Our current television system, designed in the 1930s, presents 60 pictures a second, each painted by a scanning electron beam lighting up the picture-tube phosphors from left to right, one row at a time, from top to bottom. Our system broadcasts 480 such scan lines of picture information over the face of the picture tube. This number of lines was chosen to be invisible up to a picture size measuring 19 inches diagonally. The rate of 60 pictures a second was chosen to provide for bright flicker fusion; i.e. you can make the picture bright without noticing the flicker inherent in scanning the picture over and over again. (The European system which presents frames at only 50 cycles/second requires dimmer pictures for the flicker to remain unnoticeable.)

Electronic technology of the 1930s was not capable of supporting enough information to paint 60 pictures a second, each containing 480 scan-lines. So the engineers invented interlacing as a work-around. Instead of broadcasting 60 full "frames" of picture each second, our television system broadcasts 60 "fields" a second, each field containing only half the scan lines.  Which half?  It alternates. One field transmits the odd numbered scan lines, the next field transmits the even numbered ones, back and forth. Two adjacent fields are considered to be one frame; therefore 30 frames are transmitted each second. The eye still sees 480 scan lines of information and the picture still updates at a sufficiently high 60 Hertz (cycles per second) to avoid flickering. In other words, the eye is fooled into thinking it sees 60 frames , not fields, each second.

Interlacing only fools the eye when there's little or no motion.  With significant motion, adjacent fields no longer line up perfectly and the eye perceives the individual 240-line fields. The apparently fewer scan lines across the picture become more noticeable. You can spot this yourself when viewing action scenes on any decent TV set. 

Today you might think to ask, why not just transmit 30 full frames of information each second and have the TV set show each frame twice to achieve the 60 hertz update rate?  Something similar is currently used in the better European sets to improve brightness.  Each of the 50 hertz fields is presented twice to create 100 fields/second, greatly decreasing the flicker sensitivity of the eye to a bright screen.  But these "100 hertz" sets require a fast digital memory to store each field for repainting a second time, and digital memory was hard to locate in the 1930s. 

It WAS available, however, in the 1980s, when a new type of TV briefly appeared called IDTV - improved definition television.  It did store two adjacent 240-line fields and then combine them into a single 480-line frame, which it presented twice.  It turned 60 fields a second into 60 frames a second.  Can you guess what was wrong with this scheme?  For relatively little motion this worked fine, but with significant motion the two adjacent fields did not line up.  Each frame superimposed two different 1/60 second snapshots of the motion.  IDTV was a flop.  And frankly, with direct view sets up to even a 35-inch diagonal picture size, scan lines weren't very objectionable anyway.

Enter large front and rear projection TVs for home theater.  If it is of decent resolution and focus, their scan lines become obnoxiously visible (unless you sit far away, like in a sports bar, which is where they first caught on).  On big projection systems we'd like to fill in the information between the scan lines, but IDTV showed that simply storing adjacent fields and combining them wouldn't work.  And HDTV with its greater number of scan lines had not yet been completely developed.The solution is line doublers and their progeny.

While our current system performs satisfactorily for direct-view sets with up to 35 inches of diagonal picture size, large screen rear and front projection displays blatantly reveal the line structure of the picture.  It's like looking at the world through leveler blinds. Line multipliers and scalars are very fast computers that generate information to go between those horizontal scan lines by elaborate and very fast guesswork. There is a unique feature of some line doublers called "inverse telecine".  This allows artifact-free line-doubling when watching filmed material.

Most home theater installations are primarily used for movies, and movies provide the potential to make line doubling simpler.  Movies are shot at only 24 frames a second - fewer than the 30 frames a second and 60 fields a second of standard TV.  Film is converted to video on a "telecine" which creates the 60 TV video fields from the 24 film frames every second.  Dividing 24 into 60 reveals that each film frame should take 2 1/2 video fields.  This is awkward, so the telecine uses 3 fields for one frame, 2 fields for the next, 3 fields for the next, and back and forth to get the same result:  1 second of film occupies 1 second of video. 

The frame sequence of fields would look like this:  (odd, even, odd), (even, odd), (even, odd, even), (odd, even), (odd, even, odd), etc.  where a parenthesis represents a single film frame, and the "odd" and "even" refer to the single interlaced video fields it is converted to, comprised of either odd or even scan lines, respectively.

Can you see a line doubling opportunity here?  Remember the simplistic IDTV concept where two adjacent video fields were simply combined to make a single frame, which was then presented twice at the 60 cycles per second field presentation rate?  That didn't work with motion because each video field represented a different snapshot in time, and therefore didn't combine cleanly with its neighbor. However with filmed material that has gone through the telecine process, 2 video fields, then 3 fields, etc. are taken from the same snapshot in time - a single frame of film. 

If a smart line doubler would recognize a film source at 24 frames per second, it could recombine video fields belonging to the same film frame to create a 480-line progressively scanned picture with no artifacts. To do this it would have to recognize which video fields belonged to which film frame; this means it must recognize which film frames were converted to 3 video fields and which frames were converted to 2.   In the three-field frames, the third field is a repetition of the first, and is therefore ignored. 

Recombined frames:  (odd, even, odd), (even, odd), (even, odd, even), (odd, even), (odd, even, odd), etc.

Now we've reconstructed the 24 film frames representing a second's worth of motion, each at full 480 line resolution.  This is called "inverse telecine."  If we then displayed the film frames at the video rate of 60 cycles/second, we would more than double the speed of the film presentation!  What to do?  If the line doubler showed each reconstructed film frame twice, that would still give us only 48 frames and we'd still be running fast at 60 cycles/second.  Showing each film frame 3 times would give us 72 frames - longer than a second's worth of video, now making the displayed film run too slow .  The answer: a 3X, 2X, 3X, 2X ... alternating display of successive film frames at 60 cycles per second restores us to getting a second's worth of video from a second's worth of film.  This is called "three-two pull-down", and it looks like this.

Recombined Frame Sequence (1 frame presented every 1/60 of a second):  (odd, even, odd) X3, (even, odd) X 2, (even, odd, even) X 3, (odd, even) X 2, (odd, even, odd) X 3, etc.

We get 60 full 480-line frames of video every second from the 24 frames of film.  There are no motion artifacts, and due to progressive scanning, all the phosphors of the television picture tubes or projector guns are lit every 1/60th of a second, brightening the picture. 

Inverse telecine with the accompanying three-two pull-down is thus an important feature for any line multiplier or scaler you consider purchasing to watch movies or filmed TV shows.