Preproduction issues in streaming

If you have to shoot in DV and need to deliver in many different formats, cut and edit in the native format, and then only transcode the final material.
Publish date:
Social count:

If you have to shoot in DV and need to deliver in many different formats, cut and edit in the native format, and then only transcode the final material.

Preproduction for streaming is an interesting topic and one that is not commonly covered. Streaming, unlike broadcast video production, requires extra care when capturing images for encoding and packetization.

It is important to plan how your content will be delivered. Knowing ahead of time whether your viewing audience will be receiving your content over the Internet with 28.8 modems or on corporate LANs with 10MB to their desks makes all the difference in the world. Once the technical parameters of delivery are known, you need to give thought to how the content will be used. Determine what kind of information must be gathered and what metadata needs to be incorporated during production. Additionally, consider the difference in how the various media players interact with the data.

Most corporate networks today are large and, comparatively, the Internet is a behemoth. But it's still not big enough or robust enough to handle uncompressed audio/video. We can't easily stream data at 670MB/s over it. So we must capture our content and be prepared to compress it. The worst thing you can do to a digital signal is to compress in one format and then decode and re-encode in another — talk about garbage in garbage out.

Keep your compression signal path clean. What does this mean? First, try to ensure that the devices used are not additive to the signal. In the digital chain you should keep the transcoding from one codec to another to a minimum. If you are going to capture your content in a compressed format try to deliver in that format.

If you have to shoot in DV and need to deliver in many different formats, cut and edit in the native format, and then only transcode the final material. Just prior to final transcoding you may want to use various adjustments to the content to guarantee high-quality images and more efficient file size. Below are explanations of some of the adjustments that can be made using encoding tools on the market.

Gamma: used mainly to adjust for differences between Macintosh displays and PCs

Brightness: controls the overall luminance value of the image.

Black/white restore: functions enable better compression of the solid black and white areas of the video by reducing the noise in the luminance.

Image size: converts non-square pixel formats to square so that video will display properly on computer monitors.

Telecine: removes the 3:2 pulldown and only encodes the original 24 frames, which reduces the bit rate by eliminating the video artifacts.

Blur: allows better compression by softening the edges of the images and reducing the differences between adjacent pixels.

Watermarking: lays a graphic file over the top of the video. Warning-text and graphic overlays usually cause the encoder to spend more bits to get these areas right. You can get away with this at higher bit rates but avoid it if you are trying to squeeze every last bit.

Codec: allows you to choose from a number of compression algorithms. A lot of people have been using the International Standards Organization (ISO) MPEG-4 codec lately to deliver high-quality video compression at low data rates.

Key frames: forces an I frame at a specific locations such as scene changes, fast motion sequences. This gives you a much better image, of course at the cost of more bits.

Video data rate: Here is where you get to set your final data rate based on your intended audience. Set the rate too high to get better-looking images, and you limit the number of users who can access it. Set it too low and who would want to watch? As a rule of thumb broadcast quality video is achieved with as little as 500Kb/s and near-DVD quality at 750Kb/s using MPEG-4.

Adjustments can also be made to the contrast, saturation and hue of the image.

Media Cleaner Pro is my weapon of choice for making these adjustments. It has many features to help you to make adjustments to video for streaming and supports all the major codecs. Since most of my streaming content is meant for our internal network I stick to the broadband streaming settings. In the case where the content is to be downloaded and not streamed I try to use Variable Bit Rate MPEG-2 (DVD).

Media Cleaner Pro allows you to manipulate almost every aspect of an image during the encoding process. The best way to learn about these powerful settings is to practice. I usually edit a 30-second representative test piece of my content, and then start playing with the settings. Beware — a little goes a long way.

Steven M. Blumenfeld is currently the GM/CTO of AOL - Nullsoft, the creators of Winamp and SHOUTcast.