Working with Digital Video
Windows 98 has a much more efficient disk subsystem than Windows 95, which
allows better writing to the hard disk. This can significantly improve capture
quality and editing speed. Windows 2000 is preferred over any other Windows
version. Windows 95, 98 and Me will not allow AVI files over 2 GB. If you use
Windows 2000 or XP, there is no such limitation, but do turn off some of the
operating system's memory-hogging features if you use XP.
If the system is slow, it may not write some frames to the disk and dropped
frames will occur. The result is a null frame that takes up virtually no space
and simply mimics the last frame. More dropped frames result in a jerkier video
and are more apparent at low frame rates. Dropped frames usually come in bursts,
thus the video will have large jerks instead of slight jitters. If you are
compressing the video in real-time, dropped frames occur when the data rate is
the highest-when the most action occurs, which is the worst time for a jerky
video.
To avoid dropped frames, you can try decreasing the frame rate or the video
resolution. Also, optimise the hard disk and close all open applications while
capturing. If you still don't get a smooth video, turn off preview while
capturing if possible. You can also try reducing any real-time effects the
capture software may be applying to the video. If you have a good processor but
a slow hard disk, increase the compression quality so that smaller chunks of
data are written to the disk.
Many codecs compress video with keyframes. One frame (the keyframe) stores the
entire image with the colour information of every pixel. For subsequent frames
(delta frames), only the differences between those and the preceding frame are
stored. To get to a frame, the decompressor will first get information about the
keyframe, then all subsequent frames to build the final image. More keyframes
mean larger files, but lesser seek time. Also, frames further away from the
keyframe are more susceptible to low quality, since each time only an
approximate difference is stored. Thus, you should keep the keyframe interval to
as low as possible.
If your hardware allows it, plan beforehand whether you want to record sound
along with the video itself or from an external source. An external source
usually means much better quality, but can also prove to be sluggish and is a
hassle.
Since one audio file is created for every video, you have to keep track of
double the number of files in post-production. Also, synchronising audio and
video involves more time and has to be done perfectly. However, if you capture
video and audio simultaneously and the system is slow, it may be susceptible to
dropping frames as it has to process the video and the audio at the same time.
Files stored at the start of a disk are faster to access. Make sure that the
video is captured to a partition which is at the start of the disk and which is
not already too full. Note that the position of the actual data matters, not the
size or position of the partition. For a disk split in half, data at the end of
the first partition (when the drive is nearly full) will be read at the same
speed as data at the start of the second.
If you are applying multiple effects to the video, arrange them in the best
possible order if the software allows you to do so. Doing so can cut the
conversion process by half or even more. For example, if you apply a filter to
smoothen out the video and another one to resize it, place the resize filter
first-the smoothening filter will have to work on a smaller image and also, you
would want the smoothening to work on the resized image for best quality.
If you intend to compress the video with DivX, the final movie size should be in
multiples of 16 pixels. Sometimes it can be in multiples of 8 and, very rarely,
32. Thus, a 320x240 video will compress well with DivX, but 321x240 will not.
When capturing video, pick a frame size close to the final output. If you're
going to make a 320x240 video, it doesn't make much sense to capture at 640x480,
since it takes four times the space to do so. Ordinarily, do not set the frame
rate higher than 30 fps.
AVI files with CD-quality audio (176 Kbps) are quite a waste of space,
especially for home videos. For distribution, 8-bit audio at 44 KHz is good
enough. Also, compress the audio with a codec. This will reduce the size of the
audio to between half and a quarter of its original size with very little loss
in quality. MP3 is the best codec to encode audio. However, on slow machines it
can cause synchronisation problems in the video.
Use computer-generated graphics sparingly with video. Video compressors are
adapted to natural images and do not perform optimally on saturated colours and
sharp edges. Depending on the codec, title screens and overlay items can
adversely affect video quality, especially at low bitrates.
If you must include them, make openings and trailers independent-don't use
transitions to blend them with the video and have them start and end on keyframe
boundaries. Align the position and size of overlay items to 8x8 or 16x16 pixel
boundaries and make them opaque. Transparent regions on the logo will be
translated as a complex tile by the codec and will take up a lot of space. If
the logo is opaque, the codec will ignore it between keyframes, saving a lot of
overheads.