PDX10 16:9 footage in Final Cut Pro

[Juan Parra]

<<<-- Originally posted by Boyd Ostroff
If you size your video with less than 480 vertical lines (NTSC) then you're losing an important advantage that the PDX-10 has to offer and will get results more comparable to what you might expect from a VX-2000 in widescreen mode. Your choice: why not try a test both ways and see which you prefer? -->>>

i've been reading information on the matter in the last 24 hours, and this is what i've found:

the key to all these calculations is the pixel aspect ratio (PAR).

NTSC has different PAR for 4:3 and 16:9.
4:3 is 0.9091 (pixels are taller)
16:9 is 1.2121 (pixels are wider)

pixels on the computer screen are square.

so in order to represent the correctly the non-square NTSC pixels on the computer screen we must resize the video picture accordingly.

see the excellent link for more explanation on this matter:

http://www.rottentomatoes.com/vine/showthread.php?t=351788&goto=nextnewest

based on that, there is no or little resolution lost when resizing the video. am i right? i don't see a need for "resampling".

the other important considerantion is where you want the output or format delivery.

- it's clear that for 16:9 DVD delivery, we must keep anamorphic video 720x480 for correct display.

- now it we want to output our maximum full resolution 16:9 using windows media or quicktime what size should we use? 854x480, 872x480 or 720x405?

juani

[Boyd Ostroff]

<<<-- Originally posted by Juan Parra :the key to all these calculations is the pixel aspect ratio (PAR).

I agree

<<<-- based on that, there is no or little resolution lost when resizing the video. am i right? i don't see a need for "resampling".

We may be getting tangled up with semantics here. You aren't "losing" anything because you are increasng the width. You would have to decrease one of the dimensions to "lose" anything. A couple posts back someone noted that in stretching the width from 720 to 854 you are creating something out of nothing (those 134 pixels had to come from somewhere) and this may introduce something undesirable into the image. There's something to be said for that I guess, by my practical experience is that it shouldn't be much of a problem. But by definition, this process is indeed "resampling".

- now it we want to output our maximum full resolution 16:9 using windows media or quicktime what size should we use? 854x480, 872x480 or 720x405?

First, it would seem unlikely that you'd want to do this, isn't it? Maybe I just can't visualize the application however? Where did the 872 number come from, you lost me there? It would not be correctly proportioned on a square pixel device.

720x405 avoids having to stretch the width of the image, but it throws away 75 lines of vertical resolution.

I think it all comes down to what you're trying to accomplish. I would use 854x480 if for some reason you want the highest quality display on a computer screen. But to reiterate: "why not try a test both ways and see which you prefer?"

[Juan Parra]

<<We may be getting tangled up with semantics here. You aren't "losing" anything because you are increasng the width. You would have to decrease one of the dimensions to "lose" anything. A couple posts back someone noted that in stretching the width from 720 to 854 you are creating something out of nothing (those 134 pixels had to come from somewhere) and this may introduce something undesirable into the image. There's something to be said for that I guess, by my practical experience is that it shouldn't be much of a problem. But by definition, this process is indeed "resampling".>>

yep i meant to say that. losing or missing information.

<<First, it would seem unlikely that you'd want to do this, isn't it? Maybe I just can't visualize the application however? Where did the 872 number come from, you lost me there? It would not be correctly proportioned on a square pixel device.>>

the 872 comes from the link. please read carefully and tell me what you think.

http://www.rottentomatoes.com/vine/s...oto=nextnewest

<<720x405 avoids having to stretch the width of the image, but it throws away 75 lines of vertical resolution.>>

though the pixels have to be resized either way in order to get the right aspect ratio. so can we talk about missing or losing lines of resolution if we need to resize the anamorphic video anyway?

<<I think it all comes down to what you're trying to accomplish. I would use 854x480 if for some reason you want the highest quality display on a computer screen. But to reiterate: "why not try a test both ways and see which you prefer?">>

boyd, it's not why i prefer, it's what in fact is the highest resolution that can be displayed on a computer screen. that's what i want to know. so the one with the highest resolution will look best.

juani

[Boyd Ostroff]

I read through that link, and don't follow his logic, but I'm always happy to admit I'm wrong when I miss something. I don't know where he gets the 1.2121 figure. Let's just do the math which is really simple.

The computer screen pixels are SQUARE so we can ignore the whole pixel aspect ratio thing. You want the image as displayed to have the proportions of 16:9. Now 16 / 9 = 1.778. We know that the anamorphic image is 480 pixels high. To find the width multiply 480 x 1.778 = 853.44. I rounded up to 854 since it seems nicer to deal with round numbers when scaling is involved.

That's a cute kitty on his page (I have a couple of my own). One problem I find with the examples is that we don't know how fat or skinny his cat is, and it doesn't illustrate his point very well IMO. A scene with more universally recognizable objects might have been more helpful, but I'm just quibbling I suppose.

All that crosses my mind in relation to his 872 width would be 872 / 480 = 1.8167:1 aspect ratio. I believe that 1.81:1 is a common 35mm film format, so maybe he got confused? Why not write and ask him to explain if this concerns you. But personally I don't see how it could be right.

>> though the pixels have to be resized either way in order to get the right aspect ratio. so can we talk about missing or losing lines of resolution if we need to resize the anamorphic video anyway? <<

Sorry, you lost me there...

