What's In a GOP?
You can find more compression settings articles in our Supercharging Compressor series index.Update: Sorry, I seem to have erroneously published an empty draft of this article. My apologies. Here's the text ...
As you may be aware, the DVD standard uses a form of video compression called MPEG-2. There are a few aspects of this codec that can impact the quality of your finished video -- and many of them stem from the fact that the codec works not in terms of frames, like the rest of your video process, but in terms of "Groups of Pictures," or GOPs.
You see, MPEG assumes that, in most video, frames that occur close to one another are probably fairly similar. Therefore, it picks one frame every now and again to describe completely, then describes neighboring frames only in terms of how they're different from that "reference" frame.
The full frame, called an "I" or "intra" frame, is the one that contains full information. MPEG-2 divides the picture into a series of 8-pixel squares, then describes each one with some mathematical shortcuts. Normally, you encounter an "I" frame about every 15 frames in regular DVD video.
The remaining frames in a Group of Pictures are reference frames -- "P" and "B" frames, for "predictive" and "bidirectionally predictive," respectively. The technical details don't matter too much at this stage, but essentially, either kind of reference frame divides the picture into 16-pixel squares called "macroblocks." Then, it decides whether each macroblock can be summed up by looking to the same area in a nearby frame ("do what you did back there, but move it a couple pixels to the left") or by describing it as a new block.
This article is mainly laying the groundwork for some forthcoming heavy-duty discussion about GOP size and structure in DVD authoring, but there are also some side lessons. Remember, you can only place chapter markers on I-frames, and now you know why: when a DVD player tries to skip to a point in the video, it doesn't know what to display on-screen unless the frame contains all of the information necessary to draw it!
The full frame, called an "I" or "intra" frame, is the one that contains full information. MPEG-2 divides the picture into a series of 8-pixel squares, then describes each one with some mathematical shortcuts. Normally, you encounter an "I" frame about every 15 frames in regular DVD video.
The remaining frames in a Group of Pictures are reference frames -- "P" and "B" frames, for "predictive" and "bidirectionally predictive," respectively. The technical details don't matter too much at this stage, but essentially, either kind of reference frame divides the picture into 16-pixel squares called "macroblocks." Then, it decides whether each macroblock can be summed up by looking to the same area in a nearby frame ("do what you did back there, but move it a couple pixels to the left") or by describing it as a new block.
This article is mainly laying the groundwork for some forthcoming heavy-duty discussion about GOP size and structure in DVD authoring, but there are also some side lessons. Remember, you can only place chapter markers on I-frames, and now you know why: when a DVD player tries to skip to a point in the video, it doesn't know what to display on-screen unless the frame contains all of the information necessary to draw it!
Listed below are 0 links to blogs that reference this entry: What's In a GOP?.
TrackBack URL for this entry: http://www.geniusdv.com/weblog/mt-tb.cgi/1054
Leave a comment