Ask Professor Phil: Video Conversion

Ask Professor Phil: Video Conversion

Dear Professor Phil,

I’ve used video conversion programs to convert between different formats. For example, I’ve tried converting wmv files to mov files. There always seems to be some degradation in quality. Also, it’s often variable depending on the software that is used. Why does this happen?

James, Butler, PA


Video files are usually compressed using a lossy compression. That means that all of the brightness and color captured by the camera is not being preserved in the compressed file. Nearly all compression techniques depend on representing the brightness and colors by estimating them with numerical values representing their intensity. Ultimately, the highest stored values contain the most critical information and the lowest stored values represent less important information. There is also information, generally representing redundant parts of the scene, which is discarded during the compression process.

When the software process starts the conversion process, it essentially tries to reconstruct each frame. The object is to construct an image that is as close to what the camera captured as possible. The result cannot be perfect because the colors and brightness were estimated and some information was discarded. Then, the compression process is repeated using the algorithm of the second format. However, that algorithm is likely to represent the scene with slightly different estimated values and to determine somewhat different redundant information.

Figure 1 (above) shows the original video frame. Figure 2 (below) shows the scene after two conversions.

When the second format is played out (decompressed), it presents each frame based on values that have had twice been estimated and partially discarded. It cannot be as accurate as the representation that was based on the first compression algorithm.

Since there are no standards for such conversion processes, each manufacturer will handle the procedure differently. This accounts for the variation in software product performance.

Phil Hippensteel, PhD, is a professor of information systems at Penn State Harrisburg. His columns appear every month in AV Technology. Send your questions to