Delivering a 16:9 project using 4:3 footage

From JJBwiki

If you have any comments, questions, or suggestions, feel free to post them here.

Also see this tutorial: http://www.kenstone.net/fcp_homepage/faking_it.html

When people talk about widescreen video, they are usually talking about 16:9. This is the aspect ratio of HD, this is the only widescreen aspect ratio in which consumer and prosumer cameras can capture, and this is what popular editing programs such as Final Cut Pro expect when you tell them your footage is "anamorphic" (more on that later).

I was recently working on a project using a Canon GL2, which has 3 shooting options: Regular 4:3, Regular 4:3 + 16:9 guides, and 16:9 Anamorphic.

Regular 4:3: regular video like we've seen on television for years.

Regular 4:3 + 16:9 guides: White lines are put in the viewfinder that form a 16:9 frame. Then in post-production, you have the option to letterbox the image by putting black bars on the top and bottom of the frame, or render anamorphically when delivering on DVD (more on that later).

16:9 Anamorphic: This is the interesting one. Here, the camera captures a 16:9 image and lays it down on the entire regular 4:3 region. So as far as the tape is concerned, it is getting the same kind of signal. But really, the camera is stretching the image vertically.

I knew I wanted to do this project in 16:9. The only problem was deciding whether to acquire in 4:3 while framing the shots using the 16:9 guides, or to capture directly into anamorphic. I did some research and it seemed the consensus was that, in terms of quality, either way was equivalent. I was either going to have the camera stretch the image, or the editing software.

This was my first big project, and I decided to go the more conservative route: acquiring in 4:3, using the 16:9 guides.

In retrospect, I wish I had acquired in anamorphic. Here's why:

Anamorphic pros:

• No post-production image manipulation is needed. Just click "anamorphic" when importing and the rest of the process is seamless.
• Slightly higher quality colors and resolution. As demonstrated here, the camera scales the image BEFORE compressing it to lay it down on tape. This means that the camera can use significantly more data per pixel when doing so than your editing software does when performing the same task.
• You'll never forget to frame for 16:9. When acquiring in 4:3 with 16:9 guides, it's easy to forget to frame your shots for 16:9, especially when shooting a documentary. I had to vertically shift a lot of my shots in post-production (that process is described at the end of this article).

Anamorphic cons, with my retort:

• You don't have the option of delivering in 4:3 after the fact. This isn't really an option anyway. If you are framing all your shots with the 16:9 guides, your footage is going to look pretty lame if it's delivered in 4:3
• You don't have the option of choosing which anamorphizing (scaling) algorithm to use, you are forced to use the camera's. Well, this is very true. However, I have since read that Final Cut Pro's scaling is not very impressive [update: folks are saying that FCP5 has improved scaling. See this], and to get better results I should export my entire film into frames, distort with photoshop, and then reimport! That, or use some other third party tools. Both of these would make for a very cumbersome workflow, incur extra financial costs, and overall not be worth the trouble. So I am stuck with Final Cut Pro's scaling, which is probably as good or worse than the GL2's. Also see the note above regarding potentially higher quality colors.
• The GL2 does not compensate for the anamorphic image in the viewfinder or the swingout screen, so shots must be composed while viewing a vertically stretched image. I have no retort. There is no way around this and it is very annoying. Although... one could argue that since the image fills the entire viewfinder/screen, everything is bigger and it's easier to frame things up against the edges of the field of view. This is arguably more important than the image looking natural, when acquiring.

So that is why I think it is always better to acquire in anamorphic. But alas, I acquired in 4:3. Here is the workflow I use in Final Cut Pro.

FCP workflow for vertically shifting your video to get better framing:

If you acquired in 4:3 and are delivering in 16:9, chances are you probably want to move the image up or down to better frame your subject within the 16:9 area. In my case this was because of a few times when me or another camera operator forgot to frame the video within the 16:9 viewfinder guides, or just because I changed my mind in post-production.

This is easily done in Final Cut Pro. For this example I am going to assume that there is a timeline called Final which has many subclips in it, and another timeline called Final Wide, which only has Final dropped onto it, and then distorted, with the anamorphic flag checked in the Browser, as described above.

