ErMaC's Guide to All Things Video - Part 2
The Big Picture - Interlaced vs. Progressive, Fields vs. Frames, 3:2 pulldown and Inverse Telecine
Continued from the previous page.
3) 3:2 pulldown and Inverse Telecine
Back when VCRs first came out, the movie studios were scared out of the minds that home video recording would kill the movie market, much like the RIAA is now about Napster and the MPAA is about DeCSS. But nevermind the politics, this is a technical article. Once the movie studios came to their senses and realized the tremendous market potential of this new medium, they had to figure out how to get movies onto video tapes.
So how does one turn 24 Frames per Second film into 59.94 Fields per Second video? The process invented to do this is called 3:2 pulldown or telecining. Telecining involves manipulating the film to turn it into a format which can be watched on a TV.
The first thing that is done is the film is slowed down by 0.1% to make it 23.976 Frames per Second. This is done because 29.97 FPS is 0.1% slower than 30 FPS. So from now on we will refer to the two rates as 24 and 30 for simplicity. Now comes the problem of how do we turn 24 FPS into 30 FPS? Those of us who have taken elementary algebra (which I hope is most of you) can see that common factors amongst the two numbers is 6 (24 = 6x4 and 30 = 6x5). This means if we insert an extra frame every 4 frames from the film, we will have 30FPS video.
There's a problem, however. This causes the video to stutter slightly
as we're basically duplicating a frame every fifth of a second. So what
can we do? Well we can take advantage of the fact that television is
interlaced, and manipulate the fields which make up the 5 frames we've
created. To do this, we alternate between two and three fields for each
frame that we output (thus the term 3:2 pulldown). If we have four film
frames, which we divide up into Odd and Even fields, we get the following:
Now lets interlace together the second and third frame of every series, to give us the following:
Here we can see what telecined video looks like. We've taken the second frame and stretched out its fields across two frames, while the even field of the first frame in the series stays around for an extra 60th of a second, and the odd field of the third frame does so as well.
This gives us an interesting opportunity - if we have a video source that has undergone telecining, we put it through a process to remove this, appropriately called inverse telecining. This basically reconstructs the 4 frames from every 5 to turn the source back into progressive video. This has many many advantages - most notably that you have less frames to store thus each can be given more bits (in the case of video codecs based on bits/second) or the whole file will take less space (in the case of video codecs based on bits/frame).
Here's an example of video before and after the inverse telecining process (or after and before the telecining process, if you want to call it that):
Notice that the B Fields from the second and third frame have been reconstructed into 1 frame and that has become the 2nd frame of the series. As you can see, inverse telecining dramatically increases the video quality when viewed on a computer monitor.
If you are using a codec that supports 23.976 FPS video (i.e. you're not editing using special hardware like a DC30/DV500 or something like that) I highly suggest editing in this format. It requires an extra step of preparation, but you wind up with smaller or better looking files at the end. Once everything is done, and you want to convert the video back into 29.97 FPS, it's quite easy to simply reverse the process on the final product. But you will end up with much more "film like" video and better image quality overall.
However, if you're editing in something like DV, which does not support any framerate besides 29.97 or 25, then you do not have this option.
Next time - The Ins and Outs of Video Compression