still confused about PAR

[Séamus]

Hi,
I'm quite new to capturing
My question however is about pixel aspect ratio
I'm still a little confused on the PAR where and when

So basically I could ask you:
At what point does the PAR get changed ?
-the PAR of the movie inside the DV-camera, is it square or non-square ?
-when I'm editing my captured clips (with Premiere) I suppose I see square pixels ? Everything on PC is always square pixels ?
-does the DV camera alter the PAR when streaming to a TV ?
-finally, does a DVD authoring program alter the PAR or does the DVD player do it ?

As you can see I'm pretty confused about the matter :rolleyes:

The only thing I understand is the resizing of stills for dvd authoring.

Thx for any help,
Séamus

[nicco]

DV-standard:

PAL 4:3 Pixel Aspect Ratio: 1,066
PAL 16:9 PAR : 1,42

NTSC 4:3 PAR : 0,9
NTSC 16:9PAR : 1,2

In premiere you can manually set the PAR in project settings, but if you choose a DV-preset you arleady have the right PAR for your DV video, the same is for DVD authoring program (I think)

[Séamus]

Ok, thx, I'll sum it up in three questions then...

1. Assuming my Premiere project would have a PAR of 1.067 (PAL DV)
If I export a frame, edit it with Photoshop and re-import it, will the PAR have changed ?

2. On the other hand, if I make a new image in Photoshop I'd have to start with a size of 768x576, resize the final to 720x576 and then import into Premiere or a DVD authoring program ?

3. When exporting the whole movie to make a copy for a PC/web based movie, would I have to choose square pixels when exporting ?

Thx

[nicco]

Don't care so much about PAR, you can do a 720x576 image in photoshop and then import in premiere. New Phothoshop CS is able to manage PAR, but is not so foundamental, when you import in prem your image became automatically with the right PAR

[illCP]

I was asking about Resolution & PAR in the German Doom9-Board. I still don't understand everything about it, but I can say

PAL 4:3 Pixel Aspect Ratio: 1,066

is not correct. I also always thought that this is right, because Premiere shows the PAL-DV-PAR as 1,0667, but this is called "generic PAR".

scharfis_brain and Kika in the German Board told me how it's right :

There is no 1.066667, it's wrong.
768 x 576 is the full PAL resolution with a PAR of 1:1.
But Videos aren't digitalised with square pixels, they are digitalised with 702 active Pixels. So 768 Pixels are becoming 702 Pixels ==> so the PAR is 768 / 702 = 1,094.
MPEG needs a so called Mod16 Resolution, so 2 "black lines" with one pixel width are added on the left and the right side, so you've got 704 x 576.

720 x 576 is 704 x 576 with 2 "black-pixel-Lines" with the width of 8 pixels at each side, but the PAR remains 1.094.


But I have to admit that I still don't really understand what's up with this "generic PAR" of 1.0667. But I was told that Premiere does it wrong.

[Arachnotron]

But I have to admit that I still don't really understand what's up with this "generic PAR" of 1.0667. But I was told that Premiere does it wrong.

We hope to add some of this stuff to the next capture guide. :)

This generic PAR is the PAR you get when you use a capture card with a standard PAL capture width (52 µs wide) but cap at 720x576. Actually such cards are rare; so far, the only one that does this are BT878 based cards and then only if you use the BTWincap drivers.

To Explain further:
From ITU-R BT.470-6 we know that for PAL, an active area of 52 µs x 576 lines results in a picture with 4:3 Aspect Ratio when played back on TV.

If you put 720 x 576 pixels in this area, you get
PAR = IAR * [vert res] / [hor res]
PAR = 1.333 * 576 / 720 = 1.0666

The reason this PAR is used is that it is very common for people to have an picture that is exactly 4:3 and mistakenly put 720x576 in it. Since many people also make the mistake in assuming their capture device captures exactly the PAL active window, this PAR is referred to as generic PAR.

Now, a PAL DVD actually has an active area of 53.333 µs x 576. (or 720 pixels / 13.5 MHz = 53.333)
So a PAL DVD is not exactly 4:3; it is a bit wider. The Aspect ratio now becomes (53.333 /52) * 4 / 3 = 1.3675

So PAL DVD PAR = IAR * [vert res] / [hor res]
PAR = 1.3675 * 576 / 720 = 1.094

This is the PAR you get if you cap the same 720x576 using and ITU device like DV devices which have an capture window width that maches that of a DVD

[Amit]

Maybe this link will help.

http://www.bbc.co.uk/commissioning/branding/picturesize.shtml

-Amit

[Kika]

@Arachnotron

Um, the Active Area is always 52µs. No matter what's inside the "missing" 1.33µs, it should never be visible on a correct adjust TV. You can't change the horizontal Timing on a standard CRT-based TV.

