Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
13th January 2005, 19:36 | #1 | Link |
Confused User
Join Date: Sep 2004
Posts: 145
|
Resolution
I'm wondering, when a video is resized to a lower resolution in order to improve the compressability, doesn't the loss of information instead handicap the way the codec compute the motion and "stuff" ?
Then, what should be taken in consideration when encoding at low bitrate knowing higer resolution "helps" the codec to "understand" the animation but may (or may not ? O_o) cost more bits. |
13th January 2005, 20:03 | #2 | Link |
Moderator Emeritus
Join Date: Dec 2001
Location: The NW corner of Lake Erie
Posts: 5,552
|
I assume you're asking about MPEG-4 compression technologies, because most MPEG-1/2 compression is done with fixed resolutions.
There's a great document that describes very clearly how various MPEG-4 technologies work: Motion Vectors, Quantizers, Matrices ... even though it's the DivX User Guide, it talks to MPEG-4 technologies in general. The best practice among MPEG-4 encoding members is to use techniques that choose a resolution that gives the best possible quality for any given average bitrate. This is usually done through a compressibility test. Since each video that you compress will have different compressibility, there is no one-size-fits-all resolution that works for any particular bitrate. A compressibility test uses the formula <value> = bits / (pixels * frames). This single number takes into acount both bitrate and resolution. The technique is fairly simple: A resolution is selected, and an analysis is done to determine the maximum possible bitrate at that particular resolution. The resulting b/p*f value is used to select a new resolution where the new b/p*f value is 60-80% of the maximum. This is the methodology used by AutoGK and GK for encoding DivX and XviD.
__________________
"It is easier for a camel to pass through the eye of a needle if it is lightly greased." Kehlog Albran, The Profit |
13th January 2005, 20:43 | #3 | Link |
Confused User
Join Date: Sep 2004
Posts: 145
|
I don't understand but a bit of what you'r explaining ^^' (Even if I already read much about motion vector (The whole XVId unofficial FAQ and even before that, the Dirac's algorithme explainations).
I'll try to get info on (Auto)GK. But just to know : To figure out the compressability will I have to encode over and over again ? Last edited by Sergejack; 13th January 2005 at 20:46. |
13th January 2005, 21:42 | #4 | Link |
Moderator Emeritus
Join Date: Dec 2001
Location: The NW corner of Lake Erie
Posts: 5,552
|
I don't understand but a bit of what you're explaining...
You've asked about MPEG compression technology. The Guide I pointed you to speaks to the various technologies used in MPEG-4 codecs.
...To figure out the compressability will I have to encode over and over again ? No.
Start here.
__________________
"It is easier for a camel to pass through the eye of a needle if it is lightly greased." Kehlog Albran, The Profit |
Thread Tools | Search this Thread |
Display Modes | |
|
|