Aspect Ratio and Digital Video
This page discusses how aspect ratio works in digital video and common problems associated with editing and playback of anamorphic video.
The role of aspect ratio in digital video has caused quite a bit of confusion, partly because there are actually three types of aspect ratio, not just one. The kind of aspect ratio most people know is the Display Aspect Ratio (DAR). This is the ratio of the width to the height of the display frame for the video, the aspect ratio of what we see. Typically, this ratio is 16:9 (widescreen) or 4:3 (full screen).
There are, however, two other kinds of aspect ratio that play an important role in digital video: the Pixel Aspect Ratio (PAR), and the aspect ratio of the stored data which we will call the Storage Aspect Ratio (SAR). We will try to clarify here how these three aspect ratios work and what role they play in digital video.
Anamorphic video and square vs. non-square pixels
When digital video is stored into a file or on a disc, it is stored with a particular frame size and aspect ratio, the SAR. If the viewing aspect ratio, the DAR, matches the SAR, then the process of the displaying the stored video is simply a matter of proportionally scaling it to the correct size. For this situation, DAR = SAR. An example of this might be a 16:9 display showing video stored with a frame size of 1280x720 pixels. Both have the same aspect ratio.
In other cases, the video may be stored with an aspect ratio SAR that does not match the display. Here the process of displaying the video involves distorting the stored video ratio SAR to make it match the desired viewing aspect ratio DAR. An example of this might be a 16:9 display showing video stored with a frame size of 720 x 480 pixels. The SAR is 720:480 = 3:2, an aspect ratio which does not match the 16:9 display. The stored video must be stretched horizontally or squeezed vertically to match the display correctly.
The latter situation is often referred to as "anamorphic" video. To correct for it, we introduce the third type of aspect ratio, the Pixel Aspect Ratio (PAR). The basic relationship between the three apect ratios is DAR = PAR * SAR.
In digital video, the pixels used on a display are considered to be square (i.e. with equal width and height). The pixels of the stored video are considered to be either square or non-square. If pixels are square, then the PAR is 1:1 and DAR = SAR as in the first case above. If the pixels are non-square, then the PAR is not 1:1 and acts as a correction factor for the SAR. Since DAR = SAR * PAR, in the second case above with a DAR of 16:9 and SAR of 3:2, the PAR is 32:27.
Standard definition vs. high definition video storage and compression
Anamorphic video with non-square pixels is more of an issue for Standard Definition (SD) video rather than for High Definition (HD) video. This is mostly because of the widespread use of SD compression formats such the DV camcorder formats and DVD discs. SD video for these formats is typically stored with a frame size of 720x480 pixels (NTSC video) or 720x576 pixels. These correspond to SAR values of 3:2 and 5:4 respectively. Since the video DAR display aspect ratio is either 4:3 or 16:9, the pixels are non-square and the PAR is not 1:1.
For HD video, the DAR display aspect ratio became fixed at 16:9. During the transition to HD, however, several "high-definition" compression formats were introduced for camcorders. These include formats such as DVCPRO-HD, HDV, HDCAM, XDCAM, etc. Because of data rate limitations, these formats may not necesarily store video in native 16:9 formats, but instead may use squeezed anamorphic formats. Examples include using 960x720 instead of true 720p (1280x720), and 1280x1080 or 1440x1080 instead of 1080i or 1080p (1920x1080). All of these formats therefore use non-square pixels with PAR pixel aspect ratios that are not 1:1.
More recent HD storage formats such as AVCHD, Blu-ray, DNxHD, ProRes 422, etc. all store HD video in the native frame size. In other words, 1280x720 is stored as 1280x720 and 1920x1080 is stored as 1920x1080. As a result, the DAR and SAR aspect ratios are the same and the PAR aspect ratio is 1:1. These newer HD formats are therefore not anamorphic.
Conversion of film and TV to digital video
The motion picture industry uses a variety of display aspect ratios, ranging from the nearly 4:3 "Academy" ratio of 1.37:1 used up until the 1950s, to extreme widescreen formats such as Panavision, VistaVision, or CinemaScope at 2.39:1 or even 2.67:1. Most modern films use a ratio of 1.85:1 which is somewhat flatter than 16:9 (1.78:1).
The digital video aspect of 16:9 was chosen as a compromise size to best accomodate the various widescreen sizes in use in the film industry. Conversion of film into 16:9 digital video formats is fairly straightforward. Since film aspect ratios are somewhat flatter than 16:9, in what is known as the "letterbox" process black bars are added above and below the image as padding during the digital transfer.
In the case of older films and TV broadcasts with aspect ratios closer to the "full-screen" ratio of 4:3, there are typically two methods used to convert the material into a 16:9 aspect ratio. The first method is called "pillarbox" and it involves adding black bars as padding on the left and the right of the image. This preserves the proportions of the original material, but at a cost of wasting a large portion of the 16:9 display. The second method is involves horizontally stretching the full screen material to fill the 16:9 display. This method distorts the proportions of the original content.