A 4:3 PAL DVD without generic PAR should have an active (motion) Area of 704x576 which is similar enough to the "true" 702 horizontal PAL Pixels. This Motion Area does have the exact AR of 4:3.

[Wilbert]

Um, the Active Area is always 52µs.

True.

No matter what's inside the "missing" 1.33µs, it should never be visible on a correct adjust TV. You can't change the horizontal Timing on a standard CRT-based TV.
Arachnotron is talking about the capture window of a capture card/driver. This is rarely 52 µs. It's often more (4:3 picture + some black pixels), but sometimes also less (missing a part of the 4:3 picture).

[trevlac]

Originally posted by Séamus

So basically I could ask you:
1. At what point does the PAR get changed ?
2. the PAR of the movie inside the DV-camera, is it square or non-square ?
3. when I'm editing my captured clips (with Premiere) I suppose I see square pixels ? Everything on PC is always square pixels ?
4. does the DV camera alter the PAR when streaming to a TV ?
5. finally, does a DVD authoring program alter the PAR or does the DVD player do it ?



PAR is fun ... so I'd thought I'd try to answer the original questions:

1. When the image is resampled. For DV, this PC pre-viewing, but it is not saved this way.

2. DV is non-square and it is not 4:3. Hence 720x576. 702x576 is 4:3 PAL, and 768x576 is square 4:3 PAL.

3. I don't use Premiere, but it probably does a resize for preview. For a proper AR, things are square on a PC. But if the source is not 4:3 square, the software has to know what to do to show it as 4:3 square.

4. No. The DV cam must convert to PAL analog. It has a 53.333ms x 576 picture. It crops this to 52ms x 576 (or 702x576). It knows what to do because PAL is a standard. So is DV.

5. Neither. The PAR on a DVD is the same as used by a TV. 720x576 is the full area that can be used. This is not 4:3 nor is it square pixels. 704 is cropped. 702 is cropped some more. TV Overscan crops more. SO if you feed 720x576 DV to an Mpeg encoder and author a 720x576 DVD, and then finally play it as 702x576 to your PAL TV, the par never changed.

--------------------
If you think this is a mess, look at some film options. :)

http://www.dvdaust.com/aspect.htm

An interesting thing to note is that Pre 1953 35mm movies had an aspect ratio of 1.375. I wonder if in the transfer of these to DVD they crop to 1.333 or use the full 720 samples? :D

[Arachnotron]

Arachnotron is talking about the capture window of a capture card/driver. This is rarely 52 µs. It's often more (4:3 picture + some black pixels), but sometimes also less (missing a part of the 4:3 picture).

Exactly. :)

By the way, 52 µs is the official norm. That does not mean everybody adheres to it. pre-recorded tapes, game consoles etc. may deviate a bit from it. In general, PAL sources seems to stay closer to the norm than NTSC sources do. Especially PAL broadcasts of imported ntsc stuff can have different values.

[Kika]

By the way, 52 µs is the official norm. That does not mean everybody adheres to it.

That's true... unfortunatly :(

The 16 (18) additional Pixels (1.33µs) on 720x576, aren't using a clearly defined area.

As an example: A couple of weeks ago, i had a DVB-S-Stream of the Movie Evolution. Format was 720x576, but the Picture-Area was only around 672 - with all the black stuff on the right side! :rolleyes:

[Wilbert]

As an example: A couple of weeks ago, i had a DVB-S-Stream of the Movie Evolution. Format was 720x576, but the Picture-Area was only around 672 - with all the black stuff on the right side!
This doesn't say much about your active capture window. That still can be 52 µs.

I don't know about DVB stuff and such. But, on analog caps there can be two problems:

1) The 4:3 includes black stuff, and you capture card/driver captures it all. That happens a lot for foreign broadcast material.

2) Your active capture window is wider than the 4:3 picture, which also results in capping black stuff.

[Arachnotron]

This doesn't say much about your active capture window. That still can be 52 µs.

I think DVB-S is digital broadcast (a MPEG-2 variant) and a DVB card would capture the stream digitally. As such, the active cap area would be whatever active area was used when creating the 720x576 stream by the producer.

