Need your help relating to video compression - please help

Something to keep in mind about video compression is this...it's not as much about the length of the video as it is about the information contained within the video.

And that information is related to the color of the pixels.

As an example, a 1280 X 720 (16:9 aspect ratio, also considered a HD format) contains 921,600 pixels per frame. If you are not aware of the fact that video is made up of individual frames played in sequence to give the illusion of movement, now would be a good time to understand this.

A typical web video is produced at 30 frames per second (29.97 fps) which means that for every second of video 30 images will be used to create the video. TV uses a 24 fps format. The human eye can process images at 1/20 of a second and needs a faster rate than that to give the illusion of constant motion such as many videos contain.

Since it is the changing of the color of the pixels that determines what is shown on the screen and hence what your brain will process, you need to understand that the more color changes per pixel the more information you will need encoded in the video.

A white background with text that changes every 5 seconds will have only a small fraction of pixel changes compared to a moving background/foreground such as a camera following a skateboarder, as an example. Hence the skateboarder video can be considerably shorter in length than the text video yet still be larger in file size. This is due to the amount of information that is contained in the rendering code. More and faster color changes by the pixels equals more information required which equals a larger file size, all things being equal on a compression basis.

What the compression algos do is allow for repeated pixel blocks and colors to be grouped into information blocks and that saves lines of code which translates into a smaller file size.

By applying this understanding to your video creations you can experiment with different rendering formats and compression algos to determine the optimal render setting for any given video.

For example, if it's a talking head video with you speaking into a camera you know the video will contain movement and hence will need a faster frame rate than 20 fps. However, if the video is text on a white background that only changes every 5 to 10 seconds (virtually no movement) you may be able to render the video at 10 frames per second and have it yield a very small file size.

Additionally, when you create video you will want to do all the preliminary work in as uncompressed a format as your software and computer will allow you to work with because every level of compression removes information and causes a slight deterioration of the images. Ideally, all the video work should be done in an uncompressed format until the final rendering. This however, may take a high end computer to accomplish. I know it did for me, but I do mainly 3D graphical work.

Also, when you do the final rendering for your videos be sure to use the lowest setting that will be a good compromise between file size and video quality. As an example, when final rendering using Camtasia I will start out at 35% and move up until I'm happy with the results. Generally, no more than a 45% setting is required for me to reach that threshold. But that may also be a function of the CPU/GPU I'm working with so YMMV.

~Bill