is it normal that DVD CIF is really sharper than VHS??

[Bwaak]

Ah, I forgot about 4:3. Yes, I know that TVL is for a square picture. Thanks for reminding me, guys.

[Sharc]

Once digitized the frequencies are not relevant anymore; the stored PAL signal is 720x576 pixels (only 702/704 in reality, but that's another story) at a frame rate of 50 fields per second. The 720 pixel will then display on modern TV and Monitors as 768x576 in accordance to 4:3 DAR.

how it will suddenly get those missing pixels? - there is no such thing as 768x576 in PAL - total video line in PAL is 64us long, line blanking interval is somewhere around 12..11.7us i.e. active video line is between 52..52.3us and as pixel sample rate is 13.5MHz (i.e. pixel time 74.074ns) so 52.3us/74.074ns is approximately 706 pixels - there is no place for 768 pixels there unless you are using higher sample rate (around 14.436MHz) but then you are not ITU.

The luma sampling at 13.5 MHz results in non-square pixels. For geometrically undistorted viewing the picture has to be stretched horizontally by the Pixel Aspect Ratio (aka SAR Sampling Aspect Ratio) of about 59/54. The stretching is done by the player.
It's getting off topic though ......

[anonymoustly]

Hide there is also another problem/question, a black/grey screen/collor get greeny (green tint), how this can be?

vhs

[Attachment 71622 - Click to enlarge]
vhs "black" screen

[Attachment 71623 - Click to enlarge]
YouTube:

[Attachment 71624 - Click to enlarge]

[pandy]

it is square pixel concept, pixel aspect ratio and display aspect ratio relations. Extrapolated to full HD or higher resolution monitors.

This is your private square pixel concept - despite your own claim that "there is no pixels in analog signals" and we not talking about extrapolation(?) "to full HD or higher resolution monitors"

Square pixel concept may exist if you are doing some video format conversion but in VHS you have no square pixels, in S-VHS ditto - there is no resampling from 720 to 768... also 4:3 is in NTSC but PAL is rather 5:4 so once again 768 may don't work...

and frequencies are very important - digitized or not...

They do not exist anymore in the digitized frame.

So in other words you are saying that those video compression technologies based on DCT

H.261 - 1988
Motion JPEG - 1992
MPEG-1 Video - 1993
MPEG-2 Video - 1995
DV - 1995
H.263 - 1996
H.264 - 2003
VP9 	2010
H.265 - 2013
AV1 - 2018

are abusing your claim... but they exist and works quite well isn't?

[pandy]

Once digitized the frequencies are not relevant anymore; the stored PAL signal is 720x576 pixels (only 702/704 in reality, but that's another story) at a frame rate of 50 fields per second. The 720 pixel will then display on modern TV and Monitors as 768x576 in accordance to 4:3 DAR.

how it will suddenly get those missing pixels? - there is no such thing as 768x576 in PAL - total video line in PAL is 64us long, line blanking interval is somewhere around 12..11.7us i.e. active video line is between 52..52.3us and as pixel sample rate is 13.5MHz (i.e. pixel time 74.074ns) so 52.3us/74.074ns is approximately 706 pixels - there is no place for 768 pixels there unless you are using higher sample rate (around 14.436MHz) but then you are not ITU.

The luma sampling at 13.5 MHz results in non-square pixels. For geometrically undistorted viewing the picture has to be stretched horizontally by the Pixel Aspect Ratio (aka SAR Sampling Aspect Ratio) of about 59/54. The stretching is done by the player.
It's getting off topic though ......

This depends on your display and pixel numbers may vary - depends on display structure and only under assumption that you are using display with addressable technology... but then why bother with analog signal anyway?

[lollo]

Square pixel concept may exist if you are doing some video format conversion but in VHS you have no square pixels, in S-VHS ditto

That's what I said at the beginning. VHS is analog. When you digitize it you have pixels (and not frequencies as you were trying to highlight).

there is no resampling from 720 to 768... also 4:3 is in NTSC but PAL is rather 5:4 so once again 768 may don't work...

In the analog world originating the pixels, the PAR does not exist. Only the 1:33 aspect ratio.

In the digital world the display takes the digitized 720x576 frame and provides a 4:3 image, equivalent to 768x576 if mapping pixels 1:1. Because the original aspect ratio (coded as DAR) in the digital file (or other flags, but that's another story). 576 * 4 / 3 = 768.