That probably equals 53.333 µs, but I could not find a hard definition of it anywhere. ETSI TR 101 154 simply states that it is 4:3 but that may be a simplification analogue to stating 720x576 DVD is 4:3. (it isn't really, 702x576 is, 720x576 is a bit wider, but since the extra bits are chopped of on a TV the resulting picture will have 4:3 AR so nobody cares :) )

If DVB 720x576 is truly 4:3, you would have to resize to 702x576 if you wanted to convert it to DVD with correct AR.

[Kika]

On DVB you can find (nearly) anything:
The "720-Group":
480x576
544x576
720x576

And the "704-Group":
528x576
704x576

All this resolutions are used - with a complete mixup of PARs/ARs.
But as far as i know, there are the same AR/PAR-Rules for DVB like for DV or DVD. And you will find the same confusion: Some Movies with generic PAR, some with the "correct" PAR.

If i have to re-encode DVB (or other MPEG2-Source), i'm looking for round Objects in the Movie and try to get it round in my results. Guees that's the only way to be sure to do it the right way (more or less). :sly:

[SeeMoreDigital]

A lot of people are also getting confused and maybe forgetting that when capturing say from European DVB broadcasts or DV camcorders that the streams are in fact anamorphic.

DVB broadcasts can be either 16:9 anamorphic or 4:3 anamorphic. And most DV camcorder recordings are 4:3 anamorphic.

That said I really do think that the terminologies and abbreviations we often like to use are a little misleading.

Take PAR (Picture Aspect Ratio) for-instance. Is this just the overall image pixel frame size (say 720x576), or is it referring to the aspect ratio of the viewable picture, with or without mattes?

I put some info on my website a while ago because many people don't understand what an anamorphic image really is, what they look like, and why they are being used?

Some people might find this thread (http://forum.doom9.org/showthread.php?s=&postid=485577#post485577) in the DivX section useful, together with some info I've put in 'Area 03' of my web site.


Cheers

[Wilbert]

4:3 anamorphic. And most DV camcorder recordings are 4:3 anamorphic.
I hope this is just a typo. Anamorphic always refers to 16:9.

[SeeMoreDigital]

Not necessarily Wilbert!

Say you connected the composite/s-video analog outputs from your PAL VHS video recorder, played an pre-recorded movie and recorded/captured the images with your PAL DV camcorder. These images will contain 720x576 pixels, which is an 1.25:1 frame. But your source was 1.33:1. So what you've actually captured is this: -

http://82.2.167.24/03_Video_Only_Info/Anamorphicly_Encoded_Images/PAL_Fig_04.jpg

But if you you connect the composite/s-video analog outputs from an DigiTV box, tune into a 16:9 widescreen broadcast and record/capture the images with your PAL DV camcorder, what you would capture is this: -

http://82.2.167.24/03_Video_Only_Info/Anamorphicly_Encoded_Images/PAL_Fig_01.jpg

It's the electronics in the camcorder that corrects the output AR, prior to sending the image to your TV, which always assumes is a 4:3 TV. This is why when you connect a DV camcorder to a 16:9 TV you have to 'manually' adjust the TV's AR to 4:3.

Some camcorders are thick, but some work like DVD players and DigiTV boxes and allow you to choose between 4:3 or 16:9 TV output.


Cheers

[Kika]

OK, ok, now the Confusion is perfect... ;)

PAR means Pixel Aspect Ratio, not Picture Aspect Ratio.
In the digital Video World there are thre Kinds of ARs:

DAR = Display Aspect Ratio (the Screen in short: AR)
PAR = Pixel Aspect Ratio (the Pixels themself)
SAR = Sample Aspect Ratio (how a Capture device is sampling)

SAR isn't important for us an can be ignored here.

A PAL Fullscreen-MPEG2-Video does have the DAR (AR) of 4:3 and the PAR of 1.092 (1.094). The Resolution ist 704(702)x576.


Anamorphic means a "squeezed" Video in general! A 4:3 SVCD contains (in this use of the term) a anamorphic Video.

OK, but i think, that's not really usable, because the truth is, that SVCD uses a different PAR. The better way to use the term anamorpic is imho to use it only for anamorphic 16:9-Video.

[SeeMoreDigital]

Exactly... I knew my PAR comment would spark somebody off!

