Post processing of NBTV

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Re: Timebase correction...

Postby Steve Anderson » Tue Apr 29, 2008 10:39 pm

gary wrote:Just looking at this post again, and also to bounce a few ideas around, I was wondering why you think that lines have to be "inserted/duplicated or dropped"?..... that all of the line information should be there, just compressed or expanded in length due to variations in motor speed.


This is all very tricky, and I take on board both Gary's and Chris's comments. Where to start? Let's look at the simple ones first.

DC restoration, Chris is 100% correct, few monochome TVs employed it, even as a kid I wondered why things faded to grey and not black. With the advent of colour that had to change to at least retain some semblance of colour purity. However, DC restoration 'generally' works in FSTV due to the large difference in line rate and the LF response of the system, 15.625kHz to (say) 25Hz. A 625:1 ratio, in 400Hz NBTV that ratio is much less when using an audio circuit. Say around 5-20:1. In addition FSTV clamps the back-porch/black reference, not the sync-tips.

As for Gamma, I'll be blunt here. Within NBTV there's no reason for it I can see unless you're using a noisy (RF) circuit. Our detectors in a mechanical camera are generally linear nowadays, as are the devices used in mechanical displays. OK, luxeons need a bit more care in feeding, but the end effect is linear. I wouldn't call an audio CD or even an eight-bit .wav file exactly noisey.

Now, Graham, "Ancient Brit", has written a piece of display software that makes the presumtion that there are exactly 120 samples (I think) in a line. It doesn't use the sync pulses at all. So sources that are slightly off 'roll through'. It's here that is the focus of my attention on this matter.

If a camera is producing 400+ lines per second, then those additional ones (or parts of them, read headache) need to be dropped or interpolated before the input buffer fills up. The converse if the source is running slow.

All of this may be seemingly pedantic and unnecessary in a mechanical system. But if the source can be tidied up, it gives the display less gymnastics to do. With all the talk of being unable to syncronize for so many, perhaps it's not a daft thing after all.

A slight poke in the ribs here...If the motor in the camera were syncronous then none of this discussion would be needed.

...I'm gonna make a run for it...

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby chris_vk3aml » Tue Apr 29, 2008 11:44 pm

Yeah, sure, but do you know how much a 50 Hz eight-pole syncro motor capable of rotating a 16 inch Nipkow disc exactly at 750 rpm syncro speed COSTS?

Quote from Pacific motors here in Melbourne some years back was around $400 Australian (about 160 pounds sterling) and it would have to be specially made to order. OUCH! Shaded pole motors were cheaper - but they slip below syncro speed and often hunt up and down as the rotor poles align and de-align.

Six pole motors and four pole syncro motors slightly cheaper, but the bandwidth required for the video goes up proportionately with the rotational speed.

So I used sewing machine motors, Danny Gosson uses windscreen wiper motors. The later, being hefty dc motors with fairly good speed regulation are probably a better bet - but they AIN'T synchronous because it's something we just could not, and still can't afford!

Furthermore, after you record these signals on analogue tape, there are problems introduced by recorder wow, flutter, tape slippage and tape shrinkage. Digital recording is so EASY by comparison.

Sounds so simple - is so difficult. Infinite money would be a cure, but I ain't got that!

All the best,

Chris Long VK3AML
chris_vk3aml
 

Postby Steve Anderson » Wed Apr 30, 2008 1:27 am

chris_vk3aml wrote:Sounds so simple - is so difficult. All the best, Chris Long VK3AML


Agreed. I'm still wrestling with which way to go next, at the moment it's a toss-up between 96 and 120-128 lines. It's the signal recording/storage/playback that is giving me grief. The display in comparision is easy. This is not being helped by my requirement for YUV MAC colour compatabilty in the future.

I'm tempted to also place this on the 'back burner' and get on with something else until a moment of inspiration comes along. (Fat chance).

I have, quite literally, been pacing the floor up and down this afternoon debating with myself what to do next. (He was no help). Thankfully I have the place to myself during the day, otherwise the rest of the household would realise just how crackers I really am.

Steve A.

