`full
`
`ISA, full
`
`I
`
`ISA,
`
`I
`
`PCI slot
`
`86, 4MB RAM
`
`386/25, 4MB
`
`386, 8MB RAM
`
`486, 8MB RAM
`
`D
`
`0
`
`I
`
`IRQ, I I/O address
`
`I memory address
`
`I I/O address,
`memory address
`
`I
`
`386
`
`8MB
`
`386, 4MB RAM
`
`486, 4MB RAM
`
`3
`
`I
`
`0
`
`Video Packer
`
`DVA-4000
`
`Xingltl
`
`ISA,
`
`full
`
`386/33,
`4MB RAM
`8m
`
`MediaSpace
`M-JPEG card
`ISA, 2 slots
`
`386
`
`o
`
`ISA, full
`
`386, 4MB RAM
`
`3
`
`IRQ,
`I/O address
`
`)
` IRQ, I I/O
`idress, I
`emory address
`
`I.•
`
`SSC, PAL
`
`RCA
`
`16 4:1:1
`
`Tan
`
`o/o
`
`0/0
`
`o/3
`
`If SVGA not installed/3
`
`• /•
`
`•/•
`
`3/3
`
`• /•
`
`o/0
`
`I I/O address, I DMA,
`IRQ
`
`I IRQ,
`2 I/O addresses,
`I memory address
`
`2 IRQ, I I/O address, I
`DMA
`
`2 IRQ, video
`DMA, audio DMA,
`memory address
`
`2 IRQ, 2 memory
`addresses
`
`I
`I
`
`NTSC, PAL
`
`NTSC, PAL, SECAM
`
`NTSC, PAL SECAM
`
`RISC, PAL
`I RCA, I S-video
`
`YUV 4:2:2
`
`Intel i150
`
`Splitter options
`
`I RCA,
`
`I S-video
`
`YUV 4:2:2
`
`C-Cube CL-450
`
`YUV 4:2:2
`
`LSI 64702
`
`M-JPEG
`
`I RCA, 3 S-video in,
`I S-video out
`YUV 4:2:2
`
`NTSC, PAL
`I RCA, I S-video
`
`NTSC, PAL SECAM
`
`I RCA, 1 S-video
`
`RISC, PAL
`
`I RCA, audio
`
`YUV 4:2:2
`
`YUV 4:2:2
`
`YUV 4:2:2
`
`NTSC, PAL
`
`2 RCA
`
`YUV 4:1:1
`
`C-Cube CL-550
`
`C-Cube CL-550
`
`M-JPEG
`
`M-JPEG
`
`a
`
`o
`
`C-Cube CL-55O
`
`C-Cube CL-550
`
`M-/PEG
`
`M-JPEG
`
`NTSC, PAL SECAM
`I RCA
`
`4:2:0
`
`DSP
`
`M PEG
`
`PEG I ndeo
`o
`
`ightness,
`mtrast, Hue
`
`Brightness, Contrast,
`Saturation, Tint
`
`•
`
`0
`
`o
`
`a
`
`,)
`
`M-JPEG
`
`o
`
`•
`
`3
`
`D
`
`o
`
`o
`
`o
`
`o
`
`Brightness, Hue,
`Contrast, Saturation,
`Sharpness
`•
`
`•
`•
`•
`•
`
`Brightness, Contrast,
`Hue, Saturation
`
`•
`•
`•
`•
`
`miroMov e Pro
`
`Brightness, Contrast, Hue,
`Saturation, Red, Green,
`Blue
`•
`
`•
`•
`
`Optional SVGA
`
`o
`
`•
`
`0
`
`Brightness, Contrast,
`Hue, Saturation
`
`Brightness, Contrast, Hue,
`Saturation
`
`Brightness, Contrast,
`Hue
`
`Brightness, Saturation
`
`Brightness, Hue,
`Contrast, Saturation,
`Sharpness
`
`•
`•
`•
`•
`
`3
`
`o
`
`Orchid Wave
`
`•
`
`•
`•
`
`0
`
`o
`
`o
`
`o
`
`o
`
`•
`
`J
`
`D
`
`D
`
`0
`
`•
`
`•
`•
`
`o
`
`0
`
`•
`
`Converts to AVI
`
`•
`
`o
`
`o
`
`o
`
`o
`
`o
`
`Her memory
`grade
`
`nel Multimedia,
`dia Capture
`
`Digital Video Producer
`•
`
`Compel, Gatekeeper
`TriDigital CD, MediaBlitz
`•
`
`I
`
`8570
`
`590
`
`99
`
`9
`
`Premiere
`
`•
`
`o
`
`I
`
`$195
`
`591
`
`Ulead V-Studio
`•
`
`Premiere
`option $100
`)
`
`2
`
`$899
`
`592
`
`0
`
`o
`
`0
`
`0
`
`I
`
`$995
`
`593
`
`Premiere
`•
`
`Asymetrix Compel,
`PhotoMorph
`o
`
`4
`
`$899
`
`594
`
`0
`
`o
`
`iPhoto Plus
`
`o
`
`2
`
`$540
`
`595
`
`Win/DOS API $99
`
`0
`
`-,
`
`D
`
`D
`
`I
`
`$1,995
`
`596
`
`Premiere
`
`Xinglt
`
`•
`
`3
`
`o
`
`I
`
`$2,995
`
`597
`
`Xing CD
`
`o
`
`2
`
`$499
`
`598
`
`81 NEWMED TA NOVEMBER 1994
`
`HTC Exhibit 1009
`
`

`

`NO OVERLAY
`
`Do You Need Video Overlay?
