DVD/MPEG White Paper
Ever wonder why the DVD’s you create don’t look as good as the original video on your edit suite?
Well that’s because DVD (Digital Video Disks) require the original video to be compressed to an MPEG (Moving Picture Experts Group) standard in order to fit a mass of information onto a relatively small disk (4.7gig single layer, 9.4 dual layer).
MPEG-2 compression is a lossy compression, that is to say, it has to sacrifice some data in order to fit the otherwise massive uncompressed video file into the space available on a DVD. In the case of MPEG this is done by using a Group Of Pictures (GOP) structure. This means that the movie is encoded by compressing several frames together up to a maximum of 15 frames for PAL or 18 frames for NTSC. Each GOP chain begins with an ‘I’ frame (full frame) followed by a number of ‘P’ and ‘B’ frames which are not full frames themselves but rather updates of the previous frame. The longer the GOP the more efficient the file size but the less robust the quality of the encode.
Bitrate is what determines both the duration of video that will fit onto a DVD as well as the quality of that video. Encoding with a low bitrate means you can fit more video onto a DVD but at lower quality as opposed to a higher bitrate - better quality, less duration. This brings us to CBR (Constant BitRate) vs VBR (Variable BitRate). CBR will encode a DVD at a fixed bitrate (6 megabits per second being average) from start to end with faster encoding times. VBR will take longer to encode, but will analyze the movie first to determine which parts need higher bitrates, (action scenes, quick camera pans) and which parts of the movie can afford to have lower bit rates (fixed camera interviews, static scenes). Going back to GOP structure where only the changed pixel information from one frame to the next is required, higher bitrates are essential to preserve picture quality especially in fast moving scenes. With CBR this could mean faster scenes will not get enough bits allocated to them while bits are being wasted in static scenes.
There are two options when it comes to VBR encoding. Single pass and multi pass. When using variable bitrate the encoder must look over the movie to map out how it intends to distribute the bits of video data. The more passes over the video the encoder does before finalizing it’s bit map the better the bit allocation. Each pass the encoder does before the final encode refines the bit distribution of the DVD a little more. Some encoders will allow up to 99 passes before finalizing. Although 99 is probably excessive, 5 or 10 passes can be very worth while as opposed to the 1 or 2 pass maximum available on some basic MPEG encoders like Compressor, DVD Studio Pro and Encore. Programs such as DVD Studio Pro and iDVD are only able to perform the most basic of encoding algorithms internally and, although convenient, produce very average results.
However, the quality of the compressed film is determined by more than just settings. All DVD encoders are based on an algorithm, that is to say, what the encoder chooses using mathematical equations to determine what is to be sacrificed and what information is retained.
Not all DVD encoders are created equal. The quality of the encoding algorithm being the main difference between high quality encoding and video suffering from artifacts, banding and poor motion, it is vital to use professional encoding equipment to avoid disappointment in the finished DVD. Good encoders will effectively determine where and when to sacrifice bitrate in order to preserve the most important video information compared with average quality compression, which may make bad choices resulting in compression artifacts and inefficient bitrate allocation (see figure 1).