Sometimes you see PAR in-correctly being decscribed as being 'Picture Aspect Ratio'

It's no wonder some people get confused :confused:


Cheers

[Wilbert]

Not necessarily Wilbert!

Say you connected the composite/s-video analog outputs from your PAL VHS video recorder, played an pre-recorded movie and recorded/captured the images with your PAL DV camcorder. These images will contain 720x576 pixels, which is an 1.25:1 frame. But your source was 1.33:1.

Suppose you capped the 4:3 image (and nothing more). What does this have to do with being anamorphic or not?

Have a look at the following links:

http://members.aol.com/ajaynejr/vidglos3.htm#Enhanced
http://members.aol.com/ajaynejr/vidglos.htm#Anamorphic
http://members.aol.com/ajaynejr/vidwide.htm#Anamorphic


Exactly... I knew my PAR comment would spark somebody off!

Sometimes you see PAR in-correctly being decscribed as being 'Picture Aspect Ratio'

It's no wonder some people get confused :confused:
Nice one. Just pretending that we are confused. "DVB broadcasts can be either 16:9 anamorphic or 4:3 anamorphic." You are clearly talking about DAR here.

@Kika,

Anamorphic means a "squeezed" Video in general!
Reference please?

[SeeMoreDigital]

Much if not all of that info refers to NTSC, not PAL.

If you connected the composite/s-video analog outputs from your NTSC VHS video recorder, played an pre-recorded movie and recorded/captured the images with your NTSC DV camcorder. The images would contain 720x480 pixels, which is an 1.50:1 frame. But the source was 1.33:1. So what you've actually captured is this: -

http://82.2.167.24/03_Video_Only_Info/Anamorphicly_Encoded_Images/NTSC_Fig_04.jpg

This image is one hell of a lot easier to resize back to a genuine 4:3 (1.33:1) image, because you can squeeze it to 640x480 pixels. To resize a PAL 4:3 (1.33:1) capture you would have to stretch it to 768x576!


Cheers

[Kika]

@SeeMoreDigital

Um, 720x480 IS a 4:3 (1.33:1) Screen. The DAR is 4:3, but the Pixels are non square (PAR isn't 1:1 like at 768x576 or 640x480).


@Wilbert
Source was one of the old toolbox.sgi.com sites which are no longer online :(

[SeeMoreDigital]

Originally posted by Kika
@SeeMoreDigital

Um, 720x480 IS a 4:3 (1.33:1) Screen. The DAR is 4:3, but the Pixels are non square (PAR isn't 1:1 like at 768x576 or 640x480). I've been carefully not to mention anything here about PAR and/or DAR. Only what the genuine shape of a 4:3 or 16:9 image looks like when captured by an DV camcorder!


Cheers

[trevlac]

Originally posted by Kika
Um, 720x480 IS a 4:3 (1.33:1) Screen. The DAR is 4:3, but the Pixels are non square (PAR isn't 1:1 like at 768x576 or 640x480).


Funny thing here is that the people posting probably know what they are talking about ... but it still does not clarify it for others.

For example, 720x480 can contain a 4:3 image, but it is NOT a 1.33:1 size. The DAR is 4:3, the Pixels are not square, AND there are extra pixels. 711x486 for NTSC and 702x576 for PAL are 4:3 TV sizes. 720x480 is ~ 1.35 NTSC and ~ 1.37 PAL. So saying 720x480 is 4:3 completely confuses people.

BTW: I think PAR is a waste. Pixels do not have an area. There is space between pixels. Talking sample rate and line length is mucho mas clear.

For example: DV/DVD sample rates are normally 13.5. NTSC line length is 52.6666, PAL 52. 13.5*52 = 702. 13.5*52.666 = 711. Full digital line length can be 53.333. 13.5*53.3333 = 720. Crop means change the line length. But you do not change the sample rate ... unless you are talking a different 'standard' ...


DOWN WITH PAR !

[Arachnotron]

Say you connected the composite/s-video analog outputs from your PAL VHS video recorder, played an pre-recorded movie and recorded/captured the images with your PAL DV camcorder. These images will contain 720x576 pixels, which is an 1.25:1 frame. But your source was 1.33:1. So what you've actually captured is this: -

It was my understanding a DV device is a 13.5 Mhz device. As such, when you play a PAL VHS tape to it, the result would be a 702 pixels x 576 picture with 4:3 AR with 18 pixels of black at the sides. I don't see those in your example pic. Or are there DV devices out there that don't sample at 13.5 MHz?