Venning, my butler after I asked for something to go with some wine:- "Your crackers my Lord." or even "Your nuts my Lord." and "How about something nice and cheesy?'. Hurrumph!
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby gary » Wed Apr 30, 2008 1:05 pm

chris_vk3aml wrote:So far as I am aware DC restoration is a post-mechanical TV concept.


I find that hard to believe, help me out here Stephen, Baird MUST have a patent on it?;-)

Actually, whilst DC restoration may not have been around then the need for it would have been - it is there as long as there is an AC coupled interface.

chris_vk3aml wrote:So far as I am aware DC restoration is a post-mechanical TV concept.
I am not aware of any 30-line receiving circuit published in the 1930s that even hinted at the usage of DC restoration. Many black and white TV sets of the 1960s and early 1970s had no DC restoration either. These simply could not do a 'fade to black', they always faded to grey! DC restoration only became absolutely essential when the mixing of three colour video components were necessary.


It depends on your definition of DC restoration (and I am including clamping in that definition), but all AC coupled video systems must have it to some extent be it in the form of DC Biasing of amplifiers or AC light sources, otherwise, on average, half of all frames would be black! It may not have been obvious from my previous post but the method I have been using is really just a small improvement over simple DC biasing, and can also be thought of as a differential unity gain block. This serves to get a mapping to a pixel for all samples but does nothing about the variation in contrast as picture content changes.

(I refer to contrast here but it could be just as easily brightness depending on how it is being handled)

chris_vk3aml wrote:As Gary says, "purely looxury"! In the 1970s, those of us working in NBTV really did live in the proverbial video "paper bag in a septic tank"!


That includes me by the way, in fact you can include the 1960s... ( I suppose that tags me as a nerd, since I seem to be one of the few people able to remember it... the 60s I mean)

chris_vk3aml wrote:I'm not sure, but I think DC restoration was another A D Blumlein/Marconi-EMI concept, like flyback EHT, but it certainly was not universally applied in the monochrome days. Even today, I have a cheap little portable 5" monochrome set which my wife and I take on holidays up country, and it very obviously has no DC restoration.


Yes I was aware of that fact but my understanding was that it's exclusion was a cost cutting exercise rather than it not being necessary. It seems to me that the pictures on these sets should have looked at little like when that horrible macrovision copyright protection scheme when it is invoked.

chris_vk3aml wrote:One problem with DC restoration lies in reproducing NBTV recordings with 'ringing', overshoot or excessive treble boost, where the restoration may actually lock to the dark tips of video overshoot - a very odd effect when DC restore is applied.


Absolutely, and this is the very thing I first came across when processing video through my soundcard, I found that the video can actually sag below the level of the sync tips! In this case the algorithm needs to be smart enough to realise it is not in a sync-pulse.

chris_vk3aml wrote:Like gamma correction, DC restoration is one of those things that is a 'nicety' but which one can really live without. I find 32 line pictures without aperture correction, or with streaky rasters due to using a round scanning spot without spot wobble (in CRT case) or via using Nipkow disc with undersized circular holes far more annoying. However it WOULD be nice if we had it.


Hmmmm, I can't agree entirely when I see people's noses poking out of a sea of grey...

chris_vk3aml wrote:The way I obtained DC restoration was simply to place a diode across my CRT grid to the CRT cathode, this would charge the video coupling capacitor feeding the CRT very approximately so that the modulation applied was always generally upwards from a black datum, but it was very approximate.


Indeed and that is pretty much exactly what I am doing in software, it's just choosing the time constant (your capacitor and whatever resistance is there) that's the difficult bit (when there are no syncs) - I feel that it should vary somehow with picture content.

chris_vk3aml wrote:Dear me, Gary, you must think that our mechanical scanners circa 1972 were built by Blumlein himself with the facilities of a major research laboratory!


No, just the ingenuity of a JLB, and the facility equivalent of a small attic over an artificial flower shop...;-)

chris_vk3aml wrote:Only joking, but I think you'll find that pictures derived from mechanical scanners will be ALL OVER THE PLACE in that regard, even today.

