You don't need to crop away black bars in Final Cut. However, 2048/1080 at 1,33:1 would be 1436/1080, cropping would need to be 306 from both sides.
You can crop in Final Cut to 1436/1080 and put this into a flat container - DCP-o-matic will fill the sides with black. Or, you can import 2048/1080 into DCP-o-matic, and crop away 25 pixels from each side, resulting in the same 1998/1080 container with an embedded 1436/1080 active image. You should always use 'to fit DCP'. No need to force the aspect ratio to a specific value if it is obvious and matches the actual pixel raster.
In both methods, there is no scaling applied, only cropping, so you get a 1:1 pixel representation in the DCP.
Always look at the pixel processor workflow indicated below 'Range': It will tell you exactly what DCP-o-matic is doing to your image, and the simpler the description, the better (usually).
You do not have the required permissions to view the files attached to this post.
I've just checked by exporting a frame, and the aspect ratio appears to be 1470 x 1080. It would appear I need to crop 287.5 pixels from each side? If I leave the image uncropped, I also see very slight black bars to and bottom, as well as those on each side. If I crop 25 pixels either side as suggested, the top and bottom black bars disappear.
Many thanks for the tips. Also, should I use Range: Full (Jpeg, 0255) or Range: Video (MPEG, 16-235)?
Sample frame:
Last edited by Paradisio on Sun Oct 23, 2022 8:53 am, edited 2 times in total.
When I am crazy about frame size, I usually just load the video into VLC and create a video snapshot (VLC menu command). That outputs a single image at source resolution. One should cue to a position in the video that has bright borders on all four sides, so you can see frame borders clearly. I then create a selection in e.g. Mac Preview, Photoshop, etc, that gives me exact numbers.
Usually, a few black pixels on top or bottom go unnoticed in a cinema. It's probably more important that at least one dimension of the image fits the J2K container resolution, while the other is cropped. That means that the image is not scaled by a small factor, but translated 1:1. If you crop e.g. 2 pixels from bottom and top, the result is 1076, and would usually be rescaled to 1080. That I would avoid. DCP-o-matic scales images nicely, but if possible, avoid it and try to set the original pixel raster within the DCP container pixel raster.
It's not that the image looks bad after rescaling. It's just good practice not to introduce scaling if unnecessary.