>> boyd, it's not why i prefer, it's what in fact is the highest resolution that can be displayed on a computer screen. that's what i want to know. so the one with the highest resolution will look best. <<

That isn't quite what you asked before, and I would have to answer that it depends on the size of your screen. Using my PowerBook G4 with a native LCD resolution of 1280x 854 I often display video at 1280x720 full screen and it looks great. There's a little Mac shareware program called BTV pro that will let you display video any size and proportion desired. Works nicely as a "poor man's monitor" in the field.

On my desktop machine I have a CRT monitor running at 1600x1200 resolution. I have rendered Quicktime files at 1600x900 and they looked very cool.

I have an LCD widescreen monitor that I use for editing. The native resolution is 1280x768. When you feed it either s-video or component video it rescales the image in hardware to fill the screen.

In my mind all of these are valid answers to your question. But the important principle to me is that the VERTICAL RESOLUTION should not be any less than 480. If it is then you've thrown away some of your data.

I get the feeling I'm starting to repeat myself here. Sorry if you don't agree or understand, but I think this is about all I can offer on the topic. Hope it helps you in some way...

[Juan Parra]

<<The computer screen pixels are SQUARE so we can ignore the whole pixel aspect ratio thing. You want the image as displayed to have the proportions of 16:9. Now 16 / 9 = 1.778. We know that the anamorphic image is 480 pixels high. To find the width multiply 480 x 1.778 = 853.44. I rounded up to 854 since it seems nicer to deal with round numbers when scaling is involved.>>

that computation is very obvious, but remember that the video pixels are NOT square. for instance, when we display a video grab on a computer screen we must resize it so its proportion will display accordingly.

the 872 comes for multiplying 720 by 1.2121 (PAR for 16:9).
my guess the reason this is done is to compensate for the NON square pixels so the grab will display without distorsion on the computer screen.

btw, don't get me wrong i'm not completely sold on that, it seems to me something else is still missing.

>>>> though the pixels have to be resized either way in order to get the right aspect ratio. so can we talk about missing or losing lines of resolution if we need to resize the anamorphic video anyway? <<<<

sorry, i meant to say if the anamorphic video 720x480 must be resized in order to display it properly in a computer screen, then how can we talk about losing or missing lines of resolution if that step (resizing) has to be done anyway? if we keep 480 and resize 720 -> 854 and we miss lines of resolution ("resampling" is needed), and if we keep 720 and resize 480 -> 405 we lose lines of resolution.

<<On my desktop machine I have a CRT monitor running at 1600x1200 resolution. I have rendered Quicktime files at 1600x900 and they looked very cool.>>

that's still very subjective. have you tried projecting that image in very big screen?

<<In my mind all of these are valid answers to your question. But the important principle to me is that the VERTICAL RESOLUTION should not be any less than 480. If it is then you've thrown away some of your data.>>

i'm still not sure about that (see above).

anyway, i didn't mean to make you sweat. it's all good.
though i'd like somebody else to step up.
cheers

juani

[Boyd Ostroff]

No sweat, but I think you've failed to grasp some basic concepts here. The sensor in the camera that captures the image should be in the proportion of 16:9. When you view that image it must also fit that same proportion. What happens in between is not really relevant. Video pixel aspect ratios don't matter because we're talking about a display with a 1:1 pixel aspect ratio. Really now, don't you think there's a REASON that plasma and LCD home theatre screens are either 1280x720 or 854x480? (Well, OK, some are 1280x768 but that seems to have to do with manufacturing processes, and they make it work by chopping off the sides with overscan).

>> that's still very subjective. have you tried projecting that image in very big screen?

Yes, we used anamorphic DV in an opera on a 44' wide screen with a 10,000 lumen Barco projector. It looked surprisingly good.

I'm sorry, most of your questions ARE subjective. Doing experiments is a valuable learning experience. It's how you can form YOUR OWN opinion. Don't always look for the easy way out. And don't just blindly follow whatever some guy says on a website, whether it's me or a "film student" named "Knellotron". Really, it isn't very hard to do a quick test and see what you think. Learn to trust your own judgement, form your own opinions; especially when it comes to any sort of creative/artistic venture.

Which flavor of ice cream tastes better? Chocolate or strawberry?

[Boyd Ostroff]

<<<-- Originally posted by Juan Parra : have you tried projecting that image in very big screen? -->>>

One further thing that I should add.... if you're projecting or viewing on a widescreen TV you should not mess with the frame size at all. These devices will have hardware image scalers that do a very nice job when fed raw anamorphic 720x480.

[Sean McHenry]

My last comment, I think will be this:
I think I read that a standard 4:3 image has a PAR of 1. Square pixels are as tall as they are wide so the ratio is 1:1 for the actual pixel dimensions.

In anamorphic land, you take an image that's wide and through optics, squash it horizontally to those square pixels. To play that image so it looks right, on the playback end of things, you stretch the square pixels horizontally to get the original SCREEN aspect ratio to display correctly. I always heard (I think) that the ratio was 1.33 . That is, each of those originally square pixels is stretched sideways by about 1/3 wider than their orignal size.

If you think about this, any 45 degree or so, or even more so, any nearly vertical straight lines filmed in anamorphic, when played back and blown up sufficiently, like on a projected movie screen, should show increased but elongated stairstepping.

It really seems to come down to manipulation of the available pixels width.

One of the reasons I went with the PDX10 is the very high pixel count in 16:9. With more pixels per inch on the display, the stretching should be less noticable. Alas, this is a false assumption because as we know, the DV standard sets the displayed pixel count, not the CCD imagers pixel count.

Anyway, what were we talking about?

Sean