20 - Scaling and Deinterlacing
Video source material is never quite as good as we want it, and always seems to lag behind our display capabilities. A popular thing to do is scale and/or deinterlace our video in order to present the absolute best picture possible on the display.
Deinterlacing is the act of taking the odd and even fields of video that were transmitted, and reordering them to be presented as a whole picture. The best use for this is for film over video. A film frame, which was a photograph at one instant in time will probably arrive to you via DVD. DVDs are good old NTSC video, which is interlaced. In order to encode a 24 frame/sec film onto a 30 interlaced frames/sec video system, a process called 3:2 pulldown is performed that maps a single film frame onto an uneven cadence of odd and even fields, and half frames that might contain part of one film frame and part of the next. A deinterlacer is actually doing more than simple deinterlacing. It is actually looking for the odd cadence of the original film, and reassembling the film frame from all the odd/even fields.
Some video is shot at 1080i, and it is common to try and deinterlace the image for progressive display on a 720p display. However this has it’s own set of challenges in trying to display an image out of it’s original order. This is especially difficult when objects are moving on the screen like in sports.
Scaling is a bit easier compared to deinteralcing, but still a mathematically intensive operation. Most people can understand simply doubling the size of an image, but this doesn’t mean simply repeating the pixels. A good scaler will look left and right, up and down, and maybe even at other frames to figure out what the pixel in between should be.
Scalers and deinterlacers go under the more generic term of Video Processors. Selecting a video processor depends on several variable like what the primary source material is, how many inputs are needed, and what the output display is.