I am under no illusions there, in fact that is exactly why I am going to all the trouble. I know that it is a major problem for my software, but doesn't seem to be such a problem for mechanical (and probably analogue electronic) monitors so I trying to determine what the difference could be, but maybe simply adding a DC bias is sufficient for our purposes.

chris_vk3aml wrote:How do you INSERT the dc component into a constant sync pulse stream with a mechanical scanner, let alone RETRIEVE IT ???? !!!!


Well that may indeed be difficult, but I was asking about how to do it with sync-less streams - now THAT'S a knife!

chris_vk3aml wrote:Oh dear, dear, dear....


indeed, indeed, indeed

Thanks for the input, it helps to look at things from a different perspective sometimes even if it leads to the conclusion that it can't be done or is just unnecessary

Cheers,
Gary
Last edited by gary on Wed Apr 30, 2008 1:37 pm, edited 1 time in total.
gary
 

Re: Timebase correction...

Postby gary » Wed Apr 30, 2008 1:35 pm

Steve Anderson wrote: In addition FSTV clamps the back-porch/black reference, not the sync-tips.


Yes, but of course it doesn't really matter what the reference point is as long as there IS one, unfortunately in a lot of this old tape-based material there isn't any.

Steve Anderson wrote:Now, Graham, "Ancient Brit", has written a piece of display software that makes the presumtion that there are exactly 120 samples (I think) in a line. It doesn't use the sync pulses at all. So sources that are slightly off 'roll through'. It's here that is the focus of my attention on this matter.


Indeed, my software does that as well, and that is my point. The variations come down to pixels (in the digital domain) and if they are corrected the lines and frames look after themselves . So my intention would be to decimate or interpolate (according to whether the line is too long or too short - read line frequency is to slow or too fast) to a constant be it 120 or whatever. Because there are no reference points in a sync-less stream I guess this will come down to 'eye-balling' on a frame by frame basis using a fine line length adjustment provided by oversampling the video data.

Steve Anderson wrote:All of this may be seemingly pedantic and unnecessary in a mechanical system. But if the source can be tidied up, it gives the display less gymnastics to do. With all the talk of being unable to syncronize for so many, perhaps it's not a daft thing after all.


Exactly.

Steve Anderson wrote:A slight poke in the ribs here...If the motor in the camera were syncronous then none of this discussion would be needed.


That assumes a constant frequency which mandates a crystal locked power supply inverter or motor controller (depending on the type of synchronous motor) and it is debatable as to whether that is easier or harder to produce than a PPL closed loop motor speed controller, which reminds me I must get back to my stepper motor based monitor.

When I finally found a bicycle dynamo (they don't seem to be used anymore in Oz) I found I was just as annoyed with the bouncing up and down with mains frequency variations as I was with continually adjusting a non-locked motor.

Steve Anderson wrote:...I'm gonna make a run for it...


You can run... but you can't hide.

Thanks for the input.

G.
gary
 

Postby gary » Wed Apr 30, 2008 2:04 pm

chris_vk3aml wrote:Shaded pole motors were cheaper - but they slip below syncro speed and often hunt up and down as the rotor poles align and de-align.


I used a a record player motor back in the sixties/seventies. I didn't realise then that I could have turned them into truly synchronous motors by shaving a salient pole on the rotor. I still would have had the mains frequency variation problem tho'.

Stepper motors are now cheap (sometimes free) and CAN be pressed into service. I intend to report on this in the near future.

chris_vk3aml wrote:Sounds so simple - is so difficult. Infinite money would be a cure, but I ain't got that!


Well I can certainly afford things now more than I could in my teens but throwing money at the problem seems to me to be cheating, you might as well go out and buy a wide-screen. Making use of cheap readily available materials is 90% of the fun - very expensive in time tho'.

G.
gary
 

Postby AncientBrit » Wed Apr 30, 2008 5:35 pm

>Gary,

re Synching a pix with no syncs.

How about correlation on the data stream to look for a frame by frame match on content?

Probably would have to be a multi-pass process on the data.

(A little light reading on DSP is a dangerous thing!)

re DCR, probably the best you could do would be to extract average pix level (APL), (again multi pass), and set this to some arbitrary setting.