[edit]

You should see PAR as the aspect ratio of the pixel grid, not as the AR of an individual pixel, since it has none. Lousy term that, PAR

[Kika]

@ trevlac

For example, 720x480 can contain a 4:3 image, but it is NOT a 1.33:1 size.

Oops, now you got me. Guees those confusion spreads out... ;)
That's why i prefer to use 704 only.

But why do you think, the use of PAR is a bad thing? It can be easy used for the maths.
I don't think it's easy to understand for a newbie that PAL is 702x576 and NTSC 711x480.

@Arachnotron
I have seen some VHS-Captures done by DV-Devices (PAL only). They all had a ~ 704x576 Picture with black bars at the sides.

[Arachnotron]

What really confuses the issues is that many people, when they are talking about AR of a DVD or such, are actually referring to the AR of the target device (TV) which is not neccecarily the same as the AR of the image.

Why not use:
DAR - display aspect ratio, i.e. AR of device it is intended for
IAR - Image display ratio, 'true' AR of the actual file
PAR - Pixel aspect ratio, AR of pixel grid

4:3 720x576 and 704x576 DVD
- result in the same picture on a TV, AR and DAR are the same;
- 720 is a bit wider, so IAR is different
- both use 13.5 MHz samplerate, so PAR is the same.

I know redefining DAR is not realistic, but I think the sloppy use of the term AR is at the root of the confusion about PAR.

[Kika]

What really confuses the issues is that many people, when they are talking about AR of a DVD or such, are actually referring to the AR of the target device (TV) which is not neccecarily the same as the AR of the image.

Good Point. I do always refer to the target device.

I think an other problem is the visible Area of a CRT-TV. On PAL 52µs, 52.66µs on NTSC. How to explain to others, that only this Area is clearly defined? In german Boards i tried terms like Overscan-Area, but this often collides with the normal TV-Overscan.

I like your idea of the IAR, but it's only a dream. Or do you know a way to let this become a standard term? ;)

[Arachnotron]

I think an other problem is the visible Area of a CRT-TV. On PAL 52µs, 52.66µs on NTSC. How to explain to others, that only this Area is clearly defined? In german Boards i tried terms like Overscan-Area, but this often collides with the normal TV-Overscan.

I would prefer the term active area. Active area being the part of the video signal containing picture information, whether it will be displayed or not.
So the active area is 52 µs for PAL, but the visible area on a CRT-TV is less. visible area on TV = Active area - overscan

To be exact, 52µs is only the optimum value in an allowed range of 51.7 to 52.0 µs. Dutch broadcast seems to stay close to 52 µs.

[SeeMoreDigital]

Yes, as I said in my first post: -

"I really do think that the terminologies and abbreviations we often like to use are a little misleading"

For-instance, on an DVD you have: -

The Image Pixel Frame Size. Which for NTSC is 720x480 (1.50:1). And for PAL is 720x576 (1.25:1).

Then you have to know the AR of the anamorphic video stream, which can be either 4:3 or 16:9.

And then have the Movie's Aspect Ratio. Which could be anything, but usually, 1.33:1, 1.77:1, 1.85:1, 2.35:1 etc


Cheers

[Kika]

Active area being the part of the video signal containing picture information, whether it will be displayed or not.

I know. I think Active Area is a better Term for what i pronounced as visible Area. I used that because it is the area for valid picture informations.

I think, someone (with a better knowledge of english than me) should use this Thread to build up an AR/PAR (IAR)-Guide.

To be exact, 52µs is only the optimum value in an allowed range of 51.7 to 52.0 µs.

Yes, and that's why i have become a little angry after seeing those DVB-Stream of Evolution with all the blacknes at the right side. The left side is a much better place for black bars...


The Image Pixel Frame Size. Which for NTSC is 720x480 (1.50:1). And for PAL is 720x576 (1.25:1).

Sorry, but that's not correct or, to be precise, there is something missing.
An Example: To get an correct 4:3-Image for DVD out of a Photo, you need the Size of 768x576 (perfect 4:3). Now you can use 2 Ways:

1. Resize to 704x576, add bars to get 720x576
2. Add bars to get 786x576, resize to 720x576

[SeeMoreDigital]

Originally posted by Kika
...Sorry, but that's not correct or, to be precise, there is something missing.
An Example: To get an correct 4:3-Image for DVD out of a Photo, you need the Size of 768x576 (perfect 4:3). Now you can use 2 Ways:

1. Resize to 704x576, add bars to get 720x576
2. Add bars to get 786x576, resize to 720x576

You may well be right but it does not stop Microsoft describing it in this way: -

http://82.2.167.24/Uploaded_Files/Doom9_Forum_files/M$_image_summary.gif

So how would you describe the above, if the term 'image pixel frame size' is not correct? I suppose 'image pixel size' would do (or IPS)!

