It was once accepted wisdom that LCD monitors were ill-suited to displaying moving pictures. That’s a quaint notion from this vantage point. LCD monitors have established a firm position as mainstream display devices and have significantly improved their video playback quality over the past few years, based on advancements in various technologies targeting improved picture quality. Of these, we will focus in this session on interlace-to-progressive (I/P) conversion (also called “deinterlacing”), intended to enable smooth display of moving pictures on an LCD monitor or television. This is a key technology, especially for users who have frequent opportunities to view film content on their monitors. Note: Below is the translation from the Japanese of the ITmedia article “Smoother video with cutting-edge technologies: LCD monitor interlace-to-progressive (I/P) conversion”published June 29, 2009. Copyright 2011 ITmedia Inc. All Rights Reserved.
The process of converting interlaced video signals to progressive video signals is called interlace-to-progressive (I/P) conversion. The term I/P conversion encompasses various methods and technological levels. The completeness of the conversion technology has a major impact on the picture quality of an LCD monitor or LCD television. Since some basic knowledge is essential to any discussion of I/P conversion, let’s consider video scan lines and scan methods.
As most of us know, moving pictures are displayed on a monitor by showing a progression of still images with gradually changing content, the same concept underlying a flip book. A single frame of video is split horizontally into thin lines, and the lines are traced on screen from top to bottom. In a display device, one of these separate lines making up a video picture is called a scan line. Looking at one of these lines in even greater detail, one sees that a single line is rendered by tiny rapidly moving points of light.
A standard-definition video using the NTSC method, traditionally used for analog television broadcasting, has 525 scan lines (480 effective scan lines) per frame. High-definition video for digital broadcasting has 1125 lines (1080 effective scan lines). Put another way, a standard-definition video is split into 525 components horizontally, while a high-definition video is split into 1125 components. Naturally, an HD video, with more scan lines, displays a clearer picture.
On an actual display device, the first scan line at the top of the screen is rendered from left to right. So is the second scan line, and then the third and subsequent lines. In this way, an image split into lines from top to bottom is displayed one line at a time.
Two methods, interlaced and progressive, are used to render scan lines. These methods differ in the order in which scan lines are rendered. Incidentally, in the not-too-distant past the term “non-interlaced” was used primarily in the computer industry. The non-interlaced method is roughly equal to the progressive method; the term progressive is more common today.
Ordinarily, by the interlaced method, a single video frame is transmitted split into two fields. The field used to transmit odd-numbered scan lines (1, 3, 5, . . .) is called the odd field. The field used to transmit even-numbered scan lines (0, 2, 4, . . .) is called the even field. The odd and even fields are transmitted in alternation and are displayed on the display device in alternating fashion. In other words, a pair of odd and even frames constitutes one frame of video. Under the NTSC standard, the field transmission speed is one field per 1/60th of a second. This means 60 fields (30 frames) of still images are rewritten per second—so fast that the human eye perceives them as a moving image.
Originally, interlaced technology was used to create high-resolution images by increasing the number of times the screen is re-rendered while minimizing the volume of data transmitted. Since this system was developed for cathode-ray tube television sets, which display pictures through scanning beams of electrons, it is in principle poorly suited to LCD monitors and LCD televisions, which can display fixed numbers of pixels at one time on a single screen. Current television broadcasts and DVD titles, among other content, transmit video by the interlaced method.
In contrast, the progressive method transmits and displays the first through last scan lines in order from top to bottom. Unlike the interlaced method, this method can display a single frame at once without splitting it into two fields. However, it requires greater bandwidth for transmission than the interlaced method. It also presented compatibility issues for the traditional NTSC method in the area of home electronics, which revolved around broadcast television. For these reasons, the progressive method was not used for a long period of time. (Broadcast Satellite and Communications Satellite digital broadcasts in many countries now use the progressive method.)
However, for computer displays, the progressive method has been the mainstream since the start of the 1990s, the CRT era. Given the nature of computer displays as devices displaying fixed pixels, the progressive method was ideally suited to LCD monitors and LCD televisions.
The interlaced method displays an image by alternating between two incomplete images, making it susceptible to flicker and blurring. These weaknesses can be especially pronounced on large screens. The progressive method displays one complete image per frame and achieves clear picture quality by minimizing flicker and blurring. Due in part to this advantage, products employing the progressive method have now been in common use for years in the field of computer displays, which are often used to view high-resolution images.
The LCD monitors and LCD televisions available today are compatible with audio-video input display video content based on the progressive method. However, the majority of video content, including analog television broadcasting, 480i DVD video signals, and 1080i terrestrial digital broadcasting signals, transmit information by the interlaced method. For this reason, these devices need to convert interlaced video to progressive video.
