The sensor on 5D mk ii is 5616x3744. When recording the camera crops the image to a 5616x3159 (16:9), scaling it down to a 1920x1080. As I understand the format of the final file will always be 1920x1080... but...
Would it be possible record 5616x2390 (2.35) and scale down to 1920x1080?
This would enable the user to rezise x-axis to a 2538x1080 keeping more information in the image (since no post-cropping is needed), and also not waste any compression on things you plan to crop away in post.
Note, Im not asking for a different output than 1920x1080... just different reading from the cmos?
Probably won't gonna happen...
Also you would loose quality because of the resizing...
You are probably right, but if it was possible you would get alot higher quality (817->1080) on the y-axis if you plan to use the end result as 2.35 ratio
just turn on crop marks in overlays. use the built in cinesco2 or make your own.
Quote from: poromaa on July 30, 2012, 01:46:13 PM
The sensor on 5D mk ii is 5616x3744. When recording the camera crops the image to a 5616x3159 (16:9), scaling it down to a 1920x1080. As I understand the format of the final file will always be 1920x1080... but...
-deleted-
Btw it's 'line skipping', not 'downscaling'. ;)
QuoteBtw it's 'line skipping', not 'downscaling'.
Ok, thanks! But the idea would then be something like skipping fewer lines between every recording of the 1080 lines. (scanning a 2.35 cinescope, rather than scanning the whole 16:9 letterbox)
Quotejust turn on crop marks in overlays. use the built in cinesco2 or make your own.
Well, when you do, the movie is still recorded with 16:9 frame, and in post you will have to throw away 263 lines (1080-817). The result would get lower resolution than if you do it the way I have described above. My proposed idea (if it is possible to implement) would instead use all the 1920x1080 pixels to cover the 2,35 cinescope output. It would be like having a anamorphic lens on the camera without having the hardware.
It would not be like having a real Anamorphic Lens on your Cam, it's the same as cropping in post (only that you lose vertical information with which you could better stabilize or rearrange the image).
As said, chainging the line-skipping process might never be possible, and even if you skip less lines, the Quality can not improve.
It's simply less lines because less image, within the scanned area it's the same process unless you want some heavy distortion
The quality could improve (if you do the math) but as you say probably impossible to control what lines to include from the cmos read.
Something to consider about cropped videos:
http://www.magiclantern.fm/forum/index.php?topic=2449.msg20208#msg20208