240, 405 & 625

[neon indicator]

since line rate is 10.125 on 405 and 15.625 on 625, then

15.625 / 10.125 to normalise the MHz for same horizontal resolution.

So 3.0MHz on 405 is same horizontal detail/resolution as 4.63MHz on 625 lines. For monochrome tube that's able to do higher detail. Which oddly is about the common "cheap" Digital resolution of SD of 544 pixels (DVB). The 720 pixel wide is less common now as they pack more channels in. DVD never uses the lower 544 pixels as it's not part of spec.

Stretch to 16:9 instead of 4:3 and you need 16/12 more bandwidth or pixels for the same horizontal DPI for same screen height as the 4:3 above. = 1.33:1

4.63MHz becomes 6.1MHz
544 pixels (DVB, not DVD) becomes 725 pixel.

So 720 pixel WS, 544 pixel 4:3, 6MHz WS(16:9) 625, 4.63MHz 4:3 625 and 3MHz 405 (due to less lines) on the same height mono screen are about same horizontal resolution.

Sadly DVB (not DVD) is often using 544 pixel for WS (16:9). But all is not lost. If you have a HD source (1920 x 1080) and resample to 544 x 576 for SD transmission, this is much higher quality than a 544 x 576 camera.

Decent DVDs are made by scanning over 2000 lines of the Film frame. Then after correction, the image is down-sampled to 720 x 576 with anti-aliasing. This is why with good up-scaling (on HDTV or BD player) the quality is very good, and why 50% of BDs are not perceptibly better than DVDs in blind trial. You'll notice there is a huge quality difference even on non-HDTV betweeen ITV (usually 544 x 576) and most THX / Lucas mastered DVDs.

So "raw" resolution of transmission isn't the whole story.




Many Colour CRTs are terrible in resolution, esp on smaller tubes. The a common dot pitch was I think 0.43mm for a GOOD crt.

Only larger sets could even show penultimate frequency grating on test card.

[murphyv310]

Yes I would assume you are correct. One thing to remember as the spot speed is slower on 405 a lower maximum frequency will give the same resolution as 625 as the equivalent higher frequency, if you get my drift.

[neon indicator]

Some related thoughts

http://www.techtir.ie/tv_radio/hdtv

http://www.techtir.ie/node/1003479

http://www.boards.ie/vbulletin/showp...21&postcount=2

[neon indicator]

Originally Posted by murphyv310

Yes I would assume you are correct. One thing to remember as the spot speed is slower on 405 a lower maximum frequency will give the same resolution as 625 as the equivalent higher frequency, if you get my drift.

In sense of
1) 15.625/10.125 ratio for line frequencies. 3MHz is same H detail as 4.63MHz
2) Phosphor decay time / ghosting on LCD etc. Any display has a response time. But that may not be the limiting factor.

On Colour sets the vertical slot, trinitron stripe or spot triad (older tubes), filter mask on LCDs etc limits the horizontal resolution to between 1/2 and 1/3 of native spot size.

Some displays use
R G
G B
pattern to increase the horizontal resolution. They may use twice as many lines, or just the same number of lines (which gives some degradation). Many cameras now use this pattern, except alternate rows are offset due to space filling Octagonal pixels.

[murphyv310]

Three interesting sites neon'.
I am of the opinion though and it's a personal thing that you don't get something for nothing. Upscaling will work with good results on a still image but from what I have seen moving images are not that clever.

[murphyv310]

Without having the usual CRT LCD argument, I would think we cannot compare 405 that was always used on a CRT as was 625 for years, now 625 and higher definition systems on LCD and plasma are just about universal. Remember since TV was invented all systems are done by scanning the image and with a CRT the screen is scanned by a deflected electron beam. The technology used in LCD and Plasma is different so we cannot have a direct comparison.

[ppppenguin]

Originally Posted by neon indicator

since line rate is 10.125 on 405 and 15.625 on 625, then

15.625 / 10.125 to normalise the MHz for same horizontal resolution..