I really do think it's time for us to champion some understandable abbreviations for our newbies!


Cheers

[Arachnotron]

So how would you describe the above, if the term 'image pixel frame size' is not correct? I suppose 'image pixel size' would do (or IPS)!

ISIP (image size in pixels) :D

[trevlac]

Originally posted by SeeMoreDigital
Much if not all of that info refers to NTSC, not PAL.

If you connected the composite/s-video analog outputs from your NTSC VHS video recorder, played an pre-recorded movie and recorded/captured the images with your NTSC DV camcorder. The images would contain 720x480 pixels, which is an 1.50:1 frame. But the source was 1.33:1. So what you've actually captured is this: -

...

This image is one hell of a lot easier to resize back to a genuine 4:3 (1.33:1) image, because you can squeeze it to 640x480 pixels. To resize a PAL 4:3 (1.33:1) capture you would have to stretch it to 768x576!


This is not correct. Or at least not clear... like your 'Picture Aspect Ratio' comment. ;)

A little depends upon what the DV cam does. I would say it takes a full 53.333ms line and turns it into 720x480. Yep, sync signal and all in the xtra 53.333 - 52.666 = 0.6664ms. So you picture is 1.333, just like it was broadcast. It is never 1.5. You just can't do 720/480 = 1.5. This is because the space between the pixels is not square. And there is this xtra ~ 9 pixels.

In the end, NTSC is much more messy vs PAL. For example, DV cap a PAL signal 720x576. Crop it to 702x576 to get rid of the sync 'black' and resize to 768x576 gives you 4:3 on a PC. If you want 4:3 on a TV, you already have it.

For NTSC, 720x480 contains ~9 xtra pixels on the sides and 6 pixels cropped (probably 4 from the top and 2 from the bottom). So 711x486 is the real 4:3 picture contained within the 720x480 frame. If you want a 4:3 picture on a PC you must crop off the 9 and resize to 648x480. That is not a typo! 648x486 is the true 4:3 square pixel NTSC size. Remember we lost 6 in the DV cap.

In the end ... You can get there many ways. NTSC is more complicated than PAL. The above stuff should be clear before we start talking 16:9 or things like 1/2D1 / SVCD.

BTW: anamorphic can not contain the resolution of 4:3. This is because there are less sample for a given line length. If you understand that, you get all of the rest. :)

[trevlac]

Originally posted by Arachnotron
ISIP (image size in pixels) :D

The part about "Which for NTSC is 720x480 (1.50:1). And for PAL is 720x576 (1.25:1)." is what was wrong.

Call it what you want, but 720x480 is not 1.5:1. Unless you messed up the source.

All the dialog is saying is there are 720x480 pixels. This is far from describing an Aspect Ratio. You could stick anything in there you want. AKA you could have a sample rate of 27Mhz.

[SeeMoreDigital]

Originally posted by Arachnotron
ISIP (image size in pixels) :D Joking apart that suggestion might be more understandable than how these typical applications are describing the same thing: -

http://82.2.167.24/Uploaded_Files/Doom9_Forum_files/Gspot_description.gif

http://82.2.167.24/Uploaded_Files/Doom9_Forum_files/AviCheck_description.gif

Nero Recode2
http://82.2.167.24/Uploaded_Files/Doom9_Forum_files/Recode2_description.gif

It just reinforces my point. If something as simple as this can't be abbreviated properly and in an standardised way, how are our newbies expected to understand the more technical aspects of encoding?


Cheers

[Arachnotron]

@trevlac
The part about "Which for NTSC is 720x480 (1.50:1). And for PAL is 720x576 (1.25:1)." is what was wrong.

Call it what you want, but 720x480 is not 1.5:1. Unless you messed up the source.

All the dialog is saying is there are 720x480 pixels. This is far from describing an Aspect Ratio. You could stick anything in there you want. AKA you could have a sample rate of 27Mhz.This is another example of not defining what you are talking about leads to confusion.