`V ideo overlay is very useful for accurate video capture, but many
`
`cards don't include it. Often called video-in-a-window, this fea-
`ture lets you see live video input on-screen without burdening
`your CPU. The live display lets you make real-time decisions
`about when to start and stop capture operations while
`watching the overlay image.
`No Overlay (tog►: Without overlay, you are
`forced to work blind (which is possible if
`you use time code), use a separate
`NTSC monitor or use the Preview
`function in Video for Windows'
`VidCap, Premiere and other edit-
`ing packages.
`With Preview, the capture card
`sends the decompressed video image
`over the system bus to the graphics dis-
`play card with two unfortunate results: The
`pictures are poorer quality and much slower in frame
`rate, and the CPU cycles spent moving your uncompressed
`preview images may reduce your capture frame rate. Future
`PCI bus cards should be able to accomplish Preview without
`these penalties, but all ISA cards suffer from them.
`A better solution is to turn Preview off and
`use a separate NTSC monitor to watch
`the video input. You can either
`loop the video through the mon-
`itor or use a VCR/camcorder
`source with two outputs.
`Video Overlay (bottom
`left►: Most video overlay
`cards take the VGA signal
`from your graphics dis-
`play card via the VESA
`Feature Connector (a rib-
`bon cable to your overlay
`card), put it in a RAM
`frame buffer on the overlay
`card, replace a part of the Win-
`dows screen image with the video
`image and output the combined signal to the
`computer monitor. Your capture card frame buffer therefore should have the
`same amount of RAM as your VGA card for peak performance. Some cards
`switch the signal through an external passthrough cable when not doing video
`overlay, which eases performance problems. Overlay is much faster than pre-
`view since compressed video moves over the system bus.
`Video Inlay (bottom right►: A more efficient overlay method used in newer
`cards sends the video image, usually limited to 640 by 480, from your video
`
`card to the graphics display card for "inlay" there. Only one large frame buffer
`is needed. A new board connector, the VESA Advanced Feature Connector
`(VAFC), has been developed for this purpose. It is 32 bits wide instead of 16.
`VESA Media Channel is a multi-card version of VAFC that uses over-the-top
`ribbon cables to connect cards. Inlay is as fast as overlay and less expensive,
`and any video acceleration circuitry on the graphics card can be used to scale
`up video and perform color space conversion.
`
`a
`
`ited, you can recompress 320-by-
`240 program material with Super-
`Mac's CinePak or Intel's new Indeo
`3.2 software compressors, both of
`which come with most of these
`cards. Quarter-screen M-JPEG
`images should compress nicely on
`the new low-cost, real-time MPEG
`compression engines, since MPEG
`and M-JPEG are related compres-
`sion methods, but be on the look-
`out for unusual artifacts from
`cascading Indeo's vector quantiza-
`tion and MPEG's discrete cosine
`
`transform quantization.
`Disappointingly, QuickTime
`for Windows 2.0 drivers were not
`provided with any of these cards. A
`utility called SmartCap, available
`from Intel, allows some of these
`cards to capture directly into
`QuickTime for Windows format,
`using Indeo coders.
`The top nine cards in our
`review were all able to capture full-
`motion (30fps), quarter-screen
`(320-by-240) AVI video with good
`color and image quality.
`
`MegaMotion
`ASL's MegaMotion
`($1,095 including
`Premiere) is one of
`the most unusual
`DOES THE JOB
`cards we tested. It's aimed at
`sophisticated kiosk and teleconfer-
`encing applications in which devel-
`opers want to display up to four
`resizable, repositionable videos
`simultaneously, along with a num-
`ber of still images. One video
`stream can be compressed video
`
`from the hard disk. The others
`must be live video inputs, which
`requires a $70 auxiliary video input
`adapter that takes up a slot.
