Pixel Aspect Ratio

[Cornucopia]

Sample AR is the official one. I only still use PAR because, 1. I used that for 2 decades before they changed it, so force of habit, 2. it is less confusing to others who might confuse it with that storage "monstrosity".

I will gladly start using SAR instead of PAR, if others will also stop using Storage AR.

Btw, @hello_hello, not a good idea to do multiple resizes. Each one adds blurring/interpolation.


Scott

[Ennio]

The term "storage aspect ratio" already confuses me, no matter where its definition may have been chiseled in granite. Considering how the --sar argument is meant, the x264.exe devs must be hellbent atheists then. Or am I confusing things already now?

You're not confusing things.

Okay, I'm lost. Again feeling confused As general gesture of good intention, I hereby offer my heartmeant thanks and apologies to everyone - hellbent or not - contributing to the matter.

Once upon a time:
SAR * PAR = DAR

Don't mean to sound pedantic, but even when SAR would be mistaken for PAR and vice versa, that formula would still hold clear waters.
But I don't want to confuse things even more here...

[hello_hello]

Btw, @hello_hello, not a good idea to do multiple resizes. Each one adds blurring/interpolation.

Twas just to help illustrate inserting the source and destination pixel aspect ratios into the resizing equation.

Once upon a time:
SAR * PAR = DAR

Don't mean to sound pedantic, but even when SAR would be mistaken for PAR and vice versa, that formula would still hold clear waters.

Yes it was a simpler time.... but now when there's no PAR in the equation SAR becomes ambiguous, and the aspect ratio police will express their displeasure when it doesn't mean sample AR....

[jagabo]

All you have to do is specify what each of your acronyms mean. Then there's no confusion. For example:

https://forum.videohelp.com/threads/404855-SAR-for-544x576-4-3#post2649082

[dellsam34]

I agree, we should stop using abbreviations and use full words or abbreviation followed by full text between parentheses like I wrote in post #6 of this thread.

[_Al_]

Or I would say, even better, just to explain what one wants to do from engineering point of view, like:

I have an image that I obtained exporting from virtual dub out of 1280x720 progressive video and I want to concat it to a mpeg2 for a DVD source,
something like that. If workflow is not provided, then it turns into this, academic discussion.

[Ennio]

Or I would say, even better, just to explain what one wants to do from engineering point of view...
...If workflow is not provided, then it turns into this, academic discussion

I agree. Once again being confused here about whether for years I have mistaken acronym A for B or not, doesn't really help. It's useless.
Many years ago, actually a simple constructor's visual approach made me get the logics behind the formula. Simply taking out paper and pencil, with some sketches and simple math the formula emerged by itself. I used clear phrases. No acronyms.

[_Al_]

Yep, same approach, grabbing a pen and just draw everything down. Things just start to move forward more easily after drawing a problem down on paper, because a solution or the way towards it just start to pop out of those drawings. I have a school background to that in the first place, it is amazing how faster things go then. One can remodel the whole problem yet again, by drawing something new in seconds, and again, if things do not move forward, while if trying to do that abstractly in one's head, leads nowhere, too many dead ends, impulses and connections are lost in wiring, whereas on paper it is still there and we can visualize it from multiple dimensions looking at it. In our head we usually approach a problem only from one direction.

[mark111]

Will get to full answer later (tonight, tomorrow), but your question just now is in error.
An image/video CANNOT be 5:4 "SAR", AND 1:1 PAR, AND 1.25:1 DAR! At least one of those numbers must be wrong.
Scott

FFPROBE says this:
Stream #0:0: Video: mpeg4 (Simple Profile) (DIVX / 0x58564944), yuv420p, 720x576 [SAR 1:1 DAR 5:4], 806 kb/s, 25 fps, 25 tbr, 25 tbn, 25 tbc

Likewise, your standard image/video of 4:3 DAR, AND 4:3 PAR, AND 720x576 resolution also CANNOT be true.

Scott

Yes, sure. This is a stupid typo when copying text from the text editor.
I mean an image 720x576 (i.e. W/H 5:4 (1.25) ) with pixel_ar=1.0667 and dar=1.33

[mark111]

New Width / Original Width * Original Height = New Height
is the basic equation for calculating resizing,

It would be more logical to write it as
New Width [1920] * Original Height [540] / (Original Width [1280] = New height [810].

because
New Width [1920] Original Width [1280]
x
New height [???] Original Height [540]

You posted above

New Width * Pixel Aspect Ratio / Original Width * Original Height = New Height

I asked how did you get that ?
You answered

New Width * Output Pixel Aspect Ratio / (Original Width * Source Pixel Aspect Ratio / Original Height) = New Height

Oh Gods, another 1000501 esoteric explanation in the Internet with no real explanation. It seems that you guys all don’t know what lies at the heart of the formulas and why this happens. Someone once upon a time told you these formulas and said believe me and it. And you answered: we believe you! Hallelujah! (of course, because you had to pass the exam and complete the student session). And since then you just copy-paste the ones on the forums and also say "Hallelujah! Believe it/me/us !

Yes, there is some light trolling here, but only very little.

You don't agree? Okay, then please try again, but clearly answer the question "why?".

