Anamorphic information

Introduction

Traditionally we, and computers, think of ‘pixels’ as square dots, but in the world of video this is not necessarily the case. The width:height pixel ratio is not always 1:1. The reasons for this are historic, allowing analogue and digital video to coexist. In relation to DVD video, pixels can only be stretched along its horizontal plane (left-right, widthways). So the pixels of a DVD movie that make up its height, 576 for PAL, 480 for NTSC, are not allowed to be stretched vertically. But, in contrast, the pixels that make up its width, 720 for both PAL and NTSC, are allowed to be stretched.

Take a look at the three little video screenshots below. The first picture shows how the video is stored on the original DVD. If we were to view it like this it would have an incorrect aspect ratio, they all look too thin, it either needs a reduction in its vertical plane (top|bottom needs squashing) or stretching of its horizontal plane (left|right being pulled out). The second picture shows a correct aspect ratio, the vertical plane has been reduced, all is well, and no one looks too thin or too fat. This would be the traditional method used when converting DVD video; it is also video.NETs default, none anamorphic method used. The DVD video will be resized preserving the correct aspect ratio. But, there is another method, anamorphic encoding (the third screenshot).

Original picture as stored on DVD, incorrect aspect ratio.		Traditional resize methods, reduced height, correct aspect ratio.		Original signalled to expand horizontally, vertical preserved, correct aspect ratio.

Remember when we said 'not to alter the height of the vertical plane'... Well, that would only then allow us to alter the width, in our original 720x480 example we would need to horizontally stretch our image to 853x480 pixels to obtain the correct aspect ratio. That means 'up-scaling', resizing and encoding our video to a larger size than the original which is just silly, the file will need more bits to encode, take longer, just a complete 'no-no'.

So what do we do? Well, we do what the original DVD video does; we encode and store it as anamorphic. We encode and store it just like that in the first screenshot shown above and signal that the pixels are not square. The player viewing the video will then take this into account and display it correctly, just like in the third screenshot, and just like your DVD player would, no resizing during the encode is required.

Strict or loose

The anamorphic method we have spoken about so far is ‘strict’; it will store the exact pixel ratio within the video file, problem is, sometimes being exact is not good. Numbers like 853 are not exactly ‘computer’ numbers, or more importantly, encoders! Encoders like numbers that divide by 16, that’s usually always been the best option, but modern day codecs are much better and numbers that divide by 8 and 4, and even 2, will suffice. So what happens with loose anamorphic is the same as strict but it then slightly alters the size if need be to make sure the dimensions cleanly divide by 16, or 8 in the case of video.NET.