`A serious drawback is that the
`simultaneous videos are limited to a
`combined total of 30fps. Three
`videos would each play at only
`10fps. MegaMotion offers some
`transitions between source videos
`and special effects, but the choices
`are limited and the controls awk-
`ward. The Fast Movie Machine Pro
`is better if you need A/B roll for
`
`82 NEWMEDIA NOVEMBER 1994
`
`HTC Exhibit 1009
`
`

`

`videotape production. MegaMo-
`tion's NTSC output is limited to
`straight playback of AVI files.
`MegaMotion doubles as a
`modest-quality VGA display card.
`Accelerated Windows graphics and
`video overlay are both done on this
`card, but it does not include movie
`acceleration and is not DCI-com-
`pliant (see "Video Compression
`Technology," page 88). If you need
`good VGA capability along with
`video capture, the Matrox Marvel II
`is better and cheaper.
`On playback, MegaMotion
`skipped about 10 percent of
`frames, making it jerky and unat-
`tractive. While we had trouble with
`skipped frames on many cards,
`usually we could eliminate the
`problem after optimizing our sys-
`tems. Only the Orchid (15 percent
`drop rate) and Indeo (40-50 per-
`cent) cards were worse.
`MegaMotion's inputs and
`output are S-video only; compos-
`ite adapter cables are supplied.
`We missed having simultaneous
`composite output for monitoring
`and S-video output for recording.
`The color-bar generator function
`was useful.
`MegaMotion is a very com-
`plex and technical card. Unless you
`really need its powerful features,
`stick with a simpler card for straight
`video capturing.
`
`The Video Blaster
`RT3 0 0 ($499.95
`including Premiere)
`moves Creative Labs
`from the bottom level of perfor-
`mance last year to near the top
`today. The board uses the same
`advanced Intel i750 PE compres-
`sion chip found in the Smart Video
`Recorder Pro. It should be capable
`of the same 320-by-240 perfor-
`mance at 30fps, but we were only
`able to look at a prerelease version
`of this card.
`Like the other Indeo cards
`(from ATI and Intel), the RT300 is
`not an overlay card. If you want to
`watch your incoming video care-
`
`fully during the capture process,
`you need to hook up an NTSC
`monitor to an extra output on your
`camcorder. The Preview function
`in VidCap and Premiere can dis-
`play video as you capture, but when
`the ISA bus is busy it stops updating
`the Preview display.
`At the highest quality settings,
`images from the RT300 (and the
`Intel card) are surprisingly good,
`even comparable to M-JPEG. The
`data rates are actually somewhat
`higher than the M-JPEG cards,
`because in real-time capture the
`Indeo algorithm consists entirely of
`key frames. This is fme for offline
`editing, but you will have to recom-
`press to get down to CD-ROM data
`rates. You can set every fourth
`frame, for example, to be a key
`frame, and Indeo will store only dif-
`ference information for the inter-
`mediate "delta" frames. This is a
`very time-consuming process,
`almost as slow as CinePak recom-
`pression.
`At these data rates, the Indeo
`cards are operating at the limit of
`the ISA bus, so when the time
`comes for playback the added over-
`head of the Windows graphics dis-
`play makes them skip lots of
`frames. However, the frames have
`all been captured, so you can edit
`accurately. Playback relief should
`come with video-accelerated dis-
`play cards and the new Windows
`Display Control Interface.
`
`Diamond's Video-
`Star Pro ($529
`including Premiere)
`is one of several
`cards using the Auravision chipset
`(see "Video Compression Technol-
`ogy," page 88), so it has very
`smooth scaling of video playback
`up to full-screen size with no reduc-
`tion in the frame rate. We could
`expand clips to full-screen within
`Media Player and Premiere. The
`Pro option is a daughtercard with
`the Zoran 320-by-240 M-JPEG
`compression chip. VideoStar Pro
`uses a VESA feature connector and
`
`84 NEWMEDIA NOVEMBER 1994
`
`AVI Playback Performance*
`
`% of Frames Played 0
`
`20
`
`40
`
`60
`
`80
`
`100
`
`The top nine cards captured 30fps, 320-by-240 video without dropping
`any frames, but many did not perform as well on playback. The Video
`Blaster RT300 and Smart Video Recorder Pro rely on software decom-
`pression and only play back 60 percent or less of the available frames.
`By contrast, the M-JPEG cards, which have hardware compression and
`decompression, all rated 85 percent or better. Thus, there is a tradeoff
`between the ability to play a file on any machine (software decompres-
`sion) and maximum frame rate (hardware decompression).
