What exactly is HD?

Tim Carroll · February 24, 2006

I'm trying to figure out this HD thing and not having much luck. It seems like HD has a few sizes, but they don't make sense to me. I hear and see 480p, and 720p thrown around alot, (I understand this is the lower end of HD with 1080p and 1080i being the higher end). But what do those numbers mean exactly?

I at first assumed 480p and 720p meant footage that was 480 horizontal lines of resolution at 24 progressive frames per second and 720 horizontal lines of resolution at 24 progressive frames per second, respectively. But when I download 480p and 720p trailers from the Quicktime Movie Trailers site, they do not have that horizontal resolution. The resolution is lower than that. The only trailer that had 720 lines horizontal in the 720p clip was the one from IMAX which was a 4:3 ratio clip. Are they somehow figuring out how many pixels would be in a 4:3 clip with 720 lines horizontal and calling that total number of pixels 720p, so if you have a 16:9 ratio or something wider, you still can only have 691,200 pixels, so your horizontal lines would be lowered to compensate for more vertical lines?

Thanks,

-Tim Carroll

Stephen Williams · February 24, 2006

I'm trying to figure out this HD thing and not having much luck. It seems like HD has a few sizes, but they don't make sense to me. I hear and see 480p, and 720p thrown around alot, (I understand this is the lower end of HD with 1080p and 1080i being the higher end). But what do those numbers mean exactly?

I at first assumed 480p and 720p meant footage that was 480 horizontal lines of resolution at 24 progressive frames per second and 720 horizontal lines of resolution at 24 progressive frames per second, respectively. But when I download 480p and 720p trailers from the Quicktime Movie Trailers site, they do not have that horizontal resolution. The resolution is lower than that. The only trailer that had 720 lines horizontal in the 720p clip was the one from IMAX which was a 4:3 ratio clip. Are they somehow figuring out how many pixels would be in a 4:3 clip with 720 lines horizontal and calling that total number of pixels 720p, so if you have a 16:9 ratio or something wider, you still can only have 691,200 pixels, so your horizontal lines would be lowered to compensate for more vertical lines?

Thanks,

-Tim Carroll

Tim,

1920x1080 and 1280x720. Pal is 720x576 NTSC (I don't remember the first no) x 480

HD is 16x9 with square pixels, hope that helps.

Stephen

Matt Frank · February 24, 2006

All HD is in 16x9 as that is part of the standard.

720p is 1280x720 resolution with progressive frames.

1080 is 1920x1080 interlaced or progressive.

So if you download a 720 movie it should be 1280 lines wide by 720 lines tall.

I hope this helps.

February 24, 2006

Those are pixels, not lines of resolution. Not the same thing. Resolution refers to how many black and white line pairs can be seperated in an image. Other subjective matters enter in, too.

Actual resolution of a system depends on such things as bit-depth, ccd capture quality, lenses -- all that stuff and more. I'm not a super video tech, just ran across your question.

I do know that people buying the new HDV camcorders and whatnot will be very sorry to find that their images do not have the same "resolution" or look as the $100,000 real HD cameras used by the evil emperor Lucas. Which looks dead and lifeless, and even that isn't even up to super 16 image quality with today's awesome Kodak stocks. :lol: :lol: :lol: While consumer HDV looks like, well, better TV. Which has its place and is better than DV, I'll give it that. A hell of a lot better than DV.

Phil Rhodes · February 24, 2006

Hi,

Nyquist insists that the maximum vertical resolution of a 1080p HD clip is 520 lines. Less with anti-flicker interpolation, antialias filtering, lens mushiness, imperfectly-transparent air, noise, etc.

Phil

February 28, 2006

Hi,

Nyquist insists that the maximum vertical resolution of a 1080p HD clip is 520 lines. Less with anti-flicker interpolation, antialias filtering, lens mushiness, imperfectly-transparent air, noise, etc.

Phil

Ah, but in the 1930s Raymond D. Kell said it's about 66% of the line count with an interlaced scan :D

Basically he set up a special film projector that simulated interlaced scanning and asked volunteers to watch films and move their chairs around to the viewing positions they found most comfortable. After measuring how how far each person preferred to sit form the screen, he then used eye charts to determine their visual acucity at that distance.

He found that for most people the optimum trade-off between viewable resolution and interlace flicker occurred where they could just discern about 2/3 of the maximum possible resolution. This is known as the "Kell Factor" and is the reason NTSC has the apparently odd luminance resolution figure of 4.2MHz. They figured that any extra resolution was going to be wasted anyway as people wouldn't sit close enough to the screen to see it!

Werner Van Peppen · February 28, 2006

920x1080 and 1280x720. Pal is 720x576

that means that with 1080i the effective resolution is 1080/100 * 66 = 712.8 lines which begs me to wonder, how come the marketing machine is so good at making people drool over Interlaced HD.... One of the reasons progressive has a lower horizontal resolution is that there are no fields and therefore no jagged edges which make movement less sharp. (As i'm recollecting from my tired brain)

Filip Plesha · February 28, 2006

Hi,

Nyquist insists that the maximum vertical resolution of a 1080p HD clip is 520 lines. Less with anti-flicker interpolation, antialias filtering, lens mushiness, imperfectly-transparent air, noise, etc.

Phil

The Nyquist rule is easier to apply in audio sampling than image because, the resolution of

digital images is not uniform. It depends on the angle of sampled lines.

So 1080P can capture 540 horisontal lines yes, but not diagonal lines. It can probably capture closer to

400 or less diagonal lines.

That means that it can not capture true 540lp/mm in analog terms, but probably around 400.

This is just guessing, it takes a target.

I'm sure you understand this issue Phil because you are an expert on film scanning, but

always like to point out because time after time people keep confusing video lines/pixels with analog frequencies.

Phil Rhodes · February 28, 2006

Hi,

Yes, we're intrinsically talking about horizontal lines here.

Of course when factoring in how it resolves diagonal lines, you have to consider the finer horizontal resolution and how that impacts the ability of the thing to depict scenes. I believe it can be done as a vector calculation.

But this is all nuts anyway. Assuming a noise free CCD, distortion free lens, and atom-perfect camera alignment, all of which is of course impossible, you could put 520 horizontal line pairs on a 1080-line CCD and get a response. But put 500 line pairs up and you're going to get a very odd response indeed because it's going to alias like crazy and need a lot of filtering. This is rather like the 1980s cassette deck manufacturers who claimed a frequency response up to 28KHz, or something - and you actually found that it was reasonably linear up to something pathetic like 11K, then had several huge holes in it before an odd peak at 28.

Of course, we don't usually attempt to image horizontal lines, which helps.

Phil

Michael Collier · March 2, 2006

Ah, but in the 1930s Raymond D. Kell said it's about 66% of the line count with an interlaced scan :D

Basically he set up a special film projector that simulated interlaced scanning and asked volunteers to watch films and move their chairs around to the viewing positions they found most comfortable. After measuring how how far each person preferred to sit form the screen, he then used eye charts to determine their visual acucity at that distance.

He found that for most people the optimum trade-off between viewable resolution and interlace flicker occurred where they could just discern about 2/3 of the maximum possible resolution. This is known as the "Kell Factor" and is the reason NTSC has the apparently odd luminance resolution figure of 4.2MHz. They figured that any extra resolution was going to be wasted anyway as people wouldn't sit close enough to the screen to see it!

or maybe they reasoned people shouldnt sit that close and there fore they wouldnt. (also I dont think anyone writting the 1953 NTSC spec planned on 60" plasma screens) If only engeneers could see the future. This is why I am a big proponent of timetravel. Forget addiction to oil. Timetravel.