Aspect Ratios: 0.889 vs 0.9191

[Hass860]

Here's my confusion:

So I capture miniDV, and in gspot it says it's 720x480, with a PAR of 0.889, making it 4:3 (1.33), which makes sense.

But then in Sony Vegas, the NTSC DV template is 720x480 with a PAR of 0.9191, still saying it's 4:3, but mathematically that works out to more like 1.3786.

I don't understand this.

But then when I go to render to with the NTSC DV template, I get a final MPG (MPEG-2) that is the same as the miniDV, in gspot saying it's 720x480 with a PAR of 0.889 making it 4:3 (1.33)

I am so confused. Can someone help me understand this? What is the relationship between 0.889 and 0.9191? Am I doing something wrong in my process?

Thanks

- Hass

[poisondeathray]

There are different standards or governing bodies , one is the generic specification, the MPEG-4 specification, and the ITU-R BT.601 specification, with hundreds of pages of debate over which is "correct". They all specify something slightly different. So I guess it depends on what you're doing this for and who you're trying to please

Mathematically, what you are calling PAR, it should be 8:9, because 4/3 = 720/480 x 8/9 - so it balances the equation; but MPEG-4 specs call for 10:11 for 4/3 NTSC

They use 704width (704x480) for the calculation so 4/3 = 704/480 x 10/11

Not to confuse you even more, there are different terms, DAR, FAR, PAR, SAR, and each of the governing bodies can define them to mean different things. What you define as PAR for the w:h of the pixels is defined as SAR or sample aspect ratio in MPEG4 part 10 terminology (e.g. for blu-ray)

For MPEG2 specs (ISO/IEC 14496 Part 2), the offical list is reproduced here . If you are making a DVD which uses MPEG2, this is what I would use (i.e. 10:11 for 4/3 NTSC)
http://forum.doom9.org/showthread.php?p=685854#post685854

[Hass860]

Poison,

Thanks for the reply.

If PAR is 8/9, then what is the thing which is 10/11 called?

And if I'm bringing in 8/9 (DV-AVI) into a 10/11 project (Sony Vegas) and then rendering back to 8/9 (DVD MPEG2), is there some distortion going on?

[poisondeathray]

Where are you getting PAR 8:9 from? Gspot?

From experience, I can tell you that I used to use the mathematically correct ratios (and it filled everything perfectly and looked fine) , using the "proper " ratios tends to leave a gap. I haven't used vegas for this in quite a while, but Premiere does the exact same thing if my memory serves correct (i.e. uses the spec correct, but math incorrect ratios)

Your source is NTSC DV 4/3 DAR , but some would call it 10/11 some would call it 8/9 for the PAR ... that's the problem here

You can do a quick test render on a small clip with different settings to check it out. I don't use SD or DV anymore, so no more issues everything is square pixel

[Hass860]

Where are you getting PAR 8:9 from? Gspot?

From experience, I can tell you that I used to use the mathematically correct ratios (and it filled everything perfectly and looked fine) , using the "proper " ratios tends to leave a gap

Your source is NTSC DV 4/3 AR , but some would call it 10/11 some would call it 8/9 ... that's the problem here

I see. People calling the same thing by 2 different names. Even though it's numbers and math, so it SHOULD be the same...

[poisondeathray]

I just did a quick test on some 4/3 NTSC DV footage, and it all worked out fine using the preset values in vegas 9 , rendering to MPEG2 using the licensed mainconcept engine (i.e. no gaps)

So I think if you're consistent within the same program, and use the same values , it should work out ok

For the record, the 10/11 (or 0.9090...) is the spec correct value according to those tables, even though it's not mathematically correct...

[Hass860]

Ok good, so I guess I've been doing it right

You gotta admit- this stuff is confusing!!

And as well as you've explained it to me, I think I will be forever confused at least a little bit by it...

[poisondeathray]

There are oodles of pages of debate on this topic, and it goes into way more depth than I understand

But hey, as long as you don't get some weird letterboxing and it looks ok and works in the dvd player - that's the bottom line

[Hass860]

There are oodles of pages of debate on this topic, and it goes into way more depth than I understand

But hey, as long as you don't get some weird letterboxing and it looks ok and works in the dvd player - that's the bottom line

well I'd be more afraid of a squeezed or stretched image than of letterboxing or pillar...

[Hass860]

actually do you have a link to these pages and pages of debate?

[poisondeathray]

yeah that would be bad too.

vegas 9 does it correctly , same as input DV file

if you get some distortion, it would be an issue with your playback software or hardware

[poisondeathray]

actually do you have a link to these pages and pages of debate?

they are all scattered, some of them embedded as tangents in threads with completely different topics;

there's a whole bunch on doom9, but their search engine is a bit clunky unfortunately

I should have bookmarked them in hindsight, but some of the discussion was way too technical with spreadsheets and graphs....LOL (these discussions pop up every so often, and get very heated sometimes - sometimes very amusing)

if you check here, there are links to a few pages at the bottom of some of the discussions
http://avisynth.org/mediawiki/Aspect_ratios

But I'm still looking for that monster thread where there were charts & graphs , comparing ITU-R BT.601 (aka CCIR 601) vs. some other standard etc...

[Hass860]

crazy

http://lipas.uwasa.fi/~f76998/video/conversion/

[poisondeathray]

Good one! That's what I was talking about. A little too much technical stuff for me, but I'll bookmark it for the future

[redwudz]

Hass860, I've changed your topic title so members would know what you are asking about.

Moderator redwudz

[Alex_ander]

I am so confused. Can someone help me understand this? What is the relationship between 0.889 and 0.9191? Am I doing something wrong in my process?

As for GSpot, it uses actual parameters of file (resolution + display AR: mpeg2 file is flagged either as 4/3 or 16:9) for calculation: 4/3:720/480=0.8889. The other version uses the known fact (coming from ITU) that only a 704x480 part of that 720x480 image represents 4/3 AR, when displayed in correct proportions. By trivial proportional correction (720/704) we can get: 0.8889x720/704=0.9091

It is important to understand that both numbers describe just ONE possible interpretation of a digitally stored image (mentally divided into pixels like rectangles or squares). There is NO term 'pixel' in the basic ITU601 document from which ALL those numbers were derived later by programmers. ITU 601 describes Samples, not pixels. No one watches video on a monitor displaying as many pixels as the stored number (like 720x480) of digital samples: a digital display has its own, internal pixel count.
ITU standardized some numbers for digitally sampling analog signals with NTSC and PAL timing:
sampling frequency 13.5 MHz (both 525/60 and 625/50 systems),
number of samples per digital line: 720
number of digital lines: 480 and 576

From analog timing it appears that 720 samples represent not only visible part of analog line, but additionally - parts of horizontal blanking intervals. The border of a 4/3 analog image corresponds to 711 samples for NTSC (originally 485 active analog lines are cropped to 480) and 702 for PAL (575 active analog lines). In digital interpretation 4/3 (or 16:9) are represented by 704x480 or 704x576 samples (outside ITU 601 usually called 'pixels'). Since 'PAR' are totally dependent of more basic (usually known or directly given) numbers, you don't have to remember them at all, fewer chances to use a wrong number from wrong table at manual calculations. Neither a DVD player needs those: an ITU compliant device can restore analog samples with original timing by simply reading from memory with standard 13.5 MHz clock frequency (half of it in case of 352x480/576, VCD etc. - no black borders then).