||Imperfections, the pixalation on the frame. Caused by over compression and or low quality source. Most of the time you'll notice these on your final product, but often the underlying cause of them comes from the failure to collect data during the cap.|
||COmpression & DECompression. The software algorithm that allows the video to be stored in a compressed format, then decompressed at runtime. Even with so called "software only" capture cards, there is a built in chip that has a software codec burned in to it. Make sure you know this codec, it's limitations, and it's strong suit, and use it's strengths to the max.|
||The Rate, expressed in kilobytes (KB) or kilobits (Kb) per second, which a video & / or the video audio stream is captured at. In general, given identical frame rate, frame size and codec, a video with a higher data rate will have better quality than a video with a lower data rate. The best cappers tune their capture rate to the maximum their computer configuration can stand.|
||That dreaded loss of frames during the capture
process. If your capture software allows, you should monitor this number (
expressed as a percentage of the total ) and never accept a cap with greater
than 2% drop out.
a) Hardware Problems - compression chip, hard drive or bus controller - inability to keep up with the data stream from the feed. Simple solution is to reduce the rate of data. Best solution is replace the slowest component.
b) Software Problems - compression algorithm [codec], the software codecs inability to 'keep up' with the data rate called for. Simple solution is to change the "save" codec to one more compatible with the hardware codec chip - See Codec above
||Digital Television, and to some extent digital computing, still have
yet to come together. Check this:
January 2, 2002 – The Advanced Television Systems Committee (ATSC) has advanced its DTV Application Software Environment (DASE) specifications to the level of Candidate Standard. The DASE Standard will provide enhanced and interactive content creators the specifications necessary to ensure that their applications and data will run uniformly on all brands and models of DTV receivers while providing manufacturers the flexibility to choose hardware platforms and operating systems for receivers. The Candidate Standard stage is an explicit call for implementation and technical feedback. The DASE-1 suite of specifications is organized into eight logical parts and is available at .
The only current standard of true digital is the 16:9 frame size - a square of the current NTSC 4:3
- Frames per second
||This is a count of the number of frames meant to
captured each second.
As a point of reference,
American TV plays at 29.795 fps (60 cycles),
European standard is 25 fps (with 50 cycles).
See the discussion of Fields for more on this.
||The best way to explain fields is to start off with the review of FPS and electricity referred to below. Note "cycles". That is the measure of the number of times the electric current alternates back and forth each second. Your Tv screen is "refreshed" each 1/50th or 1/60th of a second based on this electric cycle. Each cycle the raster ( the little ray gun in the back of the tube that excites the phosphors on the interior of the front of the screen and causes them to glow ) starts at the upper left, scans an entire row of pixels from left to right, moves back to the left and down a row, and repeats the process until it reaches the bottom right. It then travels back to the upper left, in what is known as the blank interval more familiar to dos program writers who used to use this nano second of time to execute other code in the space occupied by the video memory. Most broadcast quality feeds are composed of an "a" field and a "b" field to match these cycles, with the "b" field slightly offset to create a smoother blend between the pixels of the "a" field. In terms of quality, the b tends to be the lesser of the two fields, and if your capture program allows it, and your computer is having trouble keeping up during capture, turn off the b field.|
||Quite simply, a still image that when played at the set fps produces the illusion of a moving image. Frames may be complete in and of themselves, or require information from adjoining frames to complete their image. By using information from adjoing frames, the data rate (amount of the frames' information * the rate at which the frame is to play = data rate) can often be reduced greatly.|
||Well, as of May 1st, 2002, all major broadcast stations (the top 100) should be putting out a HDTV broadcast. The FCC has ruled that June 2007 is the cut off for standard broadcast as we know it now. Because of the differing standards under which they built their digital network cable is behind, and meetings are still taking place over the American HDTV standard - all 5 of them - different than those in place currently in Japan and Europe. The common theme in all the hdtv and digital tv is the 16:9 framing standard. Stay tuned.|
|Key frames are frames that are completely built from scratch - they do
not rely on prior or future frames for information on their content - each
pixel of the frame is contained with the data. In mpg these are referred to
as "I" - complete frames, "B" - frames using unchanging
data from any number of frames before it, all the way back to the last
"I" frame, and "P" frames, frames capable of reading
pixel information from frames both behind it and in front of it - two pass
variable. When data does not have to be sent again, bandwidth is reduced, but
in higher seperations between "I" frames, most encoders will
"average" pixels - meaning in broad terms that a dark blue pixel
that has a shift to red at the next key frame might be exchanged for purple -
a gross example, not true, but it gives you the idea behind it.