`*Playback of 24-bit color, 320x240, 30fps AVI file.
`
`Video Image Quality vs. File Size
`
`miroVIDEO DC-1 tv
`Movie Machine Pro
`DVA-4000/MediaSpace
`Video Blaster RT300
`Smart Video Recorder Pro
`MegaMotion
`Vidiola Premium
`VideoStar Pro
`Picture Perfect Pro
`
`Image Quality
`
`Excellent
`Very Good
`Very Good
`Good
`Good
`Fair
`Fair
`Fair
`Poor
`
`File Size
`(KB/frame)
`21
`18
`18
`29
`30
`22
`17
`24
`15
`
`There were noticeable differences in image quality among our nine top
`boards, even those using the same compression chip. The top five
`cards all produce good results, but steer clear of the bottom four.
`While one might expect a direct correlation between image quality
`and file size, the codec also plays a large part. The Indeo codecs are
`not quite as efficient as the M-JPEG codecs. To compare image
`quality, we used the maximum quality setting that would still yield
`320-by-240, 30fps capture.
`
`HTC Exhibit 1009
`
`

`

`"1"1391? 11111111111
`
`Video Image Quality Comparison
`T o assess image quality, we compared several frames of our Diagnostic Test
`
`Tape closely, including frames with animation, text and complex and sim-
`ple video images. All frames were captured at the highest quality possible while
`maintaining a 320-by-240, 30tps frame rate and 24-bit color. Some boards may
`be capable of higher quality at a lesser frame rate, color depth or window size,
`but these frames reflect realistic results at our recommended capture rate for
`multimedia work. A summary of image quality of all the frames we saw, along
`with file sizes, is given on page 84.
`A detail from one of our video test frames, including a graphic text overlay,
`is shown here for the top nine boards in our test. All deliver reasonable quality,
`but some are clearly superior to others. ASL's MegaMotion card delivered clear
`
`numbers and had the best image in the windmill scene, although the MegaMo-
`tion's results on other frames did not match this one, and we judged it in the
`middle of the pack overall. The miroVIDEO and DVA-4000/MediaSpace images
`were more true to form. Both were very good, matching their overall perfor-
`mance. Next came the Movie Machine Pro and Vidiola Premium, which delivered
`good results but with obvious JPEG artifacts. The two new Indeo-based boards,
`the Video Blaster RT300 and Smart Video Recorder Pro, came up with only fair
`images of the windmill scene that were below their average results in other
`frames. Color vectoring is evident. The VideoStar Pro also had fair results with
`bad JPEG artifacts. The Picture Perfect Pro brings up the rear, as it did in our
`other tests, with bad JPEG artifacts.
`
`Movie Machine Pro
`
`Picture Perfect Pro
`
`Smart Video Recorder Pro
`
`an external connector to pass
`through the screen graphics from
`your display card. Three video
`inputs let you choose from two
`composite and one S-video source.
`VideoStar and the other Auravision
`board have a useful cropping func-
`tion that expands the image to
`eliminate noise at the edges of the
`picture.
`When the card was running,
`we were able to capture at com-
`pression ratios up to the highest
`quality setting of 16:1 without
`dropping any frames. VideoStar
`Pro playback was among the three
`smoothest cards tested. All other
`cards skipped at least a few of the
`6,000 frames in the test.
`As expected, picture quality
`was also high. VideoStar Pro filters
`incoming video by comparing suc-
`cessive frames for identical infor-
`mation. However, the Zoran codec
`chip means that making movies in
`Premiere will take longer with
`VideoStar Pro than with boards
`based on the C-Cube or LSI Logic
`M-JPEG chips that can use their
`codec hardware to compress transi-
`tion frames.
`
`4,5
`
`vie Machine Pro
`The Movie Machine
`Pro with the M-
`Leal
`Leal
`JPEG option from
`Fast Electronic
`AWESOME
`($940; $995 with Premiere) is quite
`unusual. In addition to capturing
`AVI files, it functions as a two-
`channel video switcher (like its big
`brother, the Fast Video Machine).
`It has a built-in time-base corrector
`that lets you dissolve, wipe and even
`do digital video effects (DVEs)
`between two video signals (NTSC
`or PAL). Your video sources can be
`analog VTRs, graphics files for titles
`and overlays, digital AVI files, or
`even television signals from a built-
`in cable-ready TV tuner! It only
`lacks machine control.
`The Movie Machine Pro has
`full luminance keying and chroma
`keying of video over video, plus
`color keying of graphics over video.
`It has easy-to-use controls for set-
`ting the key values. If the graphic
`image is a title, you can have it scroll
`(move vertically) or crawl (move
`horizontally).
