Unsqueeze Anamorphic and Export?

I followed this guide to unsqueeze anamorphic Pro Res footage.

Create a new sequence, drag a clip onto the timeline and let FCP change
the sequence settings to match the clip settings. This will create a sequence
with a frame size of 2048 x 1536 pixels.
Go to the sequence settings and assign pixel aspect ratio HD (1440 x 1080)
and check Anamorphic 16:9.

However, when I export it using current settings, I end up with the original squeezed image (4:3).

In my timeline and canvas window it's unsqueezed and looks correct but when I export it seems those settings don't translate for some reason.

Am I missing a setting?

If you're viewing it in Quicktime Pro it might be showing you what it really is, which is what you're seeing. Anamorphic settings really just tell your device how to display the footage, they don't actually change the pixels. This is all normal and correct. There's an option in QT Pro to view the footage as widesceen. If you turn this on it should all be OK.

The FCP7 Sequence Settings for Pixel Aspect Ratio are a bit weird.
There six choices given. Each one means a pixel widening factor:

1. "Square" means factor = 1
2. "NTSC - CCIR 601 / DV (720x480)" means factor = 0.8889
3. "PAL - CCIR 601 (720x576)" means factor = 1.0667
4. "HD (960x720)" means factor = 1.3333
5. "HD (1280x1080) means factor = 1.5
6. "HD (1440x1080)" means factor = 1.3333

And there is also a checkbox, which also means a pixel widening factor:

• "Anamorphic 16:9" means factor = 1.3333

Notice that #4 and #6 in the list give the same factor = 1.3333. Whether you want to make HD (16:9) from 960x720 or from 1440x1080 you use exactly the same pixel aspect ratio.

When "Anamorphic 16:9" is unchecked you get the pixel indicated by the widening factor #1-#6 chosen. But when "Anamorphic 16:9" is checked it becomes a second widening factor multiplied by the first factor.
There are altogether 9 pixel aspect ratios you can make:

0.8889:1 by using #2 without anamorphic checked
1.0000:1 by using #1 without anamorphic checked
1.0667:1 by using #3 without anamorphic checked
1.1852:1 by using #2 with anamorphic checked
1.3333:1 by using either #4 without anamorphic checked or #6 without anamorphic checked or #1 with anamorphic checked.
1.4222:1 by using #3 with anamorphic checked
1.5000:1 by using #5 without anamorphic checked
1.7778:1 by using either #4 with anamorphic checked or #6 with anamorphic checked
2.0000:1 by using #5 with anamorphic checked

Joe Riggs made pixels 1.7778:1. His 2048x1536 video displayed as 2.3704:1. Hopefully that was how it was shot to display.
To really de-anamorphise this video is to transform it to square pixels. It could be transformed to 2048x864, but this means severe downrezzing in the vertical direction. It could be transformed to 3641x1536, with no downrezzing. ProRes and FCP7 can handle this.

Non-square pixels are trouble. The .mov file exported from Joe Riggs' timeline includes no pixel aspect ratio metadata, so QuickTime can't possibly play it correctly until you add some metadata. You must add scaled size information in square pixels to the .mov file properties.

Dennis Couzin
Berlin, Germany

Anamorphic 16:9 is a 16/9 image recorded into, and then unpacked from from a 4/3 frame.

what you are dealing with is a 2.40:1 image squeezed into, then un-squeezed from a 16/9 frame.

similar concept, but old FCP7 has no pre-sets to deal with it.
to work in FCP7 with this material you need to create 16/9 rushes letterboxed to 2.40

you can fudge it by dropping your image into a 16/9 sequence, then distorting it till it looks correct.
i dont know what the mathematics of it are.
i did do this once, but the project files are not close to hand, and i have not remembered the numbers.
i most likely got it so it looked about right, then looked at the number to understand the maths, then refined them


do you have resolve? it may well have presets for this.
you would need the Pro Verson ($1K)
the free version is limited to dealing with HD frame sizes.


nick

I've gone the resolve method before but I don't want to transcode if it can be avoided,
the footage is Pro Res but squeezed.

I'm following the instructions from the whitepaper, which I listed in my first post.
By following that, I do get a timeline with the image properly desqueezed and I don't have to render!

The weird thing is when I export with current settings, I get more of a 4:3 image.

Instead of exporting at "current settings", I tried 1080p and "Anamorphic"
These resulted in a wider image, closer in look to the timeline but still squeezed a little bit.

I'm not going to pretend like I comprehended what dCouzin wrote but maybe the answer is in his post.

Quote
Nick Meyers
what you are dealing with is a 2.40:1 image squeezed into, then un-squeezed from a 16/9 frame.

Nick meant to write "a 2.40:1 image squeezed into, then un-squeezed from a 4/3 frame".

Joe's original 2048 x 1536 pixel image, when displayed with square pixels, is a 4:3 image. When displayed with non-square pixels, it can have any proportions at all. The FCP7 Sequence Settings give nine choices, between 0.8889 times as wide and 2.0000 times as wide. Joe chose 1.7778 times as wide. 1.7778 times the original's 1.3333 equals 2.37, very close to 2.40.

Instead of offering a dropdown menu with six presets and an overdrive checkbox marked "Anamorphic 16:9", yielding altogether nine choices, FCP7 could have sensibly let you choose the pixel aspect ratio as a number. It's hardly any work for FCP7 to make (simulate) unsquare pixels for display. That's why it doesn't require rendering. Unfortunately, the exported .mov file lacks the unsquare pixel shape information. Jon Chappell, our expert on the .mov file format, might explain why this is missing.

2.35:1 and 2.39:1 are standard screen shapes for theatrical movies. 2.37:1 is a good compromise. If you have a 4:3 video intended to be unsqueezed to 2.37:1 the 1.7778 pixel factor is your ticket. If you have a 16:9 video intended to be unsqueezed to 2.37:1 the 1.3333 pixel factor is your ticket. FCP7's sequence settings offers both of those -- with two different ways to do the first and three different ways to do the second! The editor, after struggling with weird sequence settings, sees the picture correctly. The troubles come downstream.

There are visual theories that horizontal resolution is less critical than vertical resolution, so anamorphic recording and storage is smart, but more and more, video is becoming a square pixel medium. Unless you're sure the next person, or device, downstream will apply the right pixel widening factor, better to make square pixel video.

This does require rescaling. Compressor's geometry pane sets this up intelligently. You can make the true image proportions or else letterbox/pillarbox to fit to standard video proportions. Compressor does good quality rescaling with the Frame Controls resize filter set at Best. For big jobs, Compressor runs fast by using all processor cores efficiently with QMaster.

Dennis Couzin
Berlin, Germany