How did you get this block [Output Pixel Aspect Ratio / (Original Width * Source Pixel Aspect Ratio / Original Height)] ?
Why we e.g. New_w multiply by 'Output Pixel Aspect Ratio' but not by something other why Original Width multiply by Source Pixel Aspect Ratio but not by something other ... ect. .. What do we get if we New_w multiply by 'Output Pixel Aspect Ratio' ... ect..
I want to understand what is happening at each specific stage and why this is happening. Why do we divide or multiply something by something and not by something else? Why do we even multiply or divide something by something?

--------------

@Cornucopia
Let's get 'New Hight' using your universal formula.

DAR_orig = DAR_target = 1.33
DAR_orig = Horiz rez / Vert rez * PAR = Horiz rez / Vert rez * PAR = DAR_target
720:576 * 16:15 = (720 : X) * 8:9
1.25 * 1.067 = (720 : X) * 0.89
1.333375 / 0,89 = 1,50
720 : X = 1.50
X=480

It's just elementary! Let's compare this with the formula from hello_hello ...
No, better to shoot myself right away.

However, how to use it (e.g. how to get Width_target) if we add a standard 4:3 footage to a 16:9 anamorphic project and vice versa: 16:9 anam. project to a standard 4:3 project ? (i.e. if we have different DAR_orig and DAR_target)

[Bwaak]

All you have to do is specify what each of your acronyms mean. Then there's no confusion.

What if there are no acronyms? Say, 720x480 is frame size in pixels (FS maybe?) What do you call 720/480 ratio? Frame Size Ratio (FSR)? Frame Aspect Ratio (FAR)? Need to make sure not to confuse it with DAR.

I want to understand what is happening at each specific stage and why this is happening. Why do we divide or multiply something by something and not by something else? Why do we even multiply or divide something by something?

This is as good as any other page to start from: A Quick Guide to Digital Video Resolution and Aspect Ratio Conversions by Jukka Aho.

[Ennio]

I think everyone who has worked it out understands. But okay, I'll play academic ball. TBH, I find the acronym PAR confusing. Because "Pixel Aspect Ratio" to me can already be easily interpreted wrongly. Wouldn't it be better to replace PAR with OPAR - One Pixel Aspect Ratio? This one-letter difference I believe can take a lot of doubts away.

It all doesn't matter, as long as you indeed know and understand what lies under.
Like jagabo said:

All you have to do is specify what each of your acronyms mean. Then there's no confusion."

Very wise words. They relate to logics and point back-to-basics.

Things just start to move forward more easily after drawing a problem down on paper, because a solution or the way towards it just start to pop out of those drawings. I have a school background to that in the first place...

Same here. Other than on paper, in physics/mechanics- & math class, we were always invited (if not dragged) to the chalkboard to draw and work things out. Visualization indeed often is key to things starting to live. Everyone can relate to a picture.
Nowadays seeing highschool kids going with a laptop to class - where sometimes there even isn't a chalkboard anymore(!) - I'm like WTF, this can't be good. I still am experiencing trouble getting these kind of things directly from a pc monitor. Maybe I'm too old-skool.

[Alwyn]

I just encode to 768x576. Square pixels just like almost very other video out there.

[Sharc]

Wouldn't it be better to replace PAR with OPAR - One Pixel Aspect Ratio? This one-letter difference I believe can take a lot of doubts away.

No,No,No ... please don't invent new "private" acronyms. The worst of all scenarios IMO.
But yes, people should explicitly spell out what they mean with "their" acronyms. The problem is that people often have a vague idea only of the problem they try to describe.

Edit:
Btw, AvsPmod has a nice resize calculator (Tools-> Resize calculator ....)

[Sharc]

I just encode to 768x576. Square pixels just like almost very other video out there.

768x576 square pixels applies for properly cropped 4:3 PAL footage only.

[lollo]

I just encode to 768x576. Square pixels just like almost very other video out there.

almost = Alwin

768x576 square pixels applies for properly cropped 4:3 PAL footage only.

Yes, (useless) resizing needs to be applied on a "correct" source anyhow, and does not help in any of Display Aspect Ratio or Pixel Aspect Ratio considerations.

Note how I did no use any acronym as requested

[Alwyn]

does not help in any of Display Aspect Ratio or Pixel Aspect Ratio considerations.

Eh? Once you have square pixels, all of this stuff PAR DAR SAR ODAR etc etc is no longer relevant. I have never heard any of this type of discussion on HD, UHD, 4k, or 8K videos.

[lollo]

You were talking about "encoding" to 768x576. You meant "resizing", because the first has no sense. It (generally) assumes starting from a SD frame (unless you downscale). Out of contest.

[Sharc]

Eh? Once you have square pixels, all of this stuff PAR DAR SAR ODAR etc etc is no longer relevant. I have never heard any of this type of discussion on HD, UHD, 4k, or 8K videos.

2 comments:
- "Once you have ..." -> requires correct resizing for all non-square sources (e.g. all DVDs, home video/VHS captures, blu-ray extras/bonus .....)
- Display Aspect Ratio and "resolution" (aka frame width/hight aka frame aspect ratio, movie aspect ratio, letterbox/pillarbox for 16:9 (or other) TV canvas) are still relevant for square pixel stuff.