So in other words you are saying that those video compression technologies based on DCT --- are abusing your claim... but they exist and works quite well isn't?

Once more, I am talking about VHS digitization, which adheres to ITU-R BT.601-4. Where at the end only a certain amount of pixels and a way to display them are present. No compression involved, it that helps you more.

And finally, I do not understand what you are trying to prove in this contest, rather than an attempt trying to show I said some wrong word or whatever. More or less the same to what happened here: https://forum.videohelp.com/threads/409707-Youtube-Workflow#post2691106

[Sharc]

Hide there is also another problem/question, a black/grey screen/collor get greeny (green tint), how this can be?

Misaligned White Balance, for example.

[pandy]

Square pixel concept may exist if you are doing some video format conversion but in VHS you have no square pixels, in S-VHS ditto

That's what I said at the beginning. VHS is analog. When you digitize it you have pixels (and not frequencies as you were trying to highlight).

Of course you have frequencies - DCT is frequency transformation not spatial pixel transformation - this is basic of signal processing - fundamental difference between analog signal and digitized signal (such as PCM) is discrete time and discrete levels (that's why you have Discrete Cosine Transform) - but frequencies are not changed - they are still there...

there is no resampling from 720 to 768... also 4:3 is in NTSC but PAL is rather 5:4 so once again 768 may don't work...

In the analog world originating the pixels, the PAR does not exist. Only the 1:33 aspect ratio.

In the digital world the display takes the digitized 720x576 frame and provides a 4:3 image, equivalent to 768x576 if mapping pixels 1:1. Because the original aspect ratio (coded as DAR) in the digital file (or other flags, but that's another story). 576 * 4 / 3 = 768.

So in your world ITU NTSC signals are downsampled to 640 pixels? ((480*4)/3=640) - PAL are always upsampled to 768 pixels and in NTSC downsampled to 640?

So in other words you are saying that those video compression technologies based on DCT --- are abusing your claim... but they exist and works quite well isn't?

Once more, I am talking about VHS digitization, which adheres to ITU-R BT.601-4. Where at the end only a certain amount of pixels and a way to display them are present. No compression involved, it that helps you more.

It is not about compression but in providing so many contradictory claims - you are brave enough to claim that after digitization frequencies don't exist anymore - i've read very carefully your message https://forum.videohelp.com/threads/409943-is-it-normal-that-DVD-CIF-is-really-sharper...HS#post2693142 and suddenly you starting introducing OT claim that every 720 pixels is upsampled to 768 - this not happening at all in existing digital devices so or your claim is based on some voodoo common for videophile community where some false claims exist happily from years or some knowledge unknown to me so i'm trying to see reasonable explanation - being electronic engineer i expect some knowledge not numbers justified in particular context but suddenly popped alone without context at all... this was my point. You started something that triggered my curiosity - that's all.

And finally, I do not understand what you are trying to prove in this contest, rather than an attempt trying to show I said some wrong word or whatever. More or less the same to what happened here: https://forum.videohelp.com/threads/409707-Youtube-Workflow#post2691106

What happened there? (few weeks ago btw), I simply said: don't waste time and energy trying to do something that is completely ignored by YT as YT will recompress your video anyway. Is that not true?

And I was not even aware of your reply to be honest (don't get this as i disrespecting you - simply many people on forum has different problems - if i can help then trying to help but this is not my life goal) so i expressed my point, didn't insulted you and anyone else, but also not trying to convince OP that i have monopoly for knowledge - this is not my goal - i wrote something and after this it is up to OP to use it or not - not my circus and not my monkeys - if someone feel more comfortable spending 6 hours more before YT will get his movie? - fine for me - i live long enough to understand that some people need to have this internal impression that if they spent more time and energy on doing something then it is for sure better - maybe it is even marginally better i.e. you improved something by 0.0027% but it costed you 20 times more resources - perfection is costly - sometimes just good enough is better - you may pursue perfection but my experience shows that this cost a lot, sometimes way more than you can get in exchange - as engineer i'm trying to provide objective: facts and knowledge, not go into personal comments and opinions - respecting you same as anyone on this forum.

