Aspect ratio edditing?

[eadsja]

Allright, I've been through the guides, I can resize my video to the wondorous 1:1 square pixel aspect ratio,

But what I don't enirely understand is why we want to do this. People say I've got two options for edditing - Staying in 720X480 or moving it to my new square resolution (in my case it is 856X480) and edditing from there.

If I do the square pixel ratio, does it improve quality for PC distribution or what? I know it all has something to do with the effects I do (is this meaning that it effects my transition effects - or specialty graphic effects I overlay onto the footage?)

Does moving it to square pixel ratio make it easier to work with or render and compress? How much quality am I going to loose for my master copy of the video?

And do most people resize their master video again back down to 640x480 video for distribution? Does having the 1:1 pixel aspect ratio make it easier to do this final re-sizing?

The guides are all like - Do this so this can happen... But they don't quite explainw why. If it is a matter of quality, cool I understand that, if it is a matter of ease of edditing - I get that too. But If I do all this crazy re-sizing, how much video quality am I going to loose?

That, and what exactly qualify's as an effect? Are these my transitions and title screens? Or are they like, computer effects graphics I insert that are translucent?

Okay, I know it's a bunch of noob questions, but bear with me, I am trying to hash it all out.

[Qyot27]

If I do the square pixel ratio, does it improve quality for PC distribution or what? I know it all has something to do with the effects I do (is this meaning that it effects my transition effects - or specialty graphic effects I overlay onto the footage?)

Does moving it to square pixel ratio make it easier to work with or render and compress? How much quality am I going to loose for my master copy of the video?

All the aspect ratio-related resizing is meant to make the video look like it's meant to look on a TV screen. Many editors find it easier to edit anamorphic footage at 720x480 and then resize out (or down) to the proper ratio when they're finished. However, when they do that, any effects (and yes, this is referring to CGI-type graphics rendering, whether used as overlays or used as transitions, not crossfades or simple transitions like that) need to be done with the final ratio in mind, or else they'll look way too horizontally stretched.

And do most people resize their master video again back down to 640x480 video for distribution? Does having the 1:1 pixel aspect ratio make it easier to do this final re-sizing?

Most people do work with 4:3 or 16:9 ratio in their distribution releases (along with other widescreen ratios which aren't explicitly 16:9). It doesn't necessarily mean 640x480, though. 512x384 is another popular framesize. For widescreen videos, widths of 720, 640, 512, and 480 are all used commonly. The 1:1 PAR doesn't make it easier or harder, it just preserves the way the image looks.

This, of course, is coming from someone who doesn't care about having 1:1 PAR or 4:3 display size. I usually work in 0.9 PAR and 3:2 (I guess its not that strange considering I do my editing for use on disc media), and have been known to distort 4:3 footage intentionally by resizing it to 16:9 without cropping.

The guides are all like - Do this so this can happen... But they don't quite explainw why. If it is a matter of quality, cool I understand that, if it is a matter of ease of edditing - I get that too. But If I do all this crazy re-sizing, how much video quality am I going to loose?

As long as you're using highly precise resizing algorithms with the master file (read: Lanczos or Lanczos4) and resize down - or not - when doing your distro encode, not much, if any. Those types of quality issues usually arise from the type of compression you decide to use.

[eadsja]

Okay, everything you just told me did one of those wonderfull little brain 'clicks' and all the complicated stuff just started to connect...

But that brings forward more questions. (arg! The agony, solutions lead to more questions! must-stop-learning-now...)

Okay, the reason we do all the resizing is to format the video to display best on our chosen media (I.E. Computer monitors in our case). We also do initial formating so once edditing is over, transitions to distribution resolutions.

Pixel Aspect Ratio is the density of color pixles towards for the whole image.

Display aspect ratio is the screen resolution reduced to the lowest common denominator.

To create the best picture quality for my video's I essentially have to convert BOTH my PAR and my DAR from DVD footage into something that when I compress it to the final desired resolution there are no unwanted effects or picture quality loss. In other words, with all of this, I am trying to create the truest "translation" of the DVD footage into a movie file to be played on a TV Screen.

Now - The guide talks about converting Pixel Aspect Ratio and Display aspect ratio and how to convert as if they were two seperate entities. But, don't we have to convert both to maintain picture quality? When you change a video's PAR don't you mess up the DAR? And when you change format the DAR you do nothing with the PAR?

The 16:9 DAR is used for widescreen distribution.
The 4:3 DAR is used for computer and TV distribution.
(What is 1:1 DAR?)

We put things into 1:1 PAR only to maintain image quality while we are edditing, and then we move it back to our desired PAR for the appropriate distribution media (NTSC or PAL). Or we leave the 1:1 PAR if we are going to use a complete squar DAR like 360X360 (right?).

Theory: In order to maintain picture quality when building a video, we first move the PAR to where we want it. Eddit the video and outsource. Using the 1:1 PAR we can then change teh PAR directly into the PAR needed for the distribution format. With this new PAR product when then change the DAR to meet the desired resolution.

How am I doing so far?

-Current Project Problem-

Let me explain what I am trying to do.

1.) Convert my SOURCE to PAR 1:1
2.) Eddit and create the video, outsource using PAR 1:1
3.) Change the Par to Desired outsource (NTSC) -High video quality
5.) Worry about the 24FPS somewhere in here...
4.) Lower the resolution to a distributable widescreen format for Xvid (which I assume is the standard distribution codec)

The Video I am using is The Studio Gibili film The_Cat_Returns: Here is where I am having a bunch of headaches.

The DAR is 16:9 (At least that's what

But on the box it says - WideScreen(1.85:1) [which I belive is the Pixel Aspect Ratio]
My DGI Index tells me that the video is true progressive FILM at 23.97 FPS and that it is wide screen at 16:9 (Still at the 720X480 resolution).

