Moving image formats

From Wikipedia, the free encyclopedia

You have new messages (last change).

This article discusses moving image capture, transmission and presentation from today's technical and creative points of view; concentrating on aspects of frame rates.

Contents

[edit] Essential parameters

The essential parameters of any moving image sequence as a visual presentation are: presence or absence of colour, aspect ratio, resolution and image change rate.

[edit] Image change rate

There are several agreed standard image change rates (or frame rates) in use today: 24 Hz, 25 Hz, 30 Hz, 50 Hz, and 60 Hz. Technical details related to the backwards-compatible addition of color to the NTSC signal caused other variants to appear: 24/1.001 Hz, 30/1.001 Hz, 60/1.001 Hz.

The image change rate fundamentally affects how "fluid" the motion it captures will look on the screen. Moving image material, based on this, is sometimes roughly divided into 2 groups: the so called film-based material, where the image of the scene is captured by camera 24 times a second (24 Hz), and the video-based material, where the image is captured 50 or ~60 times a second.

The 50 and ~60 Hz material captures motion very well, and it looks very fluid on the screen. The 24 Hz material, in principle, captures motion satisfactorily, however because it is usually displayed at least at twice the capture rate in cinema and on CRT TV (to avoid flicker), it is not considered to be capable of transmitting "fluid" motion. It is nevertheless continued to be used for filming movies due to the unique artistic impression arising exactly from the slow image change rate.

25 Hz material for all practical purposes looks and feels the same as 24 Hz material. 30 Hz material is in the middle between 24 and 50 Hz material in terms of "fluidity" of the motion it captures, however it is also handled in TV systems similarly to 24 Hz material (i.e. displayed at at least twice the capture rate).

[edit] Capture

The capture process fixes the "natural" frame rate of the image sequence. Moving image sequence can be captured at the rate which is different from presentation rate, however this is usually only done for the sake of artistic effect, or for studying fast-pace or slow processes. In order to faithfully reproduce familiar movements of persons, animals, or natural processes, and to faithfully reproduce accompaining sound, the capture rate must be equal to, or at least very close to the presentation rate.

All modern moving image capture systems use some sort of a mechanical and/or electronic shutter. Shutter allows the image for a single frame to be integrated over a shorter period of time than the image change period. Another important function of the shutter in raster-based systems is to make sure that the part of frame scanned first (e.g. the topmost part) contains image of the scene integrated over exactly the same period of time as the part of frame scanned last.

Early TV cameras (see Video camera tube) did not have a shutter. Not using shutter in raster systems may alter the shape of the moving objects on the screen. On the other hand, the video from such a camera looks shockingly "live" when displayed on a CRT display in its native format. See, for instance, the black and white PAL TV recordings of early Beatles performances.

[edit] Transmission

Analog broadcasting systems - PAL/SECAM and NTSC were historically limited in the set of moving image formats they could transmit and present. PAL/SECAM can transmit 25 Hz and 50 Hz material, and NTSC can only transmit 30 Hz and 60 Hz material (later replaced by 30/1.001 and 60/1.001 Hz). Both camps were also limited to an aspect ratio of 4:3 and fixed resolution (limited by the available bandwidth). While the wider aspect ratios were relatively straightforward to adapt to 4:3 frame (for instance by letterboxing), the frame rate conversion is not straightforward, and in many cases degrades the "fluidity" of motion, or quality of individual frames (especially when either the source or the target of the frame rate conversion is interlaced or inter-frame mixing is involved in the rate conversion).

[edit] 50 Hz camp

Material for local TV markets is usually captured at 25 or 50 Hz. Live broadcasts (news, sports, important events) are usually captured at 50 Hz. Using 25 Hz for live broadcasts makes them look like taken from an archive, so it is usually avoided.

24 Hz material from film is usually sped up by 4%. The sound is also raised in pitch slightly as a result. Recent technology allows an alternative where every 12th film frame is held for three video fields instead of two. Each of these techniques has its own drawbacks, but the former's are considered less objectionable. In the latter technique, the consecutive frames could also be intermixed with variable mixing ratio, which produced a different, though still objectionable visual artifact.

