Optimizing Your Encoding System | Microsoft Windows Media Resource Kit (Pro-Resource Kit)

Encoding requires a great deal of CPU resources. You can minimize the CPU demands, especially when capturing screens or encoding a live broadcast, by following a few simple guidelines:

A fast computer. Using a fast computer is especially important if you plan to capture and compress content in real time. A computer with two or more processors is your best bet to ensure that enough horsepower is available to process audio and video samples in real time. The encoder can use up to four processors for video and two for audio. As a general rule, if you find that you continually reach or exceed 80 percent of the CPU load (as indicated on the General tab of the Monitor panel), you may want to consider upgrading your computer.
Supported audio and video cards. For a complete list of supported cards, see the Partner Center page of the Windows Media 9 Series Web site.
Microsoft Windows XP Professional. Windows XP Professional takes full advantage of a multiprocessor computer by automatically spawning separate threads to enable processes to run different parts of their programs on different processors simultaneously. This parsing of the work enables you to perform high-end encoding tasks without maximizing your CPU usage.
Check network bandwidth. It is rare that you will achieve 100 percent throughput when sending content over a network. For optimal results, ensure that your network can accommodate the aggregate bandwidth required by your broadcast session plus the overhead for HTTP.
High-performance hard disks. For archiving or file conversion, use a high-performance hard disk, such as SCSI hard disks operating at 15,000 rotations per minute, in conjunction with a RAID 0 configuration. Using a high-performance hard disk and a RAID array can help to offload performance requirements from the main CPU, freeing up more processing power for encoding. In addition, it is useful to use the hard disk array only for encoding purposes.
Keep the encoder on a dedicated computer. If you are streaming content from a server running Windows Media Services, it is recommended that you run the encoder on a separate computer from the Windows Media server.
Turn off video preview and postview. The Input and Output areas in the main encoder window show you what your content looks like before and after it is encoded. Turning off these displays improves encoding performance. Or, to achieve a minor performance improvement, you can reduce the views.
Store captured content temporarily. If you are encoding to a file while sourcing from devices, store the content to a hard disk temporarily. In addition, use a different hard disk than the one dedicated to encoding.
Disable hardware acceleration. By disabling hardware acceleration, you lower the video quality and frame rate of the display, thus simplifying the encoding process. To disable hardware acceleration, open Control Panel, click Appearance and Themes, and click Display. On the Display Properties page, click Settings, click Advanced, and then click the Troubleshoot tab. Move the slider to the left toward None.
Use an NTFS volume. The FAT32 file system has a file size limit of 4 GB, so an NTFS volume is a better choice for storing content.
Optimize your computer. To optimize your computer, defragment your hard drive, turn off network and file sharing, close unused applications, and disable any services that you aren’t using in order to free up system resources.

When you are capturing video of the computer screen, keep the following recommendations in mind:

Use a lower frame rate, such as 5 fps. The higher the fps, the more memory required to process the data.
Set your monitor to use a lower display size. For example, capturing at a 1024 x 768 resolution yields more data than capturing at 640 x 480.
Use a fast video card and set your color display to 16-bit or 8-bit color. Or, to further improve the CPU usage set the encoder to run in 256-color mode. Keep in mind that running in 256 colors can introduce a palette change when switching between programs that have different color schemes. The encoder does not support palette changes for real-time screen capture. Therefore, you should use 16-bit colors or capture to an AVI file first, then use the encoder to convert the file to Windows Media Format.
Avoid multiple scene changes, such as opening and closing windows in rapid succession during the capture.
Capture screens or regions only from the primary monitor when using a dual-monitor scenario.
Set up the output frame size to be the same as the source. Do not use the encoder to resize the source video.
Use a solid background on your desktop. Using complex designs, pictures, gradient shading, or multiple colors requires more CPU resources because there is more data to capture.

Video Optimization

Similar to choosing a codec or encoding mode, choosing the kind of optimization to apply to your content depends on the content itself and how it will be played back.

TV Sources

Your source was shot with TV cameras, then it is interlaced. Maintain the interlacing during encoding when you intend for the content to be played on a television, such as when using a set-top box.

When the content will be played back on a computer, either maintain interlacing and allow the decoder to deinterlace the frames during playback or apply a more sophisticated deinterlacing filter prior to encoding.

Film Sources

If your source was shot on film and then converted to video, it is telecined. Apply the inverse telecine filter to eliminate the extra frames and set your frame rate in the encoder to 24 fps. Inverse telecine is intended for NTSC content only. Do not use this filter for PAL content.

Unknown Sources

If you are uncertain of the origin, use the encoder to detect the format.

Mixed Sources

If your content is a mix of film and video, then it is telecined and interlaced. Maintain the interlacing during encoding when you intend for the content to be played on a television, such as when using a set-top box.

Note that these recommendations apply only when you are encoding interlaced or mixed content of the typical 480i (NTSC), 576i (PAL), or 1080i formats without cropping any video lines (resolutions such as 640 x 480, 720 x 480, 320 x 480, 720 x 576, 360 x 576, and so on). Otherwise, you should always deinterlace before encoding.