The (1.85:1) ratio is throwing me for a loop. I can't seem to properly crop and resize the video and the settings that the Resize Calculator is still leaving two thin black bars on the left and the right of the screen when I resize to (854 or 856X480 -Numbers given to me by another AMV'er on this site) and the AMV App is telling me to crop the left and right side of the screen by 12 each. But when I start manually trying to correct the cropping, it either totally ruins the video quality or those blasted black lines are still there.

Considering the footage is FILM (not interlaced) is there a work around where I re-size the WHOLE image (both width AND hiegth). The guide seems paranoid about even touching the heighth as a rule of thumb to prevent interlacing botch jobs. But considering I am working with true progressive film, wouldn't it make more sense just to resize the whole image?

Last question, Am I approching this in the wrong method?

[Zaphod_Beeblebrox]

There's quite a lot here to answer, so bear with me...

Now - The guide talks about converting Pixel Aspect Ratio and Display aspect ratio and how to convert as if they were two seperate entities. But, don't we have to convert both to maintain picture quality? When you change a video's PAR don't you mess up the DAR? And when you change format the DAR you do nothing with the PAR?

The PAR and DAR are more or less unrelated from what i understand, if you change the DAR your picture looks wierd, and changing the PAR wrongly can make your footage look botched too, but if you change one, the other does not change, changeing PAR does not affect DAR because all that has changed is the shape of the pixels, not how many there are.

What is 1:1 DAR?

That would be a completely square picture.

We put things into 1:1 PAR only to maintain image quality while we are edditing, and then we move it back to our desired PAR for the appropriate distribution media (NTSC or PAL). Or we leave the 1:1 PAR if we are going to use a complete squar DAR like 360X360 (right?).

Re-changing your PAR back to NTSC/PAL is only of any use if you plan to submit that version of you vid to a con, if not, leave the PAR 1:1 so it looks right on computers screens, never, ever use square DAR, that will look so wrong it isn't even funny, the 1:1 PAR is just the PAR used by computer monitors, and therefore should be the PAR of your online distro version.

-Current Project Problem-

Let me explain what I am trying to do.

1.) Convert my SOURCE to PAR 1:1
2.) Eddit and create the video, outsource using PAR 1:1
3.) Change the Par to Desired outsource (NTSC) -High video quality
5.) Worry about the 24FPS somewhere in here...
4.) Lower the resolution to a distributable widescreen format for Xvid (which I assume is the standard distribution codec)

That's pretty much ok, except for an online distro version of your vid you should leave the PAR as 1:1 as that is the PAR of computer monitors.

The Video I am using is The Studio Gibili film The_Cat_Returns: Here is where I am having a bunch of headaches.

The DAR is 16:9 (At least that's what

If it's a cinema film thats probably true.

But on the box it says - WideScreen(1.85:1) [which I belive is the Pixel Aspect Ratio]
My DGI Index tells me that the video is true progressive FILM at 23.97 FPS and that it is wide screen at 16:9 (Still at the 720X480 resolution).

The box is right, that's the DAR, simple maths tells us that 1.85:1 and 16:9 are close anough to not make any real difference for online distro.

The (1.85:1) ratio is throwing me for a loop. I can't seem to properly crop and resize the video and the settings that the Resize Calculator is still leaving two thin black bars on the left and the right of the screen when I resize to (854 or 856X480 -Numbers given to me by another AMV'er on this site) and the AMV App is telling me to crop the left and right side of the screen by 12 each. But when I start manually trying to correct the cropping, it either totally ruins the video quality or those blasted black lines are still there.

You shouldn't need to resize it horizontally until you've done your editing and are ready to encode to XviD, leave the resolution as-is for editing and worry about resizing afterwards.

Considering the footage is FILM (not interlaced) is there a work around where I re-size the WHOLE image (both width AND hiegth). The guide seems paranoid about even touching the heighth as a rule of thumb to prevent interlacing botch jobs. But considering I am working with true progressive film, wouldn't it make more sense just to resize the whole image?

True, you could resize the whole image, but there is no need unless you plan to do some supersampling in pre-processing your footage.

Last question, Am I approching this in the wrong method?

It's not so much the wrong method, but you shouldn't need to resize your footage before editing if the resolution still comes out fine in DGindex.

Whooo, what an answer...i think i'm right, if anyone feels they need to disagree please do.

[Keeper of Hellfire]

Seems like you did't understand much from reading the guides.

The PAR you can't change. It's given by a) the display hardware and b) standards. If PAR of display device and standard matches, all is fine. Otherwise some transformation has to be done to keep the DAR. In case of DVD Video this transformation is done by the DVD-player (no matter if hard- or software). For AMV it's your task to keep the correct DAR for the intended display device by resizing.

You should not resize your footage as long as it is interlaced, because the scanlines don't relate any longer to pixel lines. If your footage is progressive or you have already done deinterlacing, resizing is no issue.

Even if it is possible to edit with the wrong DAR, it's easier if the DAR is corrected. Otherwise you have already to take in account that your effects will be squeezed or streched if displayed with the correct DAR. To scale it down or to resize after the editing is done, before compressing, isn't that big issue.

when I resize to (854 or 856X480 -Numbers given to me by another AMV'er on this site) and the AMV App is telling me to crop the left and right side of the screen by 12 each.

You want to encode with Xvid. Xvid needs a multiple of 16 in width and height to do its job well. 856 isn't a multiple of 16. The next are 848 or 832, so you have to crop 4 or 12 each side.

[shirohamada]

thats a whole lot of words.
i do it like,
crop->denoise->limitedsharpen()->downsize