Very strictly speaking it's the ratio of active line periods but that happens to be within 1% of the ratio of line frequencies.

What's more interesting is that we talk about pixels without much consideration of their shape. In any device using an electron beam the spot is approximately gaussian so the MTF will fall smoothly with rising spatial frequency. Tubed cameras nearly always had aperture correction to boost both H and V spatial response. This is different to ordinary HF boost because it's phase linear. Vertical correction needs 1 line delays so didn't happen until this was feasible in the 1960s. The VAK on some cameras such as the EMI 2001 was an optional extra. An LCD display, for example, has square pixels that are evenly illuminated. This corresponds to a "brick wall" MTF.

The closer we approach to brick wall functions the more likely we are to see aliasing effects. This can readily be seen using zone plate test patterns. These will ruthlessly expose any oddities in the MTF as well as any colour coding artifacts.

[neon indicator]

Very strictly speaking it's the ratio of active line periods but that happens to be within 1% of the ratio of line frequencies.

Quite right. But I thought likely a close approximation. I'd have have to look up the visible line period.


LCD may have square pixels or retangular, 3 or 4 sub pixels inline, or in group. They may not be "evenly" illuminated at all normal viewing angles. Going from CRT to Plasma, OLED, DLP, LCD or Reflective LCD each creates a different bunch of new issues.

Only CRT lets you have variable resolution, unless you have at least twice physical resolution of display (in both directions) than the highest H & V signal.

Also in horizontal direction you do need TWICE resolution of LCD /Plasma/OLED/LCOS/DLP etc compared to CRT unless it's a digital signal of matching resolution that can be exactly synchronised to same horizontal dot clock.

This is why a cheap "HD ready" 1366 x 768 LCD is much better than a 720 x 576 pure SD LCD (smaller SD are 720 x 480 to 840 x 480) and "SD" LCD on Analogue can look dreadful compared to CRT.


It all illustrates how going to 16:9 on SD rather than waiting for HD was a backward step and resulting in VERY soft pictures horizontally.

A 32" WS is about 24" visible 4:3? Thus equivalent to same height as 25" approx CRT?

For most people a 28" WS would be a drop down in quality due to limit of Digital Signal resolution, lack of WS on Analogue, and poor dot pitch of CRT.

[ppppenguin]

Originally Posted by neon indicator

Also in horizontal direction you do need TWICE resolution of LCD /Plasma/OLED/LCOS/DLP etc compared to CRT unless it's a digital signal of matching resolution that can be exactly synchronised to same horizontal dot clock.

I would agree that often appears to be the case with inexpensive TVs but the real problem is often with de-interlace and scaling. This is done in digital silicon and hence subject to Moore's law but still often done badly. For any given pair of input and display resolutions it is possible to get a good result. In practice the results are often miserable. Not usually on the H axis which is technically quite easy to get right but the V and time axes which are a lot harder, espeically where interlace is involved.

Even on the H axis there can be issues with reducing resolution. If the filter is too gradual the result will look soft. If the roll-off is too fast the result will look edgy and nasty. The ultimate compromise is subjective.

Originally Posted by neon indicator

LCD may have square pixels or retangular, 3 or 4 sub pixels inline, or in group. They may not be "evenly" illuminated at all normal viewing angles. Going from CRT to Plasma, OLED, DLP, LCD or Reflective LCD each creates a different bunch of new issues.

I agree that I was oversimplifying by saying that LCDs had square and evenly illuminated pixels. However on a decent LCD display the MTF should be a reasonable approximation to a brick wall function.

[murphyv310]

After reading the last few posts I have to say that I am baffled! MTF, Gaussing, square pixels & Moore's Law are things over my head!
To my simpleton mind if a LCD is fed with a clean Composite 625 line Video signal and can not give a semblance of a decent picture whether wide screen or not there is something wrong! The epitome of 625 line pictures in my mind occurred in the 90's when CRT technology was at its best, and even comparing the results with new sets available now there is no contest. I certainly don't want a CRT LCD bashing but as people have said to me in the past to prove my findings I ask you to disprove what I am saying!

