What chips are on the card?
Philips SAA 7185B, Digital VIdeo Encoder
Philips SAA 7111A, Enhanced Video Input Processor
Zoran ZR36057 or ZR36067, JPEG PCI Controller
Zoran ZR36060, JPEG Image Compression Processor (Codec) (??)
Advansys ASC3050B, PCI SCSI Controller
National Semiconductor PT80C525, PicoPower PCI-to-PCI Bridge
Socketed 27C512 PLCC EPROM (64k)
What connectors are on the Buz?
The card itself has two back-panel connectors. One SCSI-2 connector (in addition to the internal 50-pin IDC header), and one vga-style connector just for the external Buz Box. Peek up at the picture on this page to see the Buz Box; it explains it pretty well.
The Buz Box is very snazzy, but doesn't have the most practical connector arrangement --- cables come out of three sides. The back side has the thick (well-shielded) A/V cord to the card. The right side has the four "OUT" connectors, and the front side has the same connectors for "IN":
Right Audio
Left Audio
S-Video
Composite Video
Really though, the desktop breakout box is a fantstic improvement to reaching behind the machine to connect things. It's great to have only one thick cable with a sturdy connector go to the back of the card. Sounds simple, but hey, it's neat.
What settings for format, resolution, bitrate, and input should I use?
This is certainly a Frequenty Asked Question, but it doesn't have one answer. The best parameters for your video capture and editing depend on your own equipment and source material. No one set of values works best for everyone. With that in mind, we can try to make some sweeping generalizations, but your best bet is really to play with it; find the settings that you like best.
Consider the output media
If you know what you want to do with your video output from the start of your project, take that into account early. There's no point in digitizing everything at 720x480 if you are going to only produce a low-resolution MPEG file for web distribution. Likewise, if you're planning to produce Hi8 tape, you must use high resolution to retain as much as possible of a good original.
S-Video vs. Composite
If you have the option of using the S-Video (circular connector) input or output, use it! It's capable of much better video than the composite (yellow RCA) input because it keeps the color and brightness signals separate.
Most good Hi8 video cameras and DVD players have S-Video output. Not many VHS VCRs or TV tuners have S-Video, so sometimes you're stuck with composite video.
What format?
MJPEG is almost always the best choice. While RGB does offer the promise of zero compression loss, you pay for it in lower resolution or frame rates. RGB-15 and RGB-24 are really only good for still-image capture, and not motion video. You get more bang for your video byte with MJPEG. If you think you need RGB, you probably just need to decrease the MJPEG compression.
What frame rate?
This depends more on your output than input. If you are going back out to tape, you will almost certainly want to use the standard NTSC (29.97 f/s) or PAL (25 f/s) rates. If your system is having problems keeping up, consider lowering the K/frame (increase compression) rather than the frame rate. It will probably look better in the end.
If your intended output is not video tape, but a digital AVI or MPEG file, you have more flexibility. Most systems can't handle a 30fps MPEG file anyway, so you're better off starting with 15 fps material. Faster than 15fps usually require that the playback system have some kind of hardware assistance. If you are putting your video on the web, you probably want to target the mid-range system and make some tradeoffs.
What resolution?
Simply enough, use the highest resolution your system and disk space can handle. Be sure to select from the three standard Buz resolutions to retain the full features of the hardware codec. For NTSC, these are, of course:
176 x 120 QCIF
352 x 240 CIF
720 x 480 FULL
If you are going to output to an MPEG file for distribution, it's not likely you will need better than 352x240. Very few software-only MPEG decoders can keep up with 720x480, so all your extra effort and bytes will only contribute to poor frame rates and frustrated playback anyway. Most mid-power computer systems will choke on a even a 352x240 MPEG-1 file, so if you want the most web users to be able to view your video, you're stuck with a lousy 176x120.
Some users have reported that 352x240 looks okay for VHS output, but I find significant improvement in going to 720x480. For VHS, consider increasing the compression instead of decreasing the resolution. This will let you do more with the same disk space.
What data rate?
In VideoWave SE+, you're limited to 100k/frame, or 3MB/sec. All other software will let you select 200k/frame, or 6MB/sec. This is not a light load on any machine. To stream a continuous 6MB/sec, you must have some very responsive hard disks and an efficient system. (The 2GB AVI file size limit also restricts you to less than 6 minutes.) It's probably out of reach for your average UDMA IDE system. The good news is that 3MB/sec is usually more than enough for tape output, and you can often use less.
Below 50k/frame, compression artifacts can become noticeable or distracting. How much compression is too much? It really entirely depends on your source material. Certain factors will require that you compress less:
Sharp edges and straight lines will be softer and blurry.
High-contrast edges will look strange and blocky.
Sharp contrasting colors can bleed over the edge.
Fine detail in a still image will be noticeably absent.
Other video characteristics can work to your advantage in hiding the compression effects, and allow you to push the compression loss to lower bandwidth:
Motion
Any motion will make compression artifacts less noticeable. Unlike MPEG, M-JPEG compresses each frame independant of the others, so motion is NOT harder to compress, but easier. If your source material is in constant motion, you can usually get away with high levels of compression that would destroy a still image.
Soft edges and low contrast
These are easier to compress. Facial closeups, out-of-focus stills, and smooth contours (like a water surface) hide the compression well.
Decreased color content
Night time scenes or poor lighting will reduce the amount of color in the source material. This will sometimes help mask compression losses, since there is less information to encode, and less color bleeding.
The important thing to remember is that no single setting works for all video scenes. You will probably have to try pushing the compression down to lower and lower byte rates until it becomes annoying, and then back off a bit. It's all about compromises
sinon la page complete de la FAQ
http://www.trix.com/buz/#faq05
bonne lecture..
Message édité par cyberjack le 23-08-2002 à 08:51:10
---------------
Cyberjack [:cyberjack]Membre de la communauté Varoise VarLan Varlan
