Resolution

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Resolution

Postby Andrew Davie » Wed Jan 24, 2007 4:43 pm

I have a question about the resolution of NBTV standard, in terms of 'pixels' as we are used to dealing with today.

Obviously the width is 32 scanlines, hence 32 pixels. The aspect ratio is standardised at 3:2, so this would suggest that the vertical resolution is 48 pixels (assuming square pixels).

Using the standards that I could find online, I tried to calculate the pixel resolution using the mechanics of the system...

The resolution, I expect, would be driven by the frequency of the input signal, and the ability of the light source to respond to that signal - and, of course, the 'hole' from the scanning device sweeping over this light source. I don't know about LED response, but I'd expect it is pretty rapid.

I see from the NBTV site the club standard is 2kHz - 10kHz signals. Given a 8kHz signal bandwidth, that would give 250 individual samples or pixels per scanline per second (8000Hz/32 scanlines). At 12.5 frames per second, that would be 20 pixels per scanline...?

On the other hand, using CD WAV files @44.1kHz that would suggest we could generate video to club frame rate (but not frequency bandwidth of course) at about 110 pixels per scanline.

The online test pattern images that I see appear for all the world to be drawn at a resolution of 32 x 60 (ref: http://www.nbtv.wyenet.co.uk/sound.htm)! I note the pixels in these images are drawn in a 4:3 aspect.

So, it seems to me that apparent resolution is somewhere between 32 across (defined) and vertically between 20 pixels (from the available bandwidth), or 48 pixels (using the published picture aspect ratio, assuming square pixels), or 60 pixels, from the test-pattern images, or 110 pixels (maximum using the 44.1kHz CD signal).

Commonsense tells me that it's actually 20 pixels vertical resolution, and these test patterns drawn at 32 x 60 are wishful thinking. Perhaps we get better apparent resolution due to persistance of vision, and the 'blurriness' introduced by the mechanical scanning merging pixels together, but by my calculations none of this can make it through the bandwidth pipe of 8kHz.

So, NBTV resolution at 12.5fps using 8kHz bandwidth is 32 x 20 pixels.
Can anyone correct my misunderstandings here?

Cheers
A
User avatar
Andrew Davie
"Gomez!", "Oh Morticia."
 
Posts: 1590
Joined: Wed Jan 24, 2007 4:42 pm
Location: Queensland, Australia

Postby old_tv_nut » Thu Jan 25, 2007 1:19 pm

Andrew, you missed a factor of two, since each cycle of 8000 Hz produces two pixels, a dark one and a light one, corresponding to the two halves of the cycle.

In addition, the number of pixels is determined by the upper limit of frequency, not the bandwidth between upper and lower limits; so, at 10 kHz, your calculation should lead to 50 pixels per line. In actuality, there is a bit more effective resolution, since the pixel locations are not on a fixed grid along the line dimension. You get about 20% or 30% better effective resolution as compared to a fixed series of spots. This is the inverse of the infamous "Kell Factor" discovered by Ray Kell in the 1930s.

The lower limit of frequency response has other possible effects such as streaking in the image if the DC clamping is not perfect.

- Wayne
old_tv_nut
"Fester, enough of the light-bulbs!"
 
Posts: 2
Joined: Thu Jan 25, 2007 12:50 pm
Location: Lake County, Illinois, USA

Re: Resolution

Postby ac7zl » Thu Feb 01, 2007 6:22 am

Andrew,

I'm a recent new member to the NBTVA, and I am very glad about the appearance of this new forum.

As to your comments/question, I found this verbage in instruction sheet for the eprom pattern generator that the club sells:

"The standard aspect ratio of 3:2 would suggest that the picture would be 48 pixels x 32 lines in size. However 64 pixels is an easier number in the digital world. As we are converting from digital to analogue before displaying the signal it does not really matter how many pixels there are in a line."


Andrew Davie wrote:I have a question about the resolution of NBTV standard, in terms of 'pixels' as we are used to dealing with today.

Obviously the width is 32 scanlines, hence 32 pixels. The aspect ratio is standardised at 3:2, so this would suggest that the vertical resolution is 48 pixels (assuming square pixels).

Using the standards that I could find online, I tried to calculate the pixel resolution using the mechanics of the system...

The resolution, I expect, would be driven by the frequency of the input signal, and the ability of the light source to respond to that signal - and, of course, the 'hole' from the scanning device sweeping over this light source. I don't know about LED response, but I'd expect it is pretty rapid.

I see from the NBTV site the club standard is 2kHz - 10kHz signals. Given a 8kHz signal bandwidth, that would give 250 individual samples or pixels per scanline per second (8000Hz/32 scanlines). At 12.5 frames per second, that would be 20 pixels per scanline...?

On the other hand, using CD WAV files @44.1kHz that would suggest we could generate video to club frame rate (but not frequency bandwidth of course) at about 110 pixels per scanline.

The online test pattern images that I see appear for all the world to be drawn at a resolution of 32 x 60 (ref: http://www.nbtv.wyenet.co.uk/sound.htm)! I note the pixels in these images are drawn in a 4:3 aspect.

So, it seems to me that apparent resolution is somewhere between 32 across (defined) and vertically between 20 pixels (from the available bandwidth), or 48 pixels (using the published picture aspect ratio, assuming square pixels), or 60 pixels, from the test-pattern images, or 110 pixels (maximum using the 44.1kHz CD signal).

Commonsense tells me that it's actually 20 pixels vertical resolution, and these test patterns drawn at 32 x 60 are wishful thinking. Perhaps we get better apparent resolution due to persistance of vision, and the 'blurriness' introduced by the mechanical scanning merging pixels together, but by my calculations none of this can make it through the bandwidth pipe of 8kHz.

So, NBTV resolution at 12.5fps using 8kHz bandwidth is 32 x 20 pixels.
Can anyone correct my misunderstandings here?

Cheers
A
User avatar
ac7zl
Mad Scientist
 
Posts: 62
Joined: Thu Feb 01, 2007 5:04 am
Location: Tucson, Arizona, U.S.A.

Resolution

Postby Klaas Robers » Fri Feb 02, 2007 3:12 am

Indeed. In order to get the same resolution vertically as horizontally, i.e. 48 x 32, we need a bandwidth of 2 Hz to 10 kHz. We can use more samples per line, e.g. 64 or even 128 or 256, but then we should see that no isolated white samples are present in between black samples. This resolution is then higher than comparable to 32 lines horizontally. So in this case (NBTVA EPROM picture generator) the samples are gradually changeing from black to white and vise versa.

For the Baird Televisor with 30 lines and an aspect ratio of 3 : 7 the highest frequency should even be 14 kHz.

Square holes give an extra low pass action. Only if you flash the light behind the Nipkow disc in very short pulses, then you could obtain real white and black dots in the vertical direction. When the light is continuously changing, e.g. following a sine wave form in time, then due to smear of the size of the holes you will get no real black spots on the line. In practice this doesn't matter too much.

This effect can be over come if you make very narrow slits in the disc. But this will give an even dimmer picture.
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands


Return to Mechanical NBTV

Who is online

Users browsing this forum: Google [Bot] and 36 guests

cron