As mmasw said:
As a rule of thumb, the only advantage that a large gap between key frames brings is in smaller file size. But at a cost both in Fast Forward and ReWind, and in quality. Quite a few of the default key frame settings I've seen would appear to be at the level for maximum "streaming" ability, which would not have much bearing on their use for usenet.
Cappers seem to set their keyframe interval set to one every 10 seconds.....maybe
higher. At that rate, "fast forwarding" by dragging the cusor means that there is a huge lagtime before the clip resumes play. It's extremely cumbersome....if you want that function.
Because I think that function is important, I always set my keyframe interval to one a second.....sometimes it's even smaller. That way resumption of play after dragging the cursor is nearly instantaneous.
Even on a huge file, the increase in file size by a small keyfame interval is insignificant.
In the past, it seems, every time I've "asked" a capper about why they have a large interval, they are rather surprised that aspect is under their control.
I'm wondering if that is the case now, or if there is a material reason for the larger, more awkward intervals?
|It's involved with streaming media - the ability to read forward to
know that an abrupt change it going to occur, and conserve current
bandwidth to compensate for the need to create the entire pixel set
for the first frame.
It's net effect for us in the video groups is mainly that people with
Consider scene change threshold as an "advanced Key Frame"
MOTHERBOARD -- A quality motherboard with a high speed bus makes the data flow from cap card to hard drive easy. The data from your capture device moves from your card across the bus to the CPU and then again to the IDE controller and on to the IDE Drive (or SCSI controller and SCSI drive). Bus speed on your motherboard should be at least 100mhz, and 133 is better. Speeds above 200 are getting to market now.
CAPTURE DEVICE -- gee, that's in the next section.
the Capture Device ( cap
There are at least 2 dozen cap cards out there, from mass produced $90-$200 starters, $200 to $600 mid range, and $1000+++ true video capture cards. You wanta do Tv? - think $10.000 to start. There are three considerations to step through before narrowing the field down for the right cap card for you.
COMPUTER -- After reviewing what's been discussed above about hardware, you should quickly see that buying a $800 capture device to be used on an old PII with 16 meg of memory might be a bit of Over kill. If your desire is to film, capture, and edit material that will be going into a HDTV DVD then by all means buy a card like this, then get the computer to go with it. For Capture of a standard DVD it might work with the PII. For Capture of VCD and most net produced content it's overkill. Price for the most part equals ability, when you factor in the bundled software. Know the price of this software (even the "lite" versions) and it will give you a starting point to consider what value they place on the caputure board itself.
CONNECTION -- This is a bit of same as hardware as well. A) If you don't have any free PCI slots, it would be senseless to get a PCI card unless you know you can remove one. Under this case, and for any with a crowded computer, this is the best time to consider an "All in One" Video card / Capture card like the AT and Assess. A USB connected external like the Fast's Dazzle is also a good choice if you already have a firewire port onboard. B) It's best despite the ability of win to share Iraq's if you can get your Capture driver to run solo, reducing the need for the OS to interrupt the cards cycle to check on the need of the device sharing it's IRQ. C) For those considering USB connected devices, remember the key Firewire need. A 1394, or best 1396 USB connection is about twice as fast as a USB2 port, and most certainly will be needed. If you install one of these, consider shutting off your on board USB ports in BIOS to free up resources.
HARDWARE CODEC -- Ever since the day that intel annouced the Intel Smart Recorder, manufactors have been claiming Hardware-compression-less cap cards. I say horse feathers. Every card I've put my hands on that makes this claim has somewhere on it a big fat Compression/decompression chip, including Intels. What most of these people mean is that the through-put of the capture is fast enough that you won't have to compress the signal that much due to the capture device itself.
It's important to know what / whoose chip is on the board your considering because this will tell you what standard capture settings will give you the best results. Many of the lower end cards have older Brooktree 800 series chips, which are only 16 bit color. Cap all day at 24 bit, it still is 16 until your final recompress. Most (if not all) of the first released DV capture devices use a varient of the Quicktime chip - which should have been good, but the mpeg standard moved away from it so fast that I'm not certain anymore if it will be the leader 6 months from now.