Physically crop sensor

J

jmanord

New member
I seem to remember someone cropping the camera sensor to achieve a more anamorphic aspect ratio but can not find the post. Does anyone remember the method (I thought it might have been electrical tape) or the outcome with regards to bitrates? Thanks.
 
I have visions putting electrical tape on their camera and i just cringe. I DO NOT recommend this. Think about the residue it will leave on your chip after you take it off.
 
What about using a sharpie? Just kidding, I am not wanting to permanently damage the IR filter. I was trying to think of a reasonable way to "black out" a portion of the sensor to achieve better bitrates with the ML firmware. I remember someone attempting this but have been unable to find the relevant discussion forum or post. I believe/hope the method didn't involve anything sticky actually touching the sensor. I do appreciate the warning though.
 
Please do not try this! Whoever had the original idea seems to be under the delusion that blacking out 20% of the sensor will make the compression 20% more efficient and lead to image quality that is 20% better. He is wrong. Any increase in quality (if there is any) would probably not be noticable unless pixel peeping, and even then probably not at all after editing and exporting.


If you want cinemascope aspect ratio, just crop in post. Stop pixel peeping, and focus on things that actaully will improve your image quality like lighting.
 
well i was thinking about it after I posted. If in theory you could crop out a portion of the chip with some opaque covering then your bitrate for that area would be very minimal, and thus giving you more overhead for the rest of the image area. I shoot anamorphic by using a lens to stretch my image in post so this method of covering wouldn't work for me. The only way that comes to mind would be a filter that you affix to the rear element of the lens (i know some lenses like the 70-200 f/2.8 have a holder for this http://wapedia.mobi/thumb/ebe1506/e..._EF_17-40mm_f4L_USM_lens_mount.jpg?format=jpg ) Ive never seen a topic on this however.
 
Though the encoder will give A LOT of priority to the sharp edges around the black so it probably negate a lot of the "advantages".
 
Yes, I'm a "pixel peeper". Probably because I consider 1920 x 1080 to be a low resolution image. I conducted some tests to see what the difference would be and found that, given the exact same content, the difference between cropped and non-cropped video was a qscale of 4. Thanks for your words of encouragement deltoidjohn. If you don't mind chasing my children around with a lighting setup while I film them, then I won't mind you giving impertinent advice. I apologize for trying to get the lowest ISO and artifact free footage possible given the limitations of the 550d.
 
Try to write some zeros in HD video buffers. If you do it fast enough, it may work. You will have to compile Magic Lantern and play with the code.
 
Would make a lot more sense to do this in post. Plus you'll have the added benefit of flexibility- you can reposition the frame vertically to your liking. If you somehow mask the actual sensor, you're locked in. Super35 works exactly like this btw.

Noah
 
You must log in or register to reply here.
Back
Top