~30 or ~60 Hz material, imported from the 60 Hz camp, is usually adapted for presentation at 50 Hz by adding duplicate frames or dropping excessive frames, sometimes also involving intermixing consecutive frames. Nowadays, digital motion analysis, although complex and expensive, can produce a superior-looking conversion (though not absolutely perfect).

[edit] 60 Hz camp

Because of higher television production budgets in the US, and a preference for the look of film, many prerecoded TV shows were, in fact, captured onto film at 24 Hz.

24 Hz material is converted to ~60 Hz using the technique called 3:2 pulldown, which means, basically, inserting variable number of duplicate frames, with additional slowdown by the factor of 1.001, if needed. Occasionally, inter-frame mixing is used to smooth the judder.

Live programs are captured at ~60 Hz. In the last 15 years, 30 Hz has also become a feasible capture rate when a more "film like" look is desired, but ordinary video cameras are used. Capture on video at the film rate of 24 Hz is an even more recent development, and mostly accompanies HDTV production. Unlike 30 Hz capture, 24 Hz cannot be simulated in post production. The camera must be natively capable of capturing at 24 Hz during recording. Because the ~30 Hz material is more "fluid" than 24 Hz material, the choice between ~30 and ~60 rate is not as obvious as that between 25 Hz and 50 Hz. When printing 60 Hz video to film, it has always been necessary to convert it to 24 Hz using the reverse 3:2 pulldown. The look of the finished product can resemble that of film, however it is not as smooth, (particularly if the result is returned to video) and a badly done deinterlacing causes image to noticeably shake in vertical direction and loose detail.

References to "60 Hz" and "30 Hz" in this context are shorthand, and always refer to the "slow" 59.94 Hz or 60 x 1000/1001 rate. Only black and white video and certain HDTV prototypes ever ran at true 60.000 Hz. The US HDTV standard supports both true 60 Hz and 59.94 Hz; the latter is almost always used for better compatibility with NTSC.

25 or 50 Hz material, imported from the 50 Hz camp, can be adapted to 60 Hz similarly, by dropping or adding frames and intermixing consecutive frames. The best quality for 50 Hz material is provided by digital motion analysis.

[edit] Today's digital world

Computers and information technology have brought to light the understanding of the importance of presenting the moving image in its original capture format, as intended by the director (see article about purist); as well as the technical means to achieve this goal.

Digital transmission breaks the limitations of the analog transmission formats and the specific presentation mechanism (CRT display from 1940-s), and decouples constraints of capture process from those of presentation process. The MPEG format from its very beginning (MPEG-1) allowed encoding of a moving image sequence in its original aspect ratio, resolution and capture rate (24/1.001, 24, 25, 30/1.001, 30, 50, 60/1.001, 60 Hz). At the receiving side, the display is free to present the image sequence at the most suitable multiple of its capture rate, if it needs to avoid flicker. Most of the modern displays are "multisync": they can refresh the onscreen image at a rate most suitable for the image sequence being presented. E.g. a display supporting a range of vertical refresh rates from 50 to 72 Hz or from 96 to 120 Hz can display all of the standard capture rates without a non-integer rate conversion.

The politics, unfortunately, still get into the way of technology, and the 25/50 Hz frame rate options were demonstrably dropped from the ATSC format for apparently political reasons, even though ATSC is based on the MPEG; perhaps in order to complicate the transmission of imported material for its users.

Motion analysis is sometimes pitched as an ultimate solution to the problem of interchanging video material around the world. Motion analysis relies on a computer to detect the exact boundaries and motion vectors of every moving object across two consecutive frames. Although it sounds simple, this task, like many others from the domain of computer vision, can only be perfectly solved today by real humans. For example, in a broadcast of sporting event (like skating), converted from 60 to 50 Hz, when a person on the foreground moves quickly against a background, the person's head appears to "drag" the surrounding background along with it.

[edit] Presentation

There are two kinds of displays on the market today: those which "flash" a picture for a short part of the refresh period (CRT, cinema projector, plasma?), and those which display an essentially static image between the moments of refreshing it (LCD, DLP).