[Ennio]

Wouldn't it be better to replace PAR with OPAR - One Pixel Aspect Ratio? This one-letter difference I believe can take a lot of doubts away.

No,No,No ... please don't invent new "private" acronyms. The worst of all scenarios IMO.

Of course I agree with this. Do note that my suggestion was initially meant as a tease, stirring things a bit up academically. And to expose that such will unnecessarily add to confusion which underlying matter will not benefit from. Where examplary, a very reply like yours immediately reflects upon.

[dellsam34]

If one is going to make a lossy step and resize to obtain a square pixel may as well resize to a modern higher resolution for less destruction (it's a math thing), Otherwise the display chip is going to make another resize for you. HD is the minimum requirement because all future resolutions are multiple integer of HD, therefore no complex math is required, just line doubling.

[lollo]

If one is going to make a lossy step and resize to obtain a square pixel ... to a modern higher resolution for less destruction ...

Absolutely yes. Or just avoid the first lossy resizing step

[dellsam34]

Yes, Keeping the original file without any processing is the wise thing to do.

[hello_hello]

New Width / Original Width * Original Height = New Height
is the basic equation for calculating resizing,

It would be more logical to write it as
New Width [1920] * Original Height [540] / (Original Width [1280] = New height [810].

because
New Width [1920] Original Width [1280]
x
New height [???] Original Height [540]

It depends on your viewpoint. Using the DAR to calculate the new height when you want to resize the width to 1920.
1920 / (16 / 9) = 1080
Would become, in my head, out of habit:
1920 / 16 * 9 = 1080
Or using the original resolution instead of the DAR:
1920 / 1280 * 720 = 1080
And the way I store that in my head is:
New Width / Original Width * Original Height = New Height

You posted above
New Width * Pixel Aspect Ratio / Original Width * Original Height = New Height
I asked how did you get that ?
You answered
New Width * Output Pixel Aspect Ratio / (Original Width * Source Pixel Aspect Ratio / Original Height) = New Height

I thought I explained it, but maybe not clearly.....
Above the input and output DARs were used for the calculation, only in the form of resolution for non anamorphic video.
For an anamorphic source, the same equation applies, but to convert the resolution to a DAR the width must be multiplied by the PAR. For a 16:9 PAL DVD that's 1.4222222.

720 x 1.422222 / 576 = 1024
1024 / 576 = 1.777777 (a DAR of 16:9)

1920 / 1024 * 576 = 1080
In one step:
1920 / (720 * 1.422222) * 576 = 1080

Adding the output PAR to the equation, but because the output is non-anamorphic in this case, the output PAR is 1:1 or 1.0
(1920 * 1.0) / (720 * 1.422222) * 576 = 1080

Oh Gods, another 1000501 esoteric explanation in the Internet with no real explanation. It seems that you guys all don’t know what lies at the heart of the formulas and why this happens. Someone once upon a time told you these formulas and said believe me and it. And you answered: we believe you! Hallelujah! (of course, because you had to pass the exam and complete the student session). And since then you just copy-paste the ones on the forums and also say "Hallelujah! Believe it/me/us !

If you want to resize the same 16:9 PAL DVD to 16:9 NTSC DVD dimensions (720 x 480), instead of a PAR of 1.0 the NTSC 16:9 DVD PAR of 1.1851 must be used. If you don't know what that is, calculate it first.
480 * 16 / 9 = 853.33
853.33 / 720 = 1.1851

(720 * 1.1851) / (720 * 1.422222) * 576 = 480
(New Width * New PAR) / (Original With * Original PAR) * Original Height = New Height
In this case the new width of 720 is exactly the same as the original width, and the calculated new height is the expected 480.

I want to understand what is happening at each specific stage and why this is happening. Why do we divide or multiply something by something and not by something else? Why do we even multiply or divide something by something?

Normally when resizing, the source PAR and dimensions would be known. From there you pick the output width you want, add an output PAR to the equation if you want an anamorphic output, which could be any PAR you choose (although for a DVD output you'd use the appropriate DVD compliant PAR), but no matter what width you choose to resize to, or the PAR you want the output to have, the previous equation will give you the correct New Height to resize to (for the specified New Width).

One more example, up-scaling a 16:9 PAL DVD to 1440x1080 with a 16:9 DAR. The upscaled video must have a PAR of 4:3 to display as 1920x1080 (16:9).

1920 / 1440 = 1.3333 or 4:3

(1440 * 4 / 3) / (720 * 1.422222) * 576 = 1080
(New Width * New PAR) / (Original With * Original PAR) * Original Height = New Height

1440x1080 with a DAR of 16:9 is a valid resolution and DAR for bluray, which is why I used it as an example.
https://en.wikipedia.org/wiki/Blu-ray#Video

[Ennio]

1440x1080 with a DAR of 16:9 is a valid resolution and DAR for bluray, which is why I used it as an example.

And a tricky one for people who start learning, I might add. When working out on their meanings and running the numbers, PAR and SAR having the same value carries mad-driving potential