`You can control the speed of
`
`transition effects by moving a T-bar
`fader with the mouse, pressing the
`up and down arrow keys or just
`pressing a play button. The play
`button icon changes to show the
`effect you've selected. A slider box
`next to the play button lets you set
`the rate of the effect.
`The Movie Machine Pro has
`two buses: a background bus for
`normal video and a foreground bus
`for the incoming effect. On each
`bus you choose from Video A,
`Video B or the TV tuner. For the
`foreground source, you have two
`more options. You can playback an
`AVI file or load one of six graphic
`or title images in a mini still store
`displayed across the top of the
`screen. These can be selected from
`any number of files on your hard
`disk. Movie Machine's video
`encoder outputs recordable NTSC
`or PAL, making it useful for video-
`tape editing.
`While its switching capabilities
`alone are impressive, the Movie
`Machine Pro was also tops as an
`AVI capture and playback card. It
`never skipped a frame on playback.
`Movie Machine Pro does have
`
`some drawbacks. The M-JPEG
`compression option, which enables
`320-by-240 capture at 30fps, is a
`separate card, so it uses up a second
`ISA slot. Movie Machine Pro also
`has only composite video inputs
`and outputs (no S-video). And
`because it uses the Zoran codec
`chip, it will be relatively slow mak-
`ing movies in Adobe Premiere. The
`documentation was also weak; for-
`tunately, installation was not too
`difficult.
`
`Like the Movie
`Machine Pro, the
`In-Motion Picture
`NEEDS WORK Perfect Pro with M-
`JPEG option ($699) is a two-card
`set taking up two ISA slots. One of
`the cards has a daughtercard (the
`"Pro"), and the other, called the
`Digital VCR companion, is the M-
`JPEG option. In-Motion provides
`minimal documentation, but the
`M-JPEG documentation was the
`least we've seen. A 2-by-2-inch piece
`of paper with the IRQ and I/O Base
`jumper settings for the card fell out
`
`86 NEWMEDIA NOVEMBER 1994
`
`HTC Exhibit 1009
`
`

`

`MegaMotion
`
`Video Blaster RT300
`
`VideoStar Pro
`
`miroVIDEO DC1 tv
`
`Vidiola Premium Pro
`
`DVA-4000/MediaSpace
`
`of the box when we opened it. Even
`worse, the default interrupt settings
`for both boards were shown as IRQ
`10, which could really send begin-
`ners around in circles.
`The Digital VCR's compres-
`sion setup dialog box does not have
`a quality setting. Instead, it gives
`three different numbers with their
`typical values—bits per pixel (1.0),
`compression ratio (24:1), and com-
`pressed frame size (10,000). You
`choose one and it calculates the
`others, but there is no indication of
`how they correspond to VidCap's
`quality settings of 1 to 100.
`The Picture Perfect Pro has
`composite video input only. And
`again it uses the Zoran codec chip,
`so it will be slower than C-Cube
`and LSI Logic-based cards render-
`ing transition effects.
`
`the Intel Smart
`Video Recorder was
`our top performer
`a year ago, and the
`AWESOME
`new Pro version ($570) keeps Intel
`in the front ranks. A capture-only
`
`card, it epitomizes the Intel philoso-
`phy of leaving most functions to
`software running on future fast
`buses like PCI. The new Intel Smart
`Video Recorder Pro is an ISA card
`which captures 320-by-240 video at
`30fps with zero dropped
`frames. Since it's not an
`overlay card, you can't
`preview video at high
`frame rates on the com-
`puter screen during cap-
`ture; however, you could
`use a separate NTSC
`monitor.
`We found image
`quality to be quite high for
`real-time Indeo capture
`and compression. The
`data rates were also high.
`At the highest quality set-
`ting, each frame had 25KB
`to 30KB of data, some-
`what more than high-
`quality quarter-screen
`M-JPEG frames, which
`are around 20KB. The
`ISVR Pro can also capture
`"raw" uncompressed
`YUV files (at 320 by 240
`and 15fps; 30fps is too
`
`87 NEWMEDIA NOVEMBER 1994
`
`much for the ISA bus at 80KB per
`frame for 320 by 240), and run
`them through the new Indeo 3.2
`compressor later for improved
`overall quality.
`The playback artifacts of a
`
`vector quantization codec like
`Indeo are quite different from
`intraframe M-JPEG artifacts. The
`most noticeable artifact shows up
`as blocks of color misregistered up
`to three pixels to the right and
`down from a saturated
`color area.
`The ISVR Pro is
`bundled with
`Asymetrix's Digita