Framing for deliverables

Jayson Crothers · April 26, 2009

I'm rapidly coming up on the start of a RED feature here in LA and we just had a round-table meeting with the post house. Amongst many of the conversations, one that came up was whether we were shooting 16x9 or 2:1 (we're framing for 2:35:1) and how those choices would affect our 16x9 and 4x3 deliverable requirements. To be honest (and maybe it's the lack of sleep and excess of coffee) - it got me thinking and I'm not sure which is the better way to go. My first RED feature I shot 2:1 with a 2:35:1 matte, but the producers on that were very specific about not wanting to frame for anything else (strange to me, but who am I to complain about that?).

If we shoot 4K 16x9 with a common-top for the 2:35:1, then the 16x9 and 4x3 version should be relatively easy because we'd then just be removing the 2:35:1 matte for the 16x9 and the 4x3 would be cropping off the sides of that. If we shoot 2:1, then we have to do an extraction from within that to get a 16x9 (and a 4x3) and I wonder how well the image holds up when you start pushing in and enlarging it like that (I, of course, won't be able to test this).

I like the idea of shooting 2:1 so I don't have to worry too much about the bottom of the frame in regards to equipment (and with all our camera movement - dolly track is especially a concern), but the 16x9 is appealing because it gives me more control over the 16x9 framing.

Oh how I love composing for 3 films at once. =)

I know I'll probably wake up in the morning and the answer will be obvious, but I'm eager to hear other people's opinions or experiences.

David Mullen ASC · April 26, 2009

Well, 4x3 full frame would only be for SD broadcast, so I wouldn't worry about whether you are extracting it from 2:1 or 16:9 4K RAW.

As for shooting 2:1 versus 16:9 for a 2.40 movie that may also need a 16x9 HD full-frame version, again, I wouldn't lose sleep over it, I'm sure the quality of the HD version is going to be similar either way.

I shot 2:1 on my two RED features (framed for 2.40) simply because I heard 2:1 was less buggy a year ago, especially for editing in FCP using conversions to Quicktime. But I think those issues have been resolved more or less.

So I'd be inclined now to shoot 16:9 mode just to have access to more vertical info for making TV versions... but ideally with the 2.40 image being up higher in the 16:9 frame, not center extraction. If I was going to do center extraction, I'd probably stick to 2:1 mode, figuring that if I shot 16:9 mode for 2.40 center extraction, the sound man would have the boom dipping into 16:9 all the time and I'd end up defacto only having a 2:1 area to use anyway for home video. But at least with a semi-common topline approach, the boom could be closer to 2.40 and still protect 16:9.

Michael Most · April 27, 2009

If we shoot 2:1, then we have to do an extraction from within that to get a 16x9 (and a 4x3) and I wonder how well the image holds up when you start pushing in and enlarging it like that

You're not pushing in or enlarging anything. 4x3 is only a required deliverable for SD video release. Even if you create it from the HD video version, you're actually reducing the image, not enlarging it, because the HD frame is 6 times larger than the SD frame. It's basically a combination of creative cropping and scaling the image down.

As for the HD deliverables, this all depends on the post path for your DI. Depending upon how this is done, you're likely to either be working directly with the 4K R3d files as the source and rendering 2K files for the film recorder, or you'll be working with DPX files in either 2K or 4K form (most likely 2K). In either case, if the film release is anamorphic, you'll be cropping the sides for the 16x9 version (if you're not delivering letterboxed 2.35), which will likely result in a slight blowup for the HD video version. If you're working in a 4K environment - or you're working directly from the R3d files - you're doing an image reduction similar to what I described above for SD video. It all depends on where in the process the final RGB files are rendered.

MichaelB · April 27, 2009

I'm rapidly coming up on the start of a RED feature here in LA and we just had a round-table meeting with the post house. Amongst many of the conversations, one that came up was whether we were shooting 16x9 or 2:1 (we're framing for 2:35:1) and how those choices would affect our 16x9 and 4x3 deliverable requirements. To be honest (and maybe it's the lack of sleep and excess of coffee) - it got me thinking and I'm not sure which is the better way to go. My first RED feature I shot 2:1 with a 2:35:1 matte, but the producers on that were very specific about not wanting to frame for anything else (strange to me, but who am I to complain about that?).

If we shoot 4K 16x9 with a common-top for the 2:35:1, then the 16x9 and 4x3 version should be relatively easy because we'd then just be removing the 2:35:1 matte for the 16x9 and the 4x3 would be cropping off the sides of that. If we shoot 2:1, then we have to do an extraction from within that to get a 16x9 (and a 4x3) and I wonder how well the image holds up when you start pushing in and enlarging it like that (I, of course, won't be able to test this).

I like the idea of shooting 2:1 so I don't have to worry too much about the bottom of the frame in regards to equipment (and with all our camera movement - dolly track is especially a concern), but the 16x9 is appealing because it gives me more control over the 16x9 framing.

Oh how I love composing for 3 films at once. =)

I know I'll probably wake up in the morning and the answer will be obvious, but I'm eager to hear other people's opinions or experiences.

consider that you are windowing the chip when you shoot 2:1 and you still need to crop further to get 2.35. shooting 16x9 would give you more ability to pan and scan vertically while giving you max resolution. You could also use 16x9 ff for that deliverable rather than cropping from 2:35. same goes for 4x3. not sure how that would affect your framing, but it is similar to a film scan for 1:33 that uses more headroom than its 1.85 theatrical counterpart.

Edited April 27, 2009 by MichaelB

John Sprung · April 28, 2009

consider that you are windowing the chip when you shoot 2:1 and you still need to crop further to get 2.35. shooting 16x9 would give you more ability to pan and scan vertically while giving you max resolution.

My understanding of it is that Red makes 16:9 by shooting 2:1 and cropping the sides off. Is that incorrect?

-- J.S.

David Mullen ASC · April 28, 2009

My understanding of it is that Red makes 16:9 by shooting 2:1 and cropping the sides off. Is that incorrect?

No, the recorded 4K 16:9 mode is a taller area of the sensor than the 4K 2:1 mode, sharing the same horizontal area. The sensor itself is Super-35 in size and 16:9 in shape, 4520 x 2540 active pixels (24.4 x 13.7mm). 4K mode records 4096 pixels across, but the height is determined by whether you shoot 16:9 or 2:1 mode.

Originally people were shooting in 4K 2:1 mode because there were bugs with the higher data of 4K 16:9 mode (higher because it was a taller picture) in combination with problems of support from Apple in Quicktime with 4K 16:9 (have no idea why QT would be OK with 4K 2:1 though.)

You may be thinking of their 4K "HD mode" which is also 16:9 but is 3840 x 2160 pixels, basically double 1920 x 1080 in both directions, aka "Quad HD". That would be less wide than their normal 4K 4096 modes.