Regards,


Graham
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby chris_vk3aml » Wed Apr 30, 2008 6:06 pm

There are some misconceptions regarding the role of the DC video component in NBTV. Quoting Gary:

"It depends on your definition of DC restoration (and I am including clamping in that definition), but all AC coupled video systems must have it to some extent be it in the form of DC Biasing of amplifiers or AC light sources, otherwise, on average, half of all frames would be black!"

That statement that "half of all frames would be black" is incorrect. In a NON-DC restored TV receiver, the CRT grid is biassed at quiescent conditions producing a mid-grey luminance on the screen. When peak black and peak white are present in the video, even without sync, the signal simply modulates the screen luminance up and down fully, and the picture appears with a full range of grey tones if the average brightness (CRT grid bias) has been set at mid grey. The quiescent conditions for NON-DC restored reception are NOT normally set at BLACK, with the video modulation going down to blacker than black, and up only as far as mid-grey.

The only 'pictorial distortion' that a lack of DC restoration introduces will occur when the transmitted scene is either all-black, or all-white, with no excursions to full peak black/white in the image. In these cases the screen will reproduce a scene appearing mid-grey.

Frankly, I can live with this. The 30 line pioneers certainly did. And in the 1960s and early 70s, many of us had to, even with broadcast 625 line TV. Most of us were hardly aware of the effect, and only noticed it when colour TV came along to show us something better (1974-75 in Australia).

Baird's original 30-line video signals were fed via transformer-coupled program lines to transformer-modulated a.m. broadcast transmitters. No dc component was ever possible in the signal, or in the transmitted rf waveform. There were no sync pulses as we would know them. Rather, the picture was made very tall and narrow (3:7 ratio) so that a substantial part of each line length (between 15 and 25%) was masked off, effectively just to black level. Pictures with a white background would therefore leave a video signal which always had a predominant 375 Hz line tone, which, even without a sync separator, and even if they were sinusoidally interpreted - could actuate the phonic wheel correction motor on the receivers' universal motor shaft. That left the actual picture content with a tall picture of approximately 2:3 aspect ratio - the same as that of "modern" 35mm photographic slides and negatives. Lenses designed for that picture area and ratio were readily obtainable from old slide projectors or 35 mm cameras, and their image coverage area suited the size of a 32 line image on a Nipkow disc around 15 or 16 inches diameter. That was why Doug and others in the early NBTV days chose that particular picture ratio, being in keeping with the old Baird standard as well as suiting modern optics. The only major change was a switch to 32 lines per picture, to allow scanning discs to have their angles set by simple division; and for CRT's to have their horiz/vert sweep ratio set to 32 via binary divider circuits capable of locking over wide frequency ranges. Once that ratio was set, all that was needed for even MANUAL sync was an adjustment of line rate and picture phase. Not difficult, once you've got the hang of it.

Today, with inserted blacker-than-black sync signals, we don't need a picture as extreme as the old 3:7 ratio in order to recover sync. In fact that old system resulted in the 'throwing away' of many otherwise useful pixels of picture information, effectively demanding a greater transmission bandwidth than we now need.

I always intentionally avoided the transmission of sync pulses and the usage of dc restoration, especially when long r.f. paths on lower h.f. bands were anticipated. The problem was that selective fade and multipath would often throw ghosts or phase-shifted artifacts of such pulses into the picture area. Static crashes, having amplitudes many dB above average video level, would send dc restoration BERSERK - it was better just to set the CRT at grey level and have the CRT video level set to accommodate peak white and black, thereby ignoring the static crashes entirely.

There should be - there ARE - more intelligent ways of obtaining sync WITHOUT pulses on the basis of information previously known at the receiver. The 32 line raster and approximate line frequency are known: all that should be needed is manual control of PHASE to get the picture framed properly, line and framewise. And there is, mostly, a dominant line frequency tone in the video, which a narrow band filter probably could extract and use for sync without actual sync pulses. This problem was occupying the late Gil Miles at the time of his death, and its solution does not seem insuperable.

I think we should all be thinking far more "outside the square" of modern TV systems in pursuing this stuff - dc restoration, sync - the LOT. The pioneers certainly did, and a large part of this group's raison d'etre is in exploring new possibilities of older technology. If the frictional control of a thumb applied to the surface of a Nipkow disc can sync a picture - SURELY TO GOD we can come up with far better systems to sync via digital software.

