Does AC3 take less space than the MPEG-1 I get with TMPGEnc? Will MPEG-1
audio play on a DVD player?
AC3 and MPEG-1 both support multiple bitrates. However, The NTSC DVD format
requires that a DVD have at least one audio track that's either PCM
(uncompressed) or AC3. The PAL DVD format requires that a DVD have at least
one audio track that's PCM, AC3, or MPEG-2 audio. Neither spec supports
MPEG-1 audio. You _might_ find that some players will work with MPEG-1 audio
on a DVD (since most will play VCD's, which use MPEG-1 video and audio), but
it'd be a gamble, and you'd also have a hard time finding a DVD authoring
program that'll support MPEG-1 audio on a DVD-Video project.
On bit rate on the video, what degrades as a select a lower rate. On
TMPGEnc, it shows the same resolution and the same refresh rate. What
happens to the video as I select a lower rate?
Compression doesn't change your resolution or your frame rate. It changes
the quality of what's there.
The short answer is that you'll see blocking artifacts and noise at edges as
you reduce the bitrate.
The longer answer involves a basic explanation of how video compression
works. In a nutshell, MPEG-2 creates a number of I-frames, every X frames
(usually about 15, which I think it the largest I-frame spacing allowed on
DVD video). That I-Frame is a compressed picture. The color data in the
picture is downsampled by 2 in each direction (so you have a 720x480 black
and white picture with 360x240 color information on it), and the luma (B&W)
and luma (color) channels are DCT compressed. They're broken into 8x8 blocks
and transformed with a Discrete Cosine Transform (DCT), which results in an
image that has better energy compaction (the "most important" stuff is in the
beginning of the data stream, and the less noticeable stuff is at the end).
Then it throws away some of the less significant bits. So far, we've
basically covered how JPEG works on images, too (Except that with JPEG, the
chroma downsampling is optional. With MPEG-2-for-DVD, it's not). The
up-to-14 frames in between each I-Frame are filled with P and B frames. P
frames are "predicted" frames from the previous frames. B frames are
"bidirectionally predicted" frames based on the frames before and after it
(so they're more efficient, but require knowledge of future frames, and take
longer to generate). I don't know MPEG well enough to know exactly how the
predicted frames are generated. I expect that it's a basic interpolated
image from the previous few frames, and a compressed representation of a
difference image - the data needed to go from the prediction to the actual
picture (because if 14/15 frames were _only_ predicted, without correction
factors, they'd look terrible).
Okay, so that's a bit of a large nutshell, but there you have it.
--
Erik Harris n$wsr$ader@$harrishom$.com
AIM: KngFuJoe
http://www.eharrishome.com
Chinese-Indonesian MA Club
http://www.eharrishome.com/cimac/
The above email address is obfuscated to try to prevent SPAM.
Replace each dollar sign with an "e" for the correct address.