And if you feel uncomfortable with ITU - fine for me - take Bt478 or 878 or similar card, reprogram driver or use ancient dscaler (source should be easily available) where more insights of Bt is exposed and simple capture video with samplerate supported by BT - you can even capture CVBS in grayscale mode (VBI) and later demodulate NTSC/PAL in software so you can perform fancy DSP if you wish.

All i can say - if you do compression with ultrafast preset and/or with CQP - like qp=4..8 and GOP=1..2 - your target video will be compressed very fast (almost similarly to lossless compression), filesize will be significantly lower than losslessly compressed, your upload time will be shorter and there will be no quality loss (or it will be marginal from audience perspective - you can even use PSNR to confirm that).

(btw why the heck you guys using CRF when uploading to YT? some even with placebo preset? - this is almost like banging wall with own head)

So there is nothing personal - honestly i don't even recall you till this clash - if you feel uncomfortable please accept my apologies.

[rrats]

Why not just find the HD copies of these tv shows, I am almost certain that, except for the 4:3 video, these all are cheaply produced sitcoms that my younger cousin used to watch like 5 years ago. I might be wrong but I still believe that there are better sources out here.

Y'know 16:9 video is not common occurrence on VHS, especially not for what appears to be a pure digital image rather than a telecine

[lollo]

being electronic engineer i expect some knowledge not numbers justified in particular context

I am an electronic engineer as well, and I realize that a 720x576 capture is displayed somehow 768x576 because 4:3 aspect ratio. You can play with words as long as you wish, the fact is there.

And if you feel uncomfortable with ITU - fine for me - take Bt478 or 878 or similar card, reprogram driver or use ancient dscaler (source should be easily available) where more insights of Bt is exposed and simple capture video with samplerate supported by BT - you can even capture CVBS in grayscale mode (VBI) and later demodulate NTSC/PAL in software so you can perform fancy DSP if you wish.

Smoke in the eyes to confuse the discussion concerning a specific aspect of the VHS capture. This is what I understand from your comment.

if you feel uncomfortable please accept my apologies.

I do not feel uncomfortable, and no apologies are needed. I appreciate your comments, because they force me to review my understanding.

But I have the impression we are talking on 2 different levels. No clash at all for me, and I am happy to having triggered your curiosity. I was just trying to explain to OP some of the theory behind the VHS capture, because I have the impression he was mixing numbers and pratical aspects.

[pandy]

being electronic engineer i expect some knowledge not numbers justified in particular context

I am an electronic engineer as well, and I realize that a 720x576 capture is displayed somehow 768x576 because 4:3 aspect ratio. You can play with words as long as you wish, the fact is there.

Once again - how signal is processed in display is up to display itself - some of them using non square pixels or even some complex subpixel combinations so i would be extremely caution to add some number that in real life may be different - if i use CRT then there is no 768 pixels there - PAL active video line is 52.3us and it may be displayed in analog way without upsampling, it may be resampled after conversion from digital to analog as analog by display ADC and finally it can be transmitted in digital way over HDMI or similar interface into display VDP - each case is different and in each time your claim about 768 may be incorrect.

And if you feel uncomfortable with ITU - fine for me - take Bt478 or 878 or similar card, reprogram driver or use ancient dscaler (source should be easily available) where more insights of Bt is exposed and simple capture video with samplerate supported by BT - you can even capture CVBS in grayscale mode (VBI) and later demodulate NTSC/PAL in software so you can perform fancy DSP if you wish.

Smoke in the eyes to confuse the discussion concerning a specific aspect of the VHS capture. This is what I understand from your comment.

nope - just saying that there are different possibilities to capture analog video signal and if someone doesn't like ITU (as being not square pixel) then it can do sample acquisition in non-ITU way - later additional processing can be applied, even TBC in software, also color demodulation/decoding can be done in software (that's why ages ago proposed to capture RF signal from VCR heads, some people on this forum was highly reluctant but im quite happy to observe almost 1000 messages thread about RF VCR capture - and no, i was not innovative as LD capture idea was born in parallel similar in time).
All i'm saying - if you don't like this or this you can do other way if you follow general physics rules. But there is no "smoke in the eyes" there.

if you feel uncomfortable please accept my apologies.

I do not feel uncomfortable, and no apologies are needed. I appreciate your comments, because they force me to review my understanding.