Food for thought,

Chris Long VK3AML
chris_vk3aml
 

Postby gary » Thu May 01, 2008 11:54 am

chris_vk3aml wrote:That statement that "half of all frames would be black" is incorrect.

That statement maybe, but the full statement:
"but all AC coupled video systems must have it to some extent be it in the form of DC Biasing of amplifiers or AC light sources, otherwise, on average, half of all frames would be black!"
is not. The point I am making is there has got to be some form of DC restoration and in the case you mention the biasing this is just an example of DC restoration with an infinitely long time constant.

This is the naive first implementation one does when processing the AC coupled signal in software and my experience is that, due to the fact that the DC content of a video signal varies with picture content, the brightness and/or contrast distortion is objectionable. The reports that mechanical monitors present an acceptable picture has led me to the conclusion that there must be something else happening but perhaps it is just that there are so many other detrimental factors that it is less noticeable.

chris_vk3aml wrote:If the frictional control of a thumb applied to the surface of a Nipkow disc can sync a picture - SURELY TO GOD we can come up with far better systems to sync via digital software.


But you forget that controlling that thumb is the most powerful computer on the planet, it really is asking quite a lot to expect a simple piece of dsp processing to mimic it. Digital processing is a powerful tool but an understanding of the physical laws associated with the problem is essential, if you reduce the problem to an interface to a physical controlling device (i.e. a pot or joy stick) you still have to convert that information into something that makes sense to the data you have on hand, if you think that is easy I wish you would tell me how it's done, I've been trying to do it properly for 20 years ;-)

Thanks,

G.
gary
 

Postby AncientBrit » Thu May 01, 2008 5:26 pm

>All,

Would the subject of DSP processing (or at least some ideas on possible approaches) to synch a sync-less signal etc warrant a separate thread?

Or are we all spent?

I'm reluctant to continue hijacking this thread.


Regards,


Graham
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby gary » Thu May 01, 2008 5:51 pm

AncientBrit wrote:>All,

Would the subject of DSP processing (or at least some ideas on possible approaches) to synch a sync-less signal etc warrant a separate thread?

Or are we all spent?

I'm reluctant to continue hijacking this thread.


Regards,


Graham


Hah! I forgot what thread this was and I started it!

I must say this stuff is of great interest to me but probably not to many others (although they may be really interested in what emerges from it), I would certainly like the thread to continue if there is any further input, you never know, some real insight (maybe even incite ;-)) may come of it as from what I can discover the topic is not particularly well covered in the literature.

Is it possible to move all of the posts from when I started talking about massive oversampling etc, to a new thread entitled something like "Post processing of NBTV"?.
gary
 

Postby moderator » Thu May 01, 2008 8:38 pm

gary wrote:Is it possible to move all of the posts from when I started talking about massive oversampling etc, to a new thread entitled something like "Post processing of NBTV"?.


Yep, done.
Cheers
A
User avatar
moderator
Site Admin
 
Posts: 112
Joined: Wed Jan 24, 2007 2:40 pm

Postby gary » Thu May 01, 2008 10:45 pm

Thanks Andrew, very much appreciated.
gary
 

Postby AncientBrit » Thu May 01, 2008 10:57 pm

re Locking to video with no syncs.

I wonder if it might be possible to stand the problem on its head?

Rather than trying to match video line by line what would happen if we treated the video as noise and used correlation over a period equal to a number of NBTV lines in an attempt to extract the absence of video, ie the line blanking interval?

Would that work?

I've tried bog standard filtering in the past, the problem being that as video content changed the derived "line sync" moved in phase wrt video.


Regards,


Graham
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby gary » Thu May 01, 2008 11:38 pm

Sorry Graham, it's getting past my bedtime and my brain won't work, in what sense do you mean "line blanking interval"?

I do like the idea of using correlation however and will investigate.
Last edited by gary on Fri May 02, 2008 12:12 am, edited 1 time in total.
gary
 

PreviousNext

Return to Mechanical NBTV

Who is online

Users browsing this forum: Google [Bot] and 32 guests

cron