1.5:1 (or 3:2) is correct for 720x480, but it isn't the aspect ratio; it is the pixel ratio (if that word exists).

Also, I think saying 720x576 is 4:3 is asking for problems. It is not; it is intended for 4:3 viewing. 702x576 really is 4:3. 720x576 is 4.1:3

In that sence, PAR is better defined. With a PAR number, you know exactly what is going on (provided you understand how it works). With AR numbers, you are always left guessing......

[Kika]

1.5:1 (or 3:2) is correct for 720x480

Yes, but it does not make any sense to use this (imho). It is a factor for defining the image by only using the pixels as is. No PAR, no DAR (AR), nothing concrete.

So i think we should always use the PAR/DAR for defining what we are talking about.
Simply use the PAR (Pixel Aspect Ratio) as a factor for stretching or crunching non square pixels to square pixels and you get what you want.

[SeeMoreDigital]

Originally posted by Kika
Yes, but it does not make any sense to use this (imho). It is a factor for defining the image by only using the pixels as is. No PAR, no DAR (AR), nothing concrete.Of course it makes sense to use it. Because although the horizontal (720) figure will not correctly represent the 'movies' aspect ratio (lets call that MAR :D) it sure does for the vertical (480 or 576). As anamorphic pixels are only stretched horizontally.

Does that make any sense?


Cheers

[trevlac]

Originally posted by Arachnotron
1.5:1 (or 3:2) is correct for 720x480, but it isn't the aspect ratio; it is the pixel ratio (if that word exists).


Call it what you want ... 'Pixel Ratio' is a worthless value. I agree with Kika.

Here are some pixel ratios, 648x480, 640x480, 711x480, 720x480, 480x480.

You can calculate the pixel ratios if you want. :)

But I bet you already know the AR contained in those frames. That is because you 1st assume they all contain 4:3, unless I specified anamorphic. From that you can easily get the PC or TV output you want by a horizontal resize/crop. God help you if you want to make it PAL. If those frames do not contain 4:3, then you are out of luck. Something messed it up along the way.


I would call this the convention. It IS 4:3 unless anamorphic is specified. If you have 720,704,352,480, assume the width is a factor of 702/711 for 4:3.

After much talk about this, I'm not sure why we have to put it all in math terms. A frame size is like an Auto. There are big ones and little ones. Little ones with small motors and little ones with big motors. We all know this, and generally know what we are looking at. We don't seem to get stuck on some power to weight ratio.

But of course, I'm easy ... we will do what we must.
================

@SeeMoreDigital,

No offence ... but you really don't appear to know what you are talking about. That's fine ... but, as you keep throwing out more words to see if they 'stick', this just adds to people's confusion.


To answer you question/comment on what various dialog boxes say: Never assume a programmer has any idea of the proper use of terms. The usually don't. Especially, if you are talking about consumer level or free software. They would have to be an expert in 2 domains (video & programming) to get it all correct.

PS: I'm an expert in neither. :D

[SeeMoreDigital]

Originally posted by trevlac
SeeMoreDigital,

No offence ... but you really don't appear to know what you are talking about. That's fine ... but, as you keep throwing out more words to see if they 'stick', this just adds to people's confusion.


To answer you question/comment on what various dialog boxes say: Never assume a programmer has any idea of the proper use of terms. The usually don't. Especially, if you are talking about consumer level or free software. They would have to be an expert in 2 domains (video & programming) to get it all correct.

PS: I'm an expert in neither. :D No offense taken... and yes I do know what I'm talking about I've been converting both analogue and digital video professionally for years :D

And PAR is very important. As I said before, where alot of people get confused is it's relationship with 'anamorphic' encoding.

By the way 3ivX has generated a good explaination about PAR. You might want to give it a read: -

http://www.3ivx.com/support/par.html


Cheers and good luck

[Wilbert]

To quote some stuff on that page


To take full advantage of all pixels in the frame, some DVDs are authored anamorphicly (also referred to as optimized for 16:9). The video will take full advantage of the pixels however the video will be out of Display Aspect Ratio or DAR (i.e. anamorphic).

Thus anamorphic means intended for 16:9 display. Thus your 4:3 anamorphic doesn't exist.

I can't make any sense out of the second sentence. Anamorphic video is out of DAR. Yeah, right. I guess he means 4:3 when talking about DAR?

When encoding non-square pixel or anamorphic video to MPEG-4, you have to pre-scale the image to the correct Display Aspect Ratio before encoding or you can set the PAR.

I'm not entirely sure what he means here. You sure have to resize your video (unless the size of your anamorphic video is exactly 16:9, or the size of your DAR 4:3 video is exactly 4:3). But you don't need to resize it to a 4:3 format. (Maybe that's different for hardware DivX players.)

What is the difference between PAR and DAR?
PAR is the Pixel Aspect Ratio and defines the shape of an individual pixel. DAR is the Display Aspect Ratio and defines the shape of the entire image.

A pixel has a shape?

[trevlac]

@SeeMoreDigital

Nice little write up ... but It's hard to take anything too serious that has a link to http://www.iki.fi/znark/video/conversion/ as it's only reference. :)

Don't get me wrong ... That Jukka Aho write up is great! Really helped me understand some things. But it is hardly a professional reference.

Did you notice that Aho links to Arachnotron as a reference! Damn, our own resident expert! :D

And Shit! Arachno's page links to ME! I'm the 1st to tell you that I don't know squat! Havn't made a cent off of my stuff, and probably never will. :)

Cheers

[Arachnotron]

quote:Originally posted by Arachnotron
1.5:1 (or 3:2) is correct for 720x480, but it isn't the aspect ratio; it is the pixel ratio (if that word exists).

Call it what you want ... 'Pixel Ratio' is a worthless value. I agree with Kika.

Here are some pixel ratios, 648x480, 640x480, 711x480, 720x480, 480x480.

You can calculate the pixel ratios if you want.


You quote me out of context, which is not really fair. :( Of course pixel ratio is nonsense, if you know the resolution you know the ratio so it is null-information.

I quoted this as an example of how not being clear about what value you are giving leads to further confusion. Kika assumed seemoredigital was talking about AR, in which case 1.5:1 would be wrong.

I would call this the convention. It IS 4:3 unless anamorphic is specified. If you have 720,704,352,480, assume the width is a factor of 702/711 for 4:3.

To quote you out of context :)

This is *your* way of dealing with AR. which in my opinion is just as confusing, since different resolutions end up having the same AR, as long as the user 'knows' he has to crop off some stuff and what. Other users may take AR literally, and the end-result is a mess.

I remain with my previous conclusion: By not stating what an AR value relates to - the window shape of the intended viewing device or the actual full frame in the video - you introduce an ambiguity about the actual shape of the frame.

When you use a PAR / resolution combo however, the resulting frame size is exactly defined. And yes, you do have to use some math. Tough luck :)

[edit]
here (http://www.thedigitalbits.com/articles/anamorphic/anamorphic235demo.html) are some examples of anamorphic DVD to illustrate what the discussion is about.

[trevlac]

Originally posted by Wilbert
A pixel has a shape?

Of course we know it doesn't. Probably the main fault of PAR. Gets everyone think the wrong way. But allas we are stuck with it.

If I keep talking ... I fear I will violate the rule about having nothing 'valuable' to add. :eek:

[Arachnotron]

If I keep talking ... I fear I will violate the rule about having nothing 'valuable' to add.

But we must... We as newbies have the holy duty to re-hash these ancient discussions to follow in the footsteps of our illustrious predecessors.

A foodfight once a year about PAR is obligatory.
:devil:

[trevlac]

Originally posted by Arachnotron
But we must... We as newbies have the holy duty to re-hash these ancient discussions to follow in the footsteps of our illustrious predecessors.

A foodfight once a year about PAR is obligatory.
:devil:

Maybe you should write "An introduction to Pixel Aspect Ratio" and link to it whenever this comes up.

Did you know Aho added you to his page? All of this is too funny! Next thing you know, I'm going to be the last one to know that YOU are Jukka Aho! I've always suspected this. :sly:

[SeeMoreDigital]

Originally posted by Wilbert
...Thus anamorphic means intended for 16:9 display. Thus your 4:3 anamorphic doesn't exist. Well I'm sorry but I have to disagree.

Any PAL DVD which contains a movie with an aspect ratio of 4:3 (1.33:1), contains an Mpeg2 stream which occupies all 720x576 pixels - all image, none matte.

Therefore it can be classed as anamorphic too!


Cheers

[Wilbert]

Any PAL DVD which contains a movie with an aspect ratio of 4:3 (1.33:1), contains an Mpeg2 stream which occupies all 720x576 pixels - all image, none matte.

Therefore it can be classed as anamorphic too!

Give me one reference which uses anamorphic in this way.