Re: [Frameworks] Super 8 Frame Size question

From: David Tetzlaff <>
Date: Wed, 5 Oct 2011 00:22:08 -0400

> I thought film was 4:3 just like SD video...

Wikipedia is your friend. The page below lists the gate sizes and both the aspect ratio of the recorded image AND the aspect ratio as projected for more film gauges than I ever knew existed:

The on-film aspect ratio of 16mm is 1.37 as is the projector ratio, thus a tad wider than 4:3. 8mm is 1.32 on film and 1.33 in projection (meaning there's some slight vertical masking, probably a good idea as providing a wee bit of cover for shifting framelines / small registration errors. Super 8 is 1.48 on film and 1.36 projected, so there's masking on the sides.

If you're getting any film transferred to SD, the aspect ratio of the video image is 4:3, period. Different codecs will record this in different ways, using different pixel grids. AVI is just a container, so that tells you nothing about the codec that was used to capture your film.

768x576 is PAL rendered in square pixels (it divides to 4:3). This is what a graphics program such as Photoshop would expect. So I would guess your 768x576 file is in some kind of codec mainly used for graphics or animation. Most of the codecs used for PAL _video_ though are 720x576. Thus the pixels are slightly wider than they are tall (as opposed, say, to DV-NTSC which is 720x480 -- which means the pixels are narrower than they are tall). I would assume the lab used the graphics codec to preserve the 18 fps rate rate, so one video frame = one film frame and the image runs in 'real time'. Since PAL video is 25fps, period, had the lab used a PAL video codec, your image would probably have been 720x576, and one video frame would equal one film frame, but it would be running a good bit faster than 'real time.'

Google is your friend... pretty much everything you could ever want to know about pixel grids is explained in detail here:

> I'm going to use apple compressor to set it at 25fps for editing so I just wondered if while I'm using compressor it was worth me scaling it down to 720 x 576 (if that is the original frame size)?

If you're using any PAL preset in Compressor, it should scale the image to the right pixel grid by default, which will probably be 720x576.

The real questions are:
1. What codec did the lab use?
2. What video codec do you want to work in for editing?
3. How are you going to handle the frame rate?.

Regardless of #1, if you're working on a Mac the answer to #2 is probably ProRes 422 standard (not HQ, which is overkill).

If you want the footage to play at the same speed as the original 18fps film, you'll be slowing it down, adding pull-down and/or blending frames. The video frames will no longer correspond to the film frames, but that's just the way it goes. Individual frames of the output may look weird when you view them as stills, but in motion they'll be fine. Compressor has different settings for how to do speed changes and frame blending, and does NOT default to the best quality, so you'll probably want to go in there and edit the presets.

Going back to the video transfer, realize that there are possible variations of how your Super 8 image gets into that 4:3 SD rectangle. Unless you can see the framelines at the top and bottom of the screen, something is getting cropped. The video image might be centered in the wider rectangle on the celluloid, or it might be skewed to one side. It might have the whole height of the frsme, top to bottom, but more likely is zoomed in a bit to hide frameline shifts / registration errors. But how much exactly? Nothing new here, though. In both actual film and video on CRT displays the edges of the frame are not precise, and will vary somewhat from one projector or monitor to another. A window in a computer program such as FCP generally shows you the whole frame, but even in flat-panel digital TVs and digital projectors, there's still usually some small bit of cropping.
FrameWorks mailing list
Received on Tue Oct 04 2011 - 21:22:26 CDT