Open Final Wide and Final from the Browser. In the canvas, drag the Final Wide tab down below the canvas. You probably want to adjust your timeline accordingly. Here's how mine looks:

Flip between sequences in the timeline. Go through Final Wide until you find something you want to adjust. Now you can drill down to the clip in question by double clicking within the timeline(s) as many times as you need. Final Wide will happily remain in the lower right as you left it. When you get to the clip that needs adjusting, open it in the Viewer. Make sure that "Image + Wireframe" is selected. Now shift-drag the image up or down (holding down shift makes sure that there is no left-right movement). As you drag, you will see the changes "ripple up" to Final Wide.

Delivering non-anamorphic video as 16:9, for web and DVD:

Update: Compressor 3 is now considerably more user-friendly, and I have been using it for all of my compression. It also has nice presets for modifying video in just the way i am describing it here, without all the fuss of having to distort and then undistort. Here are the relevant settings:

If for some reason you are not using Compressor, then read on...

The main concept is that however you deliver, the "device" will take your anamorphic-4:3 video and squeeze it vertically into 16:9 (Or stretch horizontally, depending on how you look at it). When a DVD player sees your video, and the widescreen flag, it squishes the anamorphic video down. When you export to a Quicktime file, you make the the dimensions 16:9, and you effectively squeeze the anamorphic video down.*

Which means that at some point before delivery, you have to vertically stretch your video.

1. How? With the distort section of the motion tab in the viewer.
2. How much? Change the upper and lower Y coordinates from -240 to -320 and from 240 to 320, respectively. How do I know that this is the perfect stretch for 16:9? Because Boyd told me. I've done some math trying to figure it out for myself, to no avail. I do plan on doing this one day. It involves square vs. rectangular pixels. Maybe one day I'll understand the black art that is pixel width math, and write an article about it.
3. When? Immediately before delivery (compression). Example workflow: if your mega master mamma sequence is called Final, make a new sequence called Final Wide. Drag Final onto the Final Wide timeline. Right click on the resultant clip, and select "Open in Viewer". Now in the Motion tab make the Distort settings as mentioned above.
4. How do I deliver?
   • To export to Quicktime (or any other file for computer delivery): Here the compression software (not the playback software) compensates for the anamorphic video. Set up your export as you normally would, but change the dimensions to be anything 16:9. For example I often chose the Quicktime Conversion preset "Broadband - High". Then, the only thing I change is the dimensions, from 480x360 (4:3) to 480x270(16:9). When being exported the file gets scaled (and NOT cropped) to fit into those dimensions. The software does all the thinking about rectangular vs. square pixels for you.
   • To export to DVD: Here the DVD player does the anamorphic compensation. So the video is laid down on the DVD as 4:3 anamorphic. Send the sequence to Compressor as you normally would (and of course, set the widescreen flag when authoring your DVD). Again, you don't have to take any pixel ratio math into consideration, the software takes care of everything.

Some interesting information from an email exchange with Boyd Ostroff:

Here is something conceptual that confuses me: DV cameras acquire in 3:2, but consumer playback devices are 4:3, correct? How does this work out when delivering content anyway?

Shouldn't be an issue, and FCP uses some confusing terminology. The clue to solve this is that pixels on TV monitors are not square, they're rectangular such that 720x480 ends up scaling to 1.33.

And do we need to take this into consideration when stretching (when DVD players squeeze the image down to 16:9, do they expect the source material to be 4:3, or 3:2)?

No. 16:9 pixels are also non-square, but unlike 4:3 pixels which are taller than they are wide, 16:9 pixels are wider than they are tall. The only time this is a consideration is if you're displaying on a square pixel device, like a computer monitor or printer. In that case you would need to resize the image to 640x480 for 4:3 or 854x480 for 16:9.

*Some of you are thinking "when delivering to Quicktime, why stretch the video only to immediately squeeze it again?". I have not found a simple workflow for hacking off pixels (i.e., hacking off exactly the pixels in the black bars when using a 1.78 letterbox matte). If anyone knows of a slick way of doing this, let me know.