So what qualifies as 'High Definition' Video?

One basic criteria often referenced for high definition video is having at least four times the resolution of standard television (double the pixels horizontally and vertically). However...

A significant source of confusion for production professionals migrating into HD delivery is that the accepted picture sizes for digital television distribution vary and not all are “high definition” by any modern standard we could apply. In the United States, the ATSC or Advanced Television Systems Committee has issued “Table 3” of picture formats for digital television, specifying 640x480, 704x480, 1280x720, and 1920x1080 as digital video display resolutions. Both the 640x480 and the 704x480 formats have interlaced and progressive variations at multiple frame rates, with the progressive scan versions often referred to as “extended” or “enhanced definition,” the spoken shorthand being “EDTV.” (HighDef.org has the chart, as well as a simplified version here.)

The two accepted frame sizes for HD, 1920x1080 interlace/progressive, and 1280x720 progressive are 16:9 aspect ratio picture formats.

1280x720 is not 4x as many pixels as standard definition video (it even falls short of 3x), but it is used as a progressively scanned image. Most professionals would agree that progressive scanning does allow for the perception of very high detail with the tradeoff being progressive scanning does sacrifice some aspects of the smooth motion of interlaced scanning at an identical frame rate.

Advocates for the 1280x720 progressive format point out that while 1920x1080 images may be larger in size, when the 1080 frame is interlaced the picture delivers only half of the picture content every 60^th of a second, while “720p” can refresh the entire image every 60^th of a second. In this scenario then, 720p would deliver only slightly fewer pixels over time than “1080i,” and it would be temporally “sharper.”

The comparison and debate over spatial resolution (frame size) and temporal resolution (frame rate) and whether one trumps the other didn’t begin here and it won’t end here either. However, it's probably significant to note that one large stock footage library has ceased placing new footage acquired as anything less that 1080...and the trend for program format for an increasing number of broadcast and cable networks is to simply require a 1080i master with the stipulation that footage may be acquired as 720p (not mastered) with prior clearance... Many local stations will undoubtedly drag their feet on upgrading local origination to HD resolutions in the present economy, but mastering you content in HD maximizes shelflife.

Each professional needs to take a hard look at what sort of projects they do and what kind of delivery requirements are involved. It does look as if mainstream program distribution is definitely pointing toward 1080.

More General Knowledge...courtesy of Quantel

Quantel has an excellent glossary of DTV/HDTV terminology. I find myself searching for technical definitions often and this is one of my frequent stops...

Quantel Digital Factbook


TimK

Human Visual Acuity...Shoot it and they will...probably see it.

With all the carrying on about pixels in our digital imaging world, it's interesting that nobody has thought about upgrading the last step in the image viewing process...the human eye.

If you read up on the topic (or just Google it I guess...), you will find that human visual acuity is often described as "cycles per degree". To any broadcast engineers out there, this sound suspiciously like "line pairs per picture height"...and it's very similar. Humans have a given field of view...with each individual varying somewhat of course. Military pilots in many western countries are required to have a 147 degree peripheral range...unscientifically, I can see dark objects out to about 150 and bright tennis balls, flashes of light and signs that say 'beer' out to about 180 (though my wife thinks I have full 360 for signs that say 'beer').

Within this field of view, humans can only discern a fixed amount of 'resolution' or cycles per degree ( a 'cycle' being much like a line pair...one dark, one light alternating...at the point they become blurred and the light/dark turn visually to grey, you've exceeded the resolving power of the visual system). The images we watch on a screen can only be perceived with all their inherent resolution if they fill enough of our field of view to occupy enough 'degrees' to account for the image resolution.

The data on human visual acuity paints a picture that is...well, not as sharp as many people think.

Panavision's John Galt remarks in his excellent presentation in Panavision's "Demystifying Camera Specs" , Segment 4, that the visual frequencies they assess most closely to determine if a 35mm flm image is 'sharp' is "20-21 cycles per millimeter" which on a 24mm wide film frame (super 35mm stock, no sound track), he reminds us, comes out to be about 1000 pixels across. More precise (higher) frequencies are visible of course, but to a human viewer these are the 'make or break' aspects of image sharpness in this visual system.

(To understand more about the principles of how different visual frequencies are captured and retained...watch the whole series of videos. This is, in my opinion, the clearest explanation of what resolution and perceived sharpness really is that I've ever encountered...I have a permanent link in the list of links to the right... For some info on how the eye responds to different color frequencies, I find this article at Olympus Microscopy to be helpful)

As far as the maximum 'cycles per degree' attainable in the human visual acuity system, research tells me that is...somewhat unclear. I have seen information that quotes everything from 30 to as high as 60 cycles per degree...I did see that dogs apparently come in at about 12 cycles per degree (I've moved the dog bed in the living room closer to the television).

BobAtkins.com has an interesting article, focusing primarily on still photography, but I thought the graphic and the study outlined actually has bearing on our discussion. You can see that regardless of the maximum visible resolution of the human visual acuity system, our vision is at peak effectiveness at about 6 cycles per degree (moving my chair past the dog bed...).

So the interesting take away from much of this information is that our perception of 'sharpness' has less to do with peak resolution than we think. Someone with a 25" diagonal television image 10 feet away may actually have trouble discerning between 480p, 720p, and 1080p material all else being equal (and you thought that your cranky old neighbor was just too tight to buy an HDTV capable monitor). HDTV didn't catch on in Japan before the U.S. because everyone has bigger television monitors...it's because they have smaller houses.

As we all obsess over the pixel count in the sensors, video file codecs, and monitors we use...let's keep in mind that the human eye has its own limitations, and the television monitor or projection screen isn't the final stage in the visual chain.

...and since most of us produce material for humans...



TimK