[neon indicator]

With a analogue signal and many non-HD Digital signals it's easier to get a good picture on a CRT OR HD Ready/ Full HD "pixel based" screen than an actual non-HD LCD.

Even if the LCD is interlaced design. De-interlacing engines can be very good now and detect film source and then de-interlace perfectly. Also in theory any pixel based screen can display interlace mode, though if you are up scale number of lines you need to de-interlace.

Because horizontally the CRT or HDTV/HD Ready TV can display the information "perfectly". The non-HD LCD can only display the horizontal "perfectly" if the number of pixels exactly matches. 704WS, 544 WS, 544 4:3, 704 4:3, 720 4:3 are all seriously degraded on an SD 720 pixel Wide 16:9 WS compared to CRT, HD Ready or Full HD. Only a 720 pixel wide digital source can be displayed properly.

Thus really even if you NEVER watch HD, an HD Ready or Full HD is minimum to get "state of the art" CRT quality from 405, 625 analogue and all non-HD Digital sources on LCD, Plasma, OLED, LCOS, DLP and similar pixel rather than continuous raster scan technology (CRT, Optical mech poly mirror drums etc).

With Full HD LCD (42" LG is about £400) you are now in last year getting CRT quality at far cheaper price than Wide Screen CRT. Note that 28" 4:3 is about 36" for same size 4:3 picture on 16:9 TV. Given smaller bezel etc, a 42" HDTV doesn't take up much more height overall including case than 28" 4:3 CRT.

My conclusions?

405 line on a 12" screen is "high definition".
Non- HDTV LCD is a waste of money and much poorer than CRT
42" full HDTV with good up scale of SD content (old 1960s 405 line ITV episodes on DVD and 625 line analogue or SD Digital) is excellent now.
Ordinary VHS is really worse overall than 405 line, but S-VHS and 625line is "better" than 405 line.
S-VHS, old pre war films, 1960s B&W TV is all fine up scaled to 1080 line on 42", but ordinary VHS looks appalling that big. Why do no HDTVs have an un-zoom mode for poor resolution content so that it's like watching on a 17" 4:3 screen?

[BGmidsUK]

I wonder if any LCD panels are capable of displaying 405-line source material? This would be the only way of truly seeing what it would look like on a modern set. I suppose the nearest we can get is to use a computer to display such material upscaled to fill the monitor or TV screen.


Brian

[neon indicator]

Just up scale to 1440 x1080p "window boxed" into 1920 x 1080p

You can make some "synthetic" 503 x 377i files and process them according to best algorithms as it doesn't have to be real time.

You can take HD 1920x1080i content and downsample it without anti-aliasing and simulate an Orthicon 405 line camera.
Vertical
I'd actually maybe crop the 1080i lines to 754i lines and then just take alternate lines of each field to build the 377i.

Horizontal
I'd crop to 1440 pixels and then filter down to about 503 pixels by resampling in a manner simulating the Orthicon lag and electron beam / reading characteristics.

Doing a "proper" Downsample from HD to 405lines would give higher quality than any original 405 line source.

(I'd maybe cheat a little and go for 504 x 378i)

Then de-interlace the 503 x 377i (or 504 x 378i) and then interpolated upsample to 1440 x 1080p in a 1920 x 1080p frame

[ppppenguin]

Originally Posted by neon indicator

Because horizontally the CRT or HDTV/HD Ready TV can display the information "perfectly". The non-HD LCD can only display the horizontal "perfectly" if the number of pixels exactly matches. 704WS, 544 WS, 544 4:3, 704 4:3, 720 4:3 are all seriously degraded on an SD 720 pixel Wide 16:9 WS compared to CRT, HD Ready or Full HD. Only a 720 pixel wide digital source can be displayed properly.

