Some musings on aspect ratio

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

DRwyoming said:
Just a guess, but based on the semiconductor fab work I've been involved with it rectangular or square sensors may have a lot to do with yield out of a wafer. IOW, circular wafers are sliced into multiple devices(multiple dies) and you don't waste as much of the total wafer if you cut rectangles vs circles. That and dicing a wafer with straight lines is very standard and easy, cutting circles out of each wafer isn't something that I've seen before but I suppose it could be done. As it is, full frame 24x26mm sensors already lead to lower yield and less ability to pack a wafer than smaller sensors like APS-C or smaller.
Click to expand...

I thought he meant curved as in 3d. How would a round sensor map to a rectangular image?
 
Member's gallery
EricBowles said:
I'm sure you're correct - and of course the power and heat impacts complicate that problem. Still - I could appreciate a 25% larger image option and would pay for it. With a shutterless camera, alternate sensor configurations are more practical than in the past.
Click to expand...

Would you settle for an octagonal sensor? (Easier to close-pack on a wafer.)
 
Member's gallery
bleirer said:
I thought he meant curved as in 3d. How would a round sensor map to a rectangular image?
Click to expand...
As I read Eric's post the idea would be to project the round image circle from the lens onto a round sensor but then crop in camera(or even in post) to various aspect ratios. IOW, with no change in camera position you could crop to a horizontal 3:2 AR or a vertical 3:2 AR if the sensor was built to capture the entire projected image circle. Basically any rectangle that could be captured horizontally (landscape format) could also be captured vertically (portrait format) without changing the camera position or requiring vertical grips or L brackets and the like and with no change in captured image resolution.
 
Last edited:
DRwyoming said:
As I read Eric's post the idea would be to project the round image circle from the lens onto a round sensor but then crop in camera(or even in post) to various aspect ratios. IOW, with no change in camera position you could crop to a horizontal 3:2 AR or a vertical 3:2 AR if the sensor was built to capture the entire projected image circle. Basically any rectangle that could be captured horizontally (landscape format) could also be captured vertically (portrait format) without changing the camera position or requiring vertical grips or L brackets and the like and with no change in captured image resolution.
Click to expand...

Interesting idea. I guess there would be more ways to cut the pie. Even a larger square sensor would work in that case, with some waste where the image circle didn't reach. I initially thought he was talking about a 3d curved area to fit Gaussian optics better, though software and lens design fixes that pretty well.
 
Member's gallery
bleirer said:
Even a larger square sensor would work in that case, with some waste where the image circle didn't reach
Click to expand...
Yeah, that was a thought I had a while back in a different thread on where digital cameras might evolve. Basically a medium format sensor that you could crop full frame 35mm format images out of in various aspect ratios and supporting in-camera landscape to portrait switching at the touch of a button without changing the actual camera position. The circular sensor (if practical from a manufacturing standpoint) as Eric suggested could do that but so could a sort of dual MF/FF camera. Start with something like a 100 mega pixel MF sensor similar to what Fuji has and you could crop to various FF aspect ratios and orientