December 14, 2004
Modern high-resolution or high definition displays advertise “pulldown removal” for optimal display of film content, such as that found on DVD. Just what is pulldown, where does it come from, and why would you want to remove it? Pulldown is a method that was created as a way to distribute film content in the world of video.
Modern film is projected at 24 frames per second. Video, in the U.S. and many other countries, is displayed at 60 pictures per second. So, when we want to convert film content to video, as long as we can come up with a way to repeat each film frame an average of 2.5 video pictures, our job is done. One way to do this is to repeat the first film picture 2 times, the second film picture 3 times, and so on: 2, 3, 2, 3. You might have heard of 2:3 or 3:2 pulldown — this is where those numbers come from. 2:3 and 3:2 pulldown are the same thing — they are not ratios, like they first appear to be.
Of course, it’s more complicated than that.
The 60 pictures per second shown in video are not really complete pictures. In fact, they are half a picture each. Video pictures are drawn with lines. About 480 visible lines make up a standard definition video image. Half of these — every other line — are shown every 60th of a second. The other half of the lines are shown the next 60th of a second. This display method is called “interlace” and it is an early form of video compression that was invented before computers had anything at all to do with television. It counts as a form of compression because it effectively cuts the required transmission bandwidth in half for a given image resolution. As with any compression system, it has its drawbacks. Since interlace trades vertical spatial resolution for temporal (frame rate) resolution, certain types of moving images or objects with lots of near-horizontal lines can create artifacts such as flicker, moiré patterns, or other distractions.
These video half-pictures are called “fields”, and two of them make up a video frame. So, pulldown actually repeats film frames into (alternately) 2 and 3 fields. The machine that transfers film to video is called a telecine and automatically creates the pulldown sequence during the transfer process.
Video with pulldown can be displayed directly on an interlaced CRT display, and in fact, that’s pretty much what you’re limited to doing on such a display. Ever since film has been shown on television, pulldown has been displayed on interlaced displays in this manner.
However, given the modern processing and memory available in a high definition display device, we can remove the pulldown and extract the original 24 frames-per-second film content. Once the film frames have been retrieved, they can be scaled up to the display’s native resolution without any interlaced artifacts. The result is a sharper picture.
Many high-end CRT displays, and all so-called “fixed pixel” displays (such as LCD, DLP, and LCoS) can display 60 full (non-interlaced) pictures per second. You might have heard these called “progressive” displays. When connected to a standard definition video signal, these displays can up-convert the interlaced signal to progressive. In the days before pulldown removal in displays, they did this by creating the “missing” lines from each field before it got displayed. They still do this for non-film content. The missing lines usually get created by blending the lines above and below. This is an estimation, an interpolation — so-called “line doubling“ cannot create content where none existed. Let’s take a look at the process:
Figure 1 shows two sequential fields from a scene where the camera is rapidly panning. You can see how the image has shifted quite a bit between each field, since the fields were captured 1/60 of a second apart in time. By the way, in these images, I’m going to scale down the image resolution from “real” video images by a factor of 5, so these 480 line images are only going to be 96 lines tall.
[click to enlarge]
Figure 1 – two sequential interlaced fields
If these images were from a typical video sequence, a progressive display would fill in the missing lines, creating the images in figure 2.
Figure 2 – two fields with the missing lines filled in (line doubling).
While this kind of progressive display is an improvement over an interlaced display, we can do much better if the content was originally film, and transferred to video using 2:3 pulldown.
Modern displays can detect the 2:3 pulldown sequence, and recombine the fields together back into the original full frames, and then display those. Figure 3 shows a full resolution frame after pulldown removal. This image has twice as much vertical detail as those shown in figure 2.
Figure 3 – full resolution after pulldown removal
Pulldown removal techniques first must determine when pulldown is being used, and where the 2 or 3 field repeats are. To further complicate matters, if video with pulldown is further edited (which often happens when trailers for films are edited for television), or when a film being broadcast cuts to commercial, the pulldown sequence may change or disappear and then reappear later. Furthermore, if there are scenes without much image contrast, or without much motion, pulldown detection may turn off, reverting to regular line doubling. Pulldown removal systems react to these situations differently, some better, some worse.
Examples of pulldown removal systems include Sony’s “Digital Reality Creation™ (DRC) Multifunction with CineMotion®” and Mitsubishi’s “Automatic Film Mode.”
For a high-resolution progressive display device, pulldown removal is a key technique to making film content on video appear the very best that it can.
Posted by Dithermaster | | Filed Under Display Technology