In practice, a 1920x1080 source on a 1920x1080 display is not going to be pixel mapped. The digital TV standards (all of them, including 720 pixel SD) are intended to include margins round the edges to allow for muckiness due to image processing, blanking etc. Hence 52us of 625 SD (702pixels) fits nicely into 720 (53.3.us). At the receiver this means a few percent overscan to get what the broadcasters intended. A good scaling engine will do this more or less transparently. Many scaling engines are not good.Distinguishing the effects of poor scaling from compression artifacts and heaven knows what else but a big screen is always going to show up more than a small one.

Originally Posted by neon indicator

Doing a "proper" Downsample from HD to 405lines would give higher quality than any original 405 line source.

Absolutely. It's well known to the broadcast engineers that you get better 625 pictures by downconversion from HD than by direct 625 cameras and telecines. This presumes you have a good downconverter which would normally be true at the sending end.

An aside on the quality impact of trasnmission methods. All compression methods will have artifacts, the aim is to make them subjectively minimal. Interlace and PAL coding both have visible artifacts. Sometimes very visible. The results can sometimes be improved by newer technology, such as comb filter decoders. For example the inexpensive TI chip in the Aurora is truly excellent. In this respect I am sure that the Aurora delivers some of the best 405 pictures ever seen, better than the BBC's own converters used during the 405>625transition. Digital methods, such as MPEG have a different set of artifacts which many of us dislike. The visibility depends heavily on the coding algorithms at the sending end while the receiver can do very little. I am told that some coders are very much better than others. At an IEE lecture last week we were told that even the MP2 audio coding used for DAB, a standard that's widely criticised for being poor at lower bit rates, could be implemented a lot better if the makers of the coders implemented the best known algorithms. The lecturer didn't provide a reference to support his assertion. There's a claim made by this manufacturer: http://www.telos-systems.com/techtalk/mpeg/default.htm

This aside has rambled a bit far from the original topic and if anyone has any specialist information on this field perhaps we could fork this into a new thread.

[neon indicator]

Yes, MPEG4 has gone from about 1/1.8th of MPEG2 bitrate to nearly 1/3rd of MPEG2 bitrate due to better implementation (for allegedly same perceived quality). Though this may be an average on Statistical Multiplex & Encoding of several channels.

[BGmidsUK]

I use satellite receivers to feed my Aurora converters, these give very good results on BBC-1 and ITV due to the higher bitrate and more reliable signal as compared to the often horrible compression on DVB-T. DVDs recorded from satellite are best at high data rates as there is significant difference in the longer playing modes, in fact the half-resolution (LP, EP and EP+) modes are only really useful for time-shifting. I certainly wouldn't use these for archiving!


Brian

[Synchrodyne]

Originally Posted by ppppenguin

The whole business of vertical vs horizontal spatial resolution is a bit fraught. For the real world and old fashioned film, resolution is the same in both directions. AFAIK likewise for the human eye. So the aim for a TV system is to give equal subjective H and V resolution.

Somewhere I have seen it said that the eye has greater vertical than horizontal resolution, but I doubt that it was an authoritative source. Most discussions of eye resolution don’t mention the orientation, from which one might infer that there is no difference between vertical and horizontal. With film I suppose that anamorphic techniques resulted in a difference, and I think that the original CinemaScope system (2:1 horizontal compression and 2.66:1 aspect ratio) was at or beyond a reasonable limit in terms of horizontal resolution, hence the mid-1950s move to 70 mm systems. (As far as I know CinemaScope moved to a 2.35:1 aspect ratio fairly early on, but whether by reduced horizontal compression or cropping I don’t know.)

Originally Posted by ppppenguin

Some systems, such as 405, have excess H resolution. Others such as Belgian 819 are clearly deficient. It's pretty obvious that viewers hardly notice, within pretty wide margins.

