Shooting Streaming Video - Page 3

Limitations

Windows Media is not a panacea for all the problems of video delivery. As we discussed above there are limitations on quality and QoS that can only be currently overcome via satellite delivery. There are also issues of bandwidth within a corporate LAN. While a 10 or 100 base-T LAN (100 Mbits/sec) may seem huge compared to a T1 line (1.5 Mbits/sec), if many users start using it at the same time it can quickly get filled up. Multicasting over satellite (defined above) can help to alleviate this problem.

We've discussed the various markets for Windows Media, defined terms and discussed the limitations of streaming video. Now let's get down to how shooting streaming video actually differs from shooting video on a BetaCam or other camera for playback from tape or live.

When shooting video that will eventually become Windows Media it is important to remember one thing: THE CODEC.

We defined codecs previously as an algorithm that compresses video for delivery over the Internet. What we didn't mention is that to compress video you have to get rid of data. In the case of taking full screen video and compressing it down for Internet delivery you have to throw away lots of data. Thus it is important to understand how the codec works so that your video look its best when played back on the Internet.

Motion: Motion is your enemy! Codecs work best when not much changes between video frames. This is why talking heads work especially well; there is very little change except in the mouth area. A codec looks at the two frames and only needs to add enough data to describe the change to the mouth. This is because the first frame and second frame are basically the same.

If you are doing talking head video then make sure the speaker doesn't move around and avoid moving the camera. This will make it much easier to compress the video later. Avoid multiple camera cuts as this can wreck havoc during encoding.

When doing your encoding always select the setting that optimizes the type of video you have shot. Windows Media can be tweaked to better display fast motion video such as someone running. It sacrifices clarity of the image for smoothness.

Tripods: Always use a tripod instead of a handheld source. If you have to use a handheld use an image stabilizer to reduce the amount of apparent motion. With the added stability the codec can dedicate more bandwidth to the details of the speakers face instead of the moving background.

Edges: Another way codecs work is to take sharp details and blur them. Having sharp edges in a video can result in the codec blurring them. Try to avoid sharp horizontal or vertical lines.

Lighting: You may have noticed that Windows Media often looks washed out. Sometimes this is because the video is shot in small studios or in poorly lit areas. It can also be because that picky codec also throws out some color data during compression. The more colors it has to remember the more data, thus the codec will merge colors that are similar in color into one color. This results in blurry video and lack of range on the colors.

I also tend to think encoding darkens video though I've never found any data to support this. When in doubt keep things nice and bright while shooting your source video.

Transitions and Wipes: Remember that pesky motion. If you do a wipe or transition the codec must process many frames where change is present. If you do a clean cut from one scene to the next then the codec only has to deal with change in the frames.

Also avoid panning and zooming the camera; this results in lots of change to the background.

Simple Backgrounds: The simpler the background the better. With a simple background if you pan or zoom then the background doesn't change. Its just more of the solid color, the codec can use its bandwidth to concentrate on details in the face as it enlarges or moves.

Audio: Audio always seems to get overlooked in small, fast video productions. Remember, the codec is also working to compress the audio. If you have a talking head video make sure the voice is loud and clear. Avoid subtle whispering and above all avoid background music tracks. A combination of music and voice takes up more bandwidth since it occupies a greater decibel range. Windows Media Encoder allows you to select the types of audio that are present. Current codecs allow you to squeeze acceptable voice quality well below 1kbit/sec. If bandwidth is limited during playback the Media Player will drop video frames before sacrificing audio.

Subject Size: When framing, make sure your subject takes up more of the image than you normally would. Though the subject may look perfect when shooting the raw video the finished product is going to be small, probably only 160x120 pixels or 320x240 pixels on a computer screen. Increasing the size of the subject relative to the whole screen will help to bring out details that won't be present if you leave the image small.

These techniques should be used when shooting the video to analog tape or to Digital Video (DV). Once the video has been captured it needs to be moved over to the computer. If you've captured using a DV camcorder then simply dump the video across using a Firewire card. If you've recorded to tape then there is another step.

Terms - Page 2
Windows Media: Encoding
Capture Cards - Page 4