The "flashing" displays must be driven at at least 48 Hz, although today, a rate significantly below 85 Hz is not considered ergonomic.

For these displays, the 24-30 Hz material is usually displayed at 2x, 3x, or 4x the capture rate. 50 and ~60 Hz material is usually displayed at its native rate, where it delivers a very accurate motion without any smearing. It can also be displayed at twice the capture rate, although moving objects will look smeared or trailed, unless intermediate frames are calculated using the motion analysis and are not just simply duplicated.

The "continuous" display can be driven at any integer multiple of the capture rate - it won't matter for the viewer, nor can it be visually discriminated. It must be noted though that, in general, "continuous" displays show noticeable smear over quickly-moving objects in 50 and ~60 Hz video material (even if their response time is instant). However there are two emerging techniques to combat smearing of the video-based material in LCD display: it can be effectively converted into the "flashing" display by appropriately modulating its backlight; and/or it can be driven at double the capture rate while calculating intermediate frames using the motion analysis (see LCD television).

Obviously, when presentation rate is not an integer multiple of the capture rate, the "fluidity" of the motion on the screen will suffer to a varying degree (terribly for video-, unpleasantly for film-based material). This is usually the case with computer-based DVD players and PAL PC TVs, where the user does not switch the refresh rate either out of ignorance, or due to technical constraints; which sometimes are, in fact, artificial, made by manufacturers counting on that user's ignorance. For instance some laptop LCD panels cannot be (easily) switched to anything but a 60 Hz refresh rate, and some LCD displays with DVI input refuse to accept digital input signal if its vertical refresh rate does not fit between 58 and 62 Hz.

Most software DVD players do not assist with switching display modes, and even if it is switched manually, they do not hardly synchronize frame updating with display's vertical retrace periods (there is only soft synchronization using hardware double buffering, which is not enough to match hardware players in the stability of playback).

[edit] Recovering the original moving image sequence after a frame rate conversion

There is a lot of film-based material available today, which was "spoiled" by frame rate conversion in order to fit old analog transmission or presentation systems. It is sometimes possible, and makes sense to recover it to its original state before presenting to the viewer.

First, if there was frame mixing involved, it is most likely that recovery is impossible. This can also happen when there were several rate conversion steps involved, such as 24 Hz film converted to 60 Hz video, and then from 60 Hz video to 50 Hz video (a typical mistake and a typical example how it should not be done).

24 Hz material, converted to 25 Hz, can be recovered in a straightforward manner by slowing it down. Sound conversion is harder, because it is hard to guess if tone adjustment was used during initial conversion.

24/1.001 Hz material can be conversely sped up to 24 Hz. Sound is usually not a problem because of the small difference in speed.

24/1.001 Hz material, converted to 60/1.001 Hz, can be recovered using the reverse 3:2 pulldown technique.

Recovering video-based material usually does not make sense - it either looks good, or is spoiled for good.

[edit] 50 vs 60 Hz

60 Hz material captures motion a bit more accurately and "fluidly" than 50 Hz material. The drawback is that it takes approximately 1/5 more bandwidth to transmit, if all other parameters of the image (resolution, aspect ratio) are equal. "Approximately", because interframe compression techniques, such as MPEG, are a bit more efficient with higher frame rates, because the consecutive frames also become a bit more similar.

There are, however, technical and political obstacles for adopting a single worldwide video format. The most important technical problem is that quite often the lighting of the scene is achieved with lamps which flicker at a rate related to the local mains frequence. For instance the mercury lighting of the stadums (twice the mains frequency). Capturing video under such a lighting must be done at a matching rate, or the colours will flicker badly on the screen. Even an AC incandescent light may be a problem for camera if it is underpowered or near the end of its useful life.

The necessity to select a single universal video format (for the sake of the global material interchange) should anyway become irrelevant in the digital age. The director of video production would then be free to select the most appropriate format for the job, and a video camera would become a global instrument (currently the market is very fragmented).

[edit] See also

Retrieved from "http://en.wikipedia.org../../../m/o/v/Moving_image_formats.html"
In other languages