Although high-definition television has made 4:3 aspect ratio displays less and less common, conversion of widescreen film into full screen 4:3 aspect digital formats is another important problem. There are typically two methods used. The first is called cropping, and it consists simply of chopping off the left and right sides of the widescreen frame to make a full screen view centered in the middle of the widescreen frame. The problem with cropping is that often the edges of the widescreen frame show action that is important to the film.
To avoid this, a technique known as "pan and scan" is used. In pan and scan, the widescreen image is still cropped but the cropping region is not fixed. Instead, the center of the cropped full screen "window" moves back and forth across the widescreen frame so that it best follows the action of the film.
Playback and editing of anamorphic video
Anamorphic digital video poses a number of problems for editing and playback. Most of these are the result of the various container formats used to store digital video data.
The first problem is that some popular container formats do not support the required PAR pixel aspect indicator flags. Two obvious examples are AVI and DV Stream. When video is stored into these containers, only the frame size may be specified (giving the SAR). The PAR cannot be stored. Players and editors therefore have to guess as to what the correct PAR is to get the right DAR aspect ratio.
The second problme is that even if a container format supports PAR indicator flags, they may be incorrectly specified. There may also be several different indicator flags with inconsistent values. Players and editors therefore may or may not get the correct PAR and DAR aspect ratios.
The third problem is that even if a container format supports PAR indicator flags and they are encoded correctly, they may not be consistently supported by players and editors. This is because often the flags are poorly documented or not well standardized. As a result, one editor may import the video with one DAR, another editor or player may determine a different DAR, etc.
As an example of this last problem, consider QuickTime which supports true PAR flags (known as "pasp" data), but also supports alterate means of specifying the DAR using the "track matrix" as well as a specific "16x9" aspect flag for DV compression. The following table shows how various QuickTime players and editors behave with these flags.
|Application||Obeys pasp flag||Obeys track matrix||Obeys DV 16x9 flag|
|QuickTime 7 Player|
|QuickTime X Player|
|Apple Editing Tools|
|Non-Apple Editing Tools|
Conversion of anamorpic video into non-anamorphic video
In general, to convert anamorphic video into video with square pixels you render the video into a horizontally stretched or squeezed form that matches the expected display aspect ratio. This makes DAR = SAR so that the PAR becomes 1:1.
For anamorphic standard definition video, the frame size will typically be either 720x576 (for PAL) or 720x480 (for NTSC). To convert these into widescreen 16:9 non-anamorphic square pixel formats, we stretch them horizontally until the SAR becomes a 16:9 aspect ratio. For 720x576, the video will be rendered to 1024x576 with the horizontal size stretched by a factor of 1024/720 = 64/45. For 720x480, the video would normally be rendered to 853x720 with the horizontal size stretched by a factor of 853/720. Since 853 is odd and this stretch factor is a complicated irreducible fraction, a close approximation is to use a display frame size of 864x480 and stretch horizontally instead by a factor of 864/720 = 6/5. To convert to a 4:3 full screen aspect ratio, the video would be rendered to a 768x576 or 640x480 frame size by stretching or squeezing the image horizontally.
For anamorphic high definition video, we typically have camcorder recordings stored at 1440x1080 with a DAR of 16:9. To convert this into non-anamorphic square pixel form, we would render the video to a frame size of 1920x1080 by stretching it horizontally by a factor of 1920/1440 = 4/3.
|Anamorphic Frame Size||Display Aspect Ratio||Square Pixel Frame Size|
|720 x 480||4:3||640 x 480|
|720 x 480||16:9||853 x 480 or 864 x 480|
|720 x 576||4:3||768 x 576|
|720 x 576||16:9||1024 x 576|
|960 x 720||16:9||1280 x 720|
|1440 x 1080||16:9||1920 x 1080|
Best practices when working with anamorphic video
When working with anamorphic video with non-square pixels (PAR not equal to 1:1), here are some recommendations about how to get the best results.
If you are planning to eventually display the video on a TV by creating a DVD or a Blu-ray disc, it is best to maintain the video in anamorphic form throughout the editing process. Use a container format such as QuickTime that supports metadata to specify the correct pixel aspect PAR value. When importing the video into your editing software, make sure to set up your project to specify the correct display aspect ratio that matches your video.
If you are planning to eventually display the video on a computer or portable device or post it on the web, then we recommend converting the video into non-anamorphic square pixel form such that the PAR pixel aspect is 1:1. This is because computers, devices, and web players almost universally expect video to be non-anamorphic with square pixels. As much as possible, try to render the video into a frame size with a horizontal size in pixels that is divisible by 16 or 32. This will ensure optimal results for any compression you may do on your video, since nearly all video compression codecs will pad video to ensure a horizontal size that is divisible by 16 or 32.