But I have the impression we are talking on 2 different levels. No clash at all for me, and I am happy to having triggered your curiosity. I was just trying to explain to OP some of the theory behind the VHS capture, because I have the impression he was mixing numbers and pratical aspects.

Perhaps there was misunderstanding as i'm trying to limit myself purely to main topic without introducing too many details (yes i'm aware that currently it drifted way more OT) - i could probably elaborate more and provide some complex documents but at some point this will not help too much for OP - that's why i stopped replying until seeing those two numbers - 720 and 768 - this triggered my curiosity as IMHO it is better to use DAR 1.33 than say that 720 is upsampled to 768 as it can be upsampled to 1440 or 2880 pixels or other value depends on unique display construction. And more or less it was all from my side.
Once again my apologies to all.

[lollo]

if i use CRT then there is no 768 pixels there - PAL active video line is 52.3us and it may be displayed in analog way without upsampling, it may be resampled after conversion from digital to analog as analog by display ADC and finally it can be transmitted in digital way over HDMI or similar interface into display VDP - each case is different and in each time your claim about 768 may be incorrect.

Come on pandy, I never made reference about CRT when talking about 768 pixels. Re-read my posts.

And more or less it was all from my side.

Here, you are joking , it was my concept since the beginning, with or without a reference to 768.

Once again my apologies to all.

Mine as well to having posted sligthly off topic. But I think the reading of both our posts was useful for other users...

See you in the next thread.

[pandy]

Come on pandy, I never made reference about CRT when talking about 768 pixels. Re-read my posts.

This was my problem - 768 was without any reference to anything - it was one liner with "720 px upscaled to 768 px" so i started to scratch my head and asking my self "did i miss something?"

And more or less it was all from my side.

Here, you are joking , it was my concept since the beginning, with or without a reference to 768.

Well - it was not clear to me how you see signal flow in analog vs digital (720 -> 768) so expressed my concerns on this.

See you in the next thread.

See you in the next thread.

[lordsmurf]

VHS sp 720x480???
VHS has only 240 lines NTSC or 288 PAL.

You're mixing up the axis.
VHS = ~240x480 (or more, or less, depending on factors)
CIF = 352x240

But the comparisons are entirely invalid, as both very obviously originate from different sources. There is obvious generational loss in the VHS (note chroma offset, ringing/halo), but the VCD is from clean source. A low-end VCR does not help matters.

However, that said, I see bad deinterlace loss in the stills, so "better" is arguable.

240x480???

Do you not mean: 480x240??

No.

The vertical (top to bottom) is fixed at x480 (240x2 interlaced)
The horizontal (left to right) is 240 analog lines max (and it varies). This is NOT the same as pixels, which are fixed palette. That said, it can be 240 pixels, or more, sometimes less. The equivalency is not fixed.

The "240" has confused video newbies for 25 years now.

Furthermore, it is important to understand that you need much more than ~340 digital pixels to sample and store that potential. Otherwise, this would result in aliasing – or blurring.

Yes Skiller, I did not want to enter into technical details with Nyquist–Shannon sampling theorem and Kell factor.

Sometimes video purists take Nyquist/Kell far too literal, and ignore other variables at play in the conversion. It's far more wibbly-wobbly than some believe. Quite a bit of processing happens in the VCR, the TBCs, and the cards in use. You have zero control over it. So setting VirtualDub/etc to 720, 704, 640, or even 352 doesn't have the effect you think it does. Nyquist/Kell is what engineers worried about, not you as end user of the gear.

That said, with 22tb drives now under $400, just capture lossless to 720x and move on. We don't need to compress specs to DVD/etc anymore.

[lollo]

Sometimes video purists take Nyquist/Kell far too literal, and ignore other variables at play in the conversion. It's far more wibbly-wobbly than some believe.

Nothing is too literal in science. It is what it is. Sure there are other variables in the digitizing process as you properly noted, which do not change anything, just add something.

Nyquist/Kell is what engineers worried about, not you as end user of the gear.

Well, we like to understand why something happens, and why the specs are what they are. Because understanding is necessary if you plan any modification or variant, or if you want to use the best approach, otherwise you have to stick to what is there, and maybe is not optimal for your purposes.

That said, with 22tb drives now under $400, just capture lossless to 720x and move on.

Yes, no need to speculate and search useless alternatives. But this does not contradict the above.