That certainly seems to be the majority case. In that light the following recollection seems to be very contrary. In New Zealand, back in the monochrome days before the full microwave network was installed, getting the small number of network programs from Wellington to Auckland involved a several off-air links, with loss of video bandwidth. The drop in horizontal resolution as compared with locally-originated programs was noticeable, but I doubt that it was any worse was later found acceptable in the VCR era. Yet there were some complaints, letters to the editor in the NZ Listener.

Cheers,

[neon indicator]

Widescreen cinema uses three techniques, sometimes 1 2 or all three of them.

1) Panavision uses special barrel distortion lens on camera and projector. Anamorphic image as the image looks squashed horizontally. Vertical resolution is unaffected. Only one fixed aspect ratio per lens is possible and the lenses are big and expensive. Wider means less horizontal resolution.

2) Masking. On printing the the negative is cropped or masked. The projectors unlike TVs have variable masks. This means any aspect ratio is possible. Wider means less vertical resolution. This is why on 4:3 TV or VHS release of WS sometimes they can reduce the crop on sides and "open" top and bottom a little to get 4:3 frame. Compare "Willow" on DVD 16:9 and VHS 4:3, you will see more top detail. They call it "being artistic" but I think some shots are badly composed for Widescreen mask and excessive amount of top of head chopped of on closeup. Why on earth should HDTV or Cinema film have such close shots?

3) Different shape of frame of 35mm stock by different number of sprockets holes per frame. Note original 35mm film on cinema is similar to still photography "half" frame as a normal still camera gets a larger frame by rotating the film 90 degrees. Also there is then film stock which is not 35mm.

Other aspects is space taken at edge for optical soundtrack on prints originally, which means the top and bottom and one side need masked on projector to hide optical track and have same aspect as silent movie. Films moved from separate discs for sound to Optical track at edge to Magnetic and then separate magnetic reel. Now cinemas are changing to entirely digital (about 2000+ lines) and GPS locked Ultra HD cinema Satellite receivers for distribution as making film prints can cost millions per film, which is why releases staggered geographically in the past (to re-use same prints).

It's surely amazing that 405 was in theory equivalent about 500+ pixels wide (near square pixel on 378 line visible 4:3 aspect) and square pixel 625 line would be 768 x 576 for 4:3 and 1024x576 for 16:9, yet most digital broadcast of SD in UK & Ireland is 16:9 using 544 x 576. No wonder the reduced 1440 x1088 of most HD looks better even on a 32" 16:9 screen (equivalent approx 26" 4:3).
HD TX started at 1920x1080 (square pixel), but most channels are now 1440x1080 (anamorphic, one ITV region seems to be 1920x1080 on satellite).

Almost all TV content is still shot "safe" for 4:3 and SD, which certainly is why there are not enough long shots and too much in close and generally poor for HD Widescreen compared with "best" made for cinema. OTH really made for cinema WS with no thought of TV market at 2.4:1 aspect is like trying to watch Shakespeare via letterbox on far side of street on a smaller TV (less than 42", 60" would be ideal).

[Synchrodyne]

Thanks, Neon.

My understanding, hazy now, is that the original CinemaScope system was 8:3 (2.66:1) using a 2:1 anamorphic lenses and the standard 35 mm frame area, with four-track magnetic sound. But evidently few theatres were ready to equip for magnetic sound, and an optical track had to be squeezed in by reducing the frame width, which in turn reduced aspect ratio to 2.35:1. I had wondered whether this implied the use of a lower anamorphic compression ratio, but upon further thought that seems less likely than cropping. I suspect that anamorphic lenses were not made in a range of compression ratios. I can recall that back in the late 1960s/early 1970s it occurred to me that the then–current movies did not seem to be as “wide screen” as those I recalled from the 1950s when widescreen was new, which led to some research on the topic. By that time I suppose that most regular movies were around 1.85:1 non-anamorphic.

