Compensating Images to Adjust For Video’s Non-Square Pixels

Posted on by Larry

[ This article was first published in the September, 2004, issue of
Larry’s Monthly Final Cut Studio Newsletter. Click here to subscribe. ]


This article grew out of last month’s article on “Square vs. Non-Square Pixels” describing the differences between computer graphics and television graphics.

One of the mailing lists I read regularly is the Apple Final Cut Pro Trainer’s list. Recently, I was fascinated by a discussion of what’s the correct size to create graphics on your computer that import into Final Cut with the correct proportions for video.

But, first, some background. The smallest element of a picture is the pixel (short for “PICture ELement”). On your computer screen, all pixels are square. However, on a TV, pixels are rectangles. This means that if you are creating graphics on your computer to display as part of a video program, you need to compensate for the differences in shape between the two pixel sizes.

If you don’t compensate for these differences, the circles you create on your computer will display as long, thin ovals on your TV set.

To make matters worse, Final Cut, in an effort to be helpful, adjusts for these pixel differences differently, depending upon whether you are importing a single layer graphic, such as a scanned photograph, or a multi-layer PhotoShop document. AND, with the release of Final Cut Pro HD, Apple changed the math behind how it calculates this difference in shape, which means that the old numbers we used to use may no longer be correct.

Into this discussion, Carl Jacobs, both an Apple Trainer and newsletter subscriber, wrote a really nice piece that describes what you need to know. I asked him to summarize the results exclusively for this newsletter.

Carl writes:

A little-known SMPTE Recommended Practices document from 1995 (SMPTE RP 187 -1995) both anticipates the Photoshop to FCP image workflow and specifies the SMPTE recommended pixel aspect ratio for NTSC 4:3 (.904) and 16:9 Anamorphic (1.205) Unfortunately, FCP uses different pixel aspect ratios of .888888888889 (8/9) and 1.185, respectively.

 

One can guess why Apple may have used different numbers (they were unaware of the SMPTE doc and made some logical assumptions of their own), but the implication for those of us who import Photoshop and other square-pixel images into FCP is that using Apple’s numbers based on FCP pixel aspect ratio will make images appear correct (a circle is a circle) within FCP but they will be slightly distorted when output to an NTSC monitor.

 

On the other hand, if we use the SMPTE numbers (or Photoshop’s and After Effects’ presets which use SMPTE’s numbers) then the image will look right on the NTSC monitor but will be slightly distorted in FCP and will not match other images created in FCP.

 

The numbers break down like this:

 

Video Format
Final Image size
SMPTE recommends
FCP recommends
DV-NTSC 4:3
720×480
720×531
720×540
DV-NTSC 16:9 Anamorphic
720×480
868×480
853×480
601-NTSC 4:3
720×486
720×538
720×547
601-NTSC 16:9 Anamorphic
720×486
868×486
853×480

 

For more info on Apple’s numbers (including the numbers for PAL video) go here:

 

http://docs.info.apple.com/article.html?artnum=36836

 

Before you tear out whatever hair you have left trying to figure out which numbers to use, keep in mind that most if not all NTSC displays and TV sets (which ultimately is what the viewer will see your work on) are probably not geometrically calibrated anyway and certainly not within the 2% difference between the two sets of numbers.

Larry again: Carl did a great job researching these numbers. However, I especially point out his last paragraph. There’s a point where the geometry (or lack thereof) of the home viewer’s set will have more impact than the size you create your images.

And a large thank-you to Carl for taking the time to write this up for all of us!


By the way, for those of you outside North America scratching your heads about this, SMPTE (the Society of Motion Picture and Television Engineers) is the official committee that determines the engineering specifications for NTSC video. There are similar governmental bodies that set the standards for PAL video. And, like NTSC, PAL video uses a non-square pixel. However, a PAL pixel is a different size than an NTSC pixel (which should, of course, surprise no one). It’s aspect ratio is 1:1.06.

 

Apple’s web site provides the necessary conversion numbers for PAL video. You may talk amongst yourselves to decide if Apple’s numbers are correct.


 


Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

Larry Recommends:

FCPX Complete

NEW & Updated!

Edit smarter with Larry’s latest training, all available in our store.

Access over 1,900 on-demand video editing courses. Become a member of our Video Training Library today!

JOIN NOW

Subscribe to Larry's FREE weekly newsletter and save 10%
on your first purchase.