Here we’ve reached this session’s topic: I/P conversion. The majority of LCD televisions and displays currently incorporate I/P conversion technologies. While this technology first emerged in display devices, the number of playback devices featuring I/P technology has recently expanded. For example, playback devices compatible with progressive output and upscaled output of DVD video images (including consumer DVD recorders, Blu-ray Disc recorders, and the PlayStation 3) feature I/P conversion technologies.
In passing, simply filling in the gaps in an interlaced signal can sometimes make the picture look unnatural and prevent high picture quality even after converting it to the progressive method. Keep in mind that the systems used in I/P conversion and the precision of such conversion vary from product to product.
I/P conversion can be divided into two basic methods. The first is motion-adaptive conversion. The second is 2-3 pull-down conversion. These two I/P conversion systems differ at a fundamental level. It’s important to choose one based on the video source to be displayed. Neither can be said to be clearly superior. Playback devices and displays compatible with these I/P conversion methods automatically select the appropriate method. (Settings can sometimes be made manually as well.)
Let's start with a description of motion-adaptive conversion. The simplest I/P conversion method involves combining the odd and even fields of an interlaced video to create a single complete frame (inter-field supplementation). While this allows the conversion of a still image to progressive format with greater clarity, conditions differ when converting video images.
Due to the gap in movement between the odd and even fields of a moving picture, simply combining them will result in jaggy outlines and combing noise. While the time gap between odd and even fields is only 1/60th of a second, for the purposes of smooth video playback, this gap is significant. If the original source for I/P conversion is video, scan-line information making up the odd or even field is used to generate scan lines from the top and bottom to the middle of the screen, putting measures to reduce jagginess and combing noise ahead of rendering a single frame with high precision (inner-field supplementation). This is called motion-adaptive or selective blending I/P conversion.
In real-life usage, whether the original image to be subjected to I/P conversion is a still or moving picture is determined automatically. Inter-field supplementation is applied if the source is a still image (or a moving picture with very little movement). Inner-field supplementation is applied if the source is a moving picture. However, video content includes very few cases of still images and moving pictures with very little movement. In most cases, use of inner-field supplementation makes it difficult to obtain high-precision frames. Actual problems that arise include incorrect detection of whether the original image source is a still or motion image, picture quality degradation after I/P conversion, and unnatural movement. The capabilities of motion-adaptive I/P conversion have improved steadily in recent years.
This progress has been achieved by increasing the precision with which moving pictures are detected and inner-field supplementation is carried out and by performing inter-field supplementation for moving pictures. Here, every manufacturer demonstrates its own experience, expertise, and skill.
Now let's look at 2-3 (also written as 2:3) pull-down I/P conversion. This is used in the I/P conversion of a video source with a frame rate of 24 frames per second (fps). Primary examples of 24 fps video sources include motion pictures and animated films. Since today's LCD televisions and displays assume 30 fps or 60 fps video content, frame supplementation is needed to adapt 24 fps video for output on such devices.
The 2-3 pull-down method of I/P conversion first converts the first frame of a 24 fps video to two fields, the second frame to three fields, the third frame to two fields, the fourth frame to three fields (and so forth, continuing the same processing), then implements supplementation between the fields to convert the video to progressive format.
While this in principle should be a superior method, it can be difficult in practice to determine whether the original video source is 24 fps video. Mistakenly proceeding with 2-3 pull-down processing for a 30 fps video source can cause the video to appear to stop momentarily, since excess frames will be added. Recent measures have significantly improved the precision with which 24 fps and 30 fps sources are identified.
Additionally, 2-3 pull-down conversion is not perfectly compatible with LCD monitors and televisions. Since an LCD monitor or television features a 60 Hz refresh rate (the speed at which scan lines are rendered), its frame rate is 60 fps (or 60 fields per second). When displaying a video processed through 2-3 pull-down conversion, frames displayed for 2/60th of a second or 3/60th of a second can be mixed into the video, making it appear jerky and stiff.
For this reason, some newer LCD monitors and televisions have functions that display 24 fps video smoothly without 2-3 pull-down conversion. One example is Eizo Nanao's Foris FX2431 wide-screen LCD monitor, which is compatible with audio-video input. This product is capable of displaying 1080/24p images. It is based on a system that operates the display at 48 Hz and converts 24 fps video to double the original frame rate at 48 fps, then displays each frame of the video source for 2/48th of a second at a time. Since each frame is displayed for the same amount of time, movement appears natural.
To display 1080/24p video on a Foris FX2431, the video playback source must also be compatible with 24 fps output. However, many recent Blu-ray disc players and recorders are compatible with such output. Even the PlayStation 3 can output 24 fps video.