Cheers,

[Synchrodyne]

Originally Posted by murphyv310

We often criticise receiver design and blame them for poor pictures and during early dual standard days with only BBC2 on 625 there were many complaints about the Grey pictures! The receiver always got the blame, but I do remember the area engineer (Henry Chmiel) in British Relay telling me the whites were crushed for a long time until all of a sudden the problem was rectified. We used receivers that were designed for broadcast standard monitoring made by Murphy R&D, MR754 & 756, the 756 had true synchronous detectors but still the step was crushed on the VIT signal. So without a doubt this was a transmission problem and it did eventually get sorted.

Now that’s an interesting revelation; so the receiver makers were not entirely to blame for the “grey picture” phenomenon. Not-so-good practice on the part of the BBC apparently, which seems out-of-character for the period. It was still an implementation problem though, and not one that was inherent to the 625-line system as such.

Originally Posted by murphyv310

When I started this thread it was not about a debate on positive or negative modulation, both systems have advantages and disadvantages.

Regarding impulse interference UHF has made a big difference to that but also the legislation for the reduction in interference has made the car manufacturers get their act together, also mains borne interference from motors has reduced due to better suppression etc, so I doubt positive mod has that problem nowadays.

Nevertheless modulation polarity is probably germane to the issue, which is not just about the line standards but how they were implemented and how they were perceived. Had the UK opted for a positive modulation 625-line system, I somehow doubt that it would have suffered from crushed whites. In fact the positive vs. negative modulation issue was addressed some time back in this thread:
https://www.vintage-radio.net/forum/...ad.php?t=38458
the starting point for which was the assertion by Carnt and Townsend that positive modulation was much better for NTSC-type colour systems.

Perhaps a pertinent question is how in general did French viewers perceive the move from 819 to 625 lines in the pre-colour days. Maybe there were UHF reception issues, and surely some viewers noticed the loss of vertical resolution. But were there other areas of dissatisfaction? As well as impulse noise, the alleged difficulty of obtaining good black level AGC is often quoted as a strike against positive modulation. Surely though, the advent of solid state and then integrated circuitry would have eliminated or minimized that argument. That said, I remain very puzzled as to why the vision IF ICs designed for positive modulation systems (TDA2542 etc.) had only mean level AGC. (Maybe I’m missing an obvious point, but it seems to be dereliction on the part of the IC makers.)

Now back to a comment that I made in an earlier posting, had the UK opted for 625 lines either at the start of provincial expansion in 1949, or with ITV in 1955, then maybe it might have chosen a positive modulation system. In 1949, which was after the establishment of the OIRT 625-line standard but before the issue of the CCIR 625-line standard, positive modulation would have simply been an extension of established 405-line practice. In 1955, whilst negative modulation was clearly by then the majority choice for 625-lines, there was also the recent (1953) precedent of the Belgian 625-line system, plus the argument that dual-standard receiver design would be easier if both standards had the same modulation polarity. I understand that simplicity of dual-standard receiver design was why France chose positive modulation for System L. Although at around the same time it chose the negative modulation System K’ for its overseas territories where the dual-standard receiver requirement did not exist.

[QUOTE=ppppenguin;426592]

Originally Posted by neon indicator

For example the inexpensive TI chip in the Aurora is truly excellent. In this respect I am sure that the Aurora delivers some of the best 405 pictures ever seen, better than the BBC's own converters used during the 405>625transition.

Maybe what the Aurora delivers is beyond what even the best domestic 405-line receivers could fully appreciate? Short of a professional tuner and monitor combination, I wonder if the D.C. Read TV tuner design presented in Wireless World October 1975 ff could have been adapted to 405 lines. On 625 it had video bandwidth flat out to the full 5.5 MHz, with group delay correction as well. It was a pre-SAW filter design, with a block L-C IF filter that one imagines might be reconfigurable to suit the System A requirements.

Cheers,