You can use the Basic Video Settings section of the Video tab to specify the basic video characteristics and compression quality, as shown in Figure 8-8.

Figure 8-8. The Basic Video Settings specify the base video format and compression quality.

These are the options that you can adjust:

Use the Bitrate Settings to specify the target data rate for the output video, and associated encoding process. The bitrate settings are shown in Figure 8-9.

The Bitrate Encoding parameter defines two attributes of the compression process: constant or variable bit rate, and single- or two-pass encoding.

The GOP Settings are used to specify the internal structure of each Group Of Pictures used in the MPEG-2 compressed video format. These settings are shown in Figure 8-10.

Figure 8-10. The GOP Settings specify the frame structure of each Group Of Pictures.

What is a GOP? Well, the MPEG-2 format segments runs of consecutive frames into groups, called a Group Of Pictures (GOP). The frames within each group then are compressed with reference to each other. MPEG-2 takes advantage of the fact that adjacent frames, appearing only 1/30th of a second after the previous (for NTSC), often look very similar. Very often the background remains basically the same from frame to frame, with some foreground objects moving against it. Or the camera may be panning across a scene, so most of each frame also appears in the previous frame.

Some video compression algorithms compress each frame individually, using an intraframe approach to only use information within the frame itself. These approaches exploit spatial redundancy within individual frames, such as blank walls, blue sky, and smaller patches with similar color and texture. For example, Motion JPEG video compression essentially compresses each individual frame using the same JPEG compression algorithm used for still images.

MPEG-2 and other interframe approaches break the frame boundary and exploit the temporal redundancy between consecutive frames by using information from adjacent frames. These motion compensation approaches therefore can achieve significantly higher compression ratios than intraframe approaches.

MPEG-2 compression subdivides each frame into blocks, and then uses blocks of information from adjacent frames to build up the next frame (as possible), and then adds in the details that have changed in the new frame. If not much has changed between frames (as in an interview shot with a fixed camera against a still background), only a small amount of additional data is needed to compress adjacent frames (a bonus for VBR encoding). If there is a lot of change between frames (e.g., with a moving camera, fast action, trees waving in the background), then the compression will require significantly more bits to add in this new detail. And, if the bitrate is set too low, the result will be the familiar visible blocks in the compressed frames where the compression ran out of bits and could not fully recover the original frame.

The MPEG-2 compression approach uses a rather sophisticated approach to employ information not just from the previous frame, but also from several frames before, and even from the following frames. Frames are predicted from their neighbors, by finding and reusing similar areas, which may even have moved within the frame.

However, this kind of relative encoding does introduce errors, which would propagate and grow if all frames were compressed relative to each other. In addition, video does contain scene cuts and other fast changes that cannot be encoded using intraframe methods.

As a result, MPEG-2 also regularly includes stand-alone frames with intraframe encoding. Called I frames in MPEG-2, these frames can be decoded independently, and therefore are useful for providing random access for jumping into the video stream (i.e., for DVD chapter points), and for providing fast forward and reverse playback.

MPEG-2 is designed to encode video using a regular pattern of frame types:

The trick with MPEG encoding is to find a good mix of these frame types to trade off the compression quality and size. I frames are significantly larger, but provide the best quality. P frames are smaller, but introduce some errors from motion compensation. B frames are smallest, but also have the most errors as they are generated from frames which themselves may be predicted. At low bitrates, you may even see a visible pulsing pattern when you play back the compressed frames, between the higher-quality I frames and the lower-quality B frames.

A Group Of Pictures (GOP), then, is a contiguous run of frames, starting with an I frame, and with a regular pattern of P and B frames. This pattern is specified in the GOP Settings section of the Project Transcode Presets using two values, M and N:

Use the Audio Format Settings menu to select the audio compression format, as shown in Figure 8-12.

Figure 8-12. Select the base Audio Format for encoding the audio.

The available options are:

When choosing audio formats, remember that NTSC discs must contain at least one track in either PCM or Dolby Digital format. This was an issue before Dolby Digital encoders were generally available in DVD authoring tools, since the uncompressed PCM format can take up a lot of space that you might rather use for more video, and the only available alternative, MPEG, was officially not permitted to be used as the only audio track on a DVD-Video disc. Dolby Digital is now a much better choice, since it is highly compressed and also provides great quality.

The Basic Audio Settings display the base characteristics for the selected format (see Figure 8-13). This information varies for each format.

Figure 8-13. Encore displays the characteristics for the selected audio format under the Basic Audio Settings.

These are the settings:

Use the Bitrate Settings to specify the data rate of the audio data stream, in thousands of bits per second (kbps). This applies only to the compressed formats (Dolby Digital and MPEG), and not to PCM, which is uncompressed and therefore uses a fixed data rate. See Figure 8-14.

Figure 8-14. Set the bitrate for the compressed audio formats.

Video Noise Reduction removes random noise from the input frames, such as introduced during video capture or from analog tape dropouts. Ideally, noise reduction is performed by finding spurious pixels that are dramatically different from the adjacent area or neighboring frames. However, values that look like noise may well be real information from a textured scene.

You can adjust the Noise Reduction slider between 0.0 up to 100.0 to apply more aggressive filtering to larger areas around each pixel. The result of this filtering is that blurring of the frames may occur, more visibly with the higher values. See Figure 8-17.

Figure 8-17. Apply Video Noise Reduction to reduce random noise in the video, although this can blur the frames.

DeInterlace filters the two interlaced fields that make up a NTSC or PAL video frame to remove the interlacing and create a non-interlaced, or progressive, frame.

Interlaced video, such as DV, creates each frame from two consecutive fields, captured one after another. Each frame of DV video is 720 pixels by 480 lines at approximately 30 fps for NTSC, while each field is half the height, at 720 pixels by 240 lines, and double the rate, at around 60 fps. Each pair of alternating lines in an interlaced video frame comes from different fields captured at slightly different times: the upper line of each pair from one field, and the lower pair from another. However, depending on the video format, the upper or the lower lines could be captured first.

Consequently, interlaced video frames with fast motion can look very bad when viewed as a still image, with every other line ripped apart by a panning camera or a fast foreground object.

Interlacing was initially introduced in the early days of television as a work-around that enabled the existing technology to capture video at a fast enough frame rate. Its legacy carries forward today in television broadcasts and displays, and the associated video cameras and recording equipment (such as VHS tape and DV camcorders).

As a result, the MPEG-2 digital compression format supports encoding each field of a frame separately, and the DVD-Video format supports storing and displaying interlaced compressed digital video. This makes great sense, of course, when displaying video on television displays. However, computer displays and high-end video monitors are native progressive displays, refreshing an entire full frame at a time, with no interlacing of alternate fields. Interlaced DVD material is therefore converted to progressive on these displays using software or hardware digital processing, and also taking advantage of the fact that movies on DVD originated as progressive frames of film stock.

If your DVD is intended for display on a computer or other progressive monitor, and not on regular television, you can perform de-interlacing as a pre-encoding step to store the compressed video on the DVD in a native progressive format. Specify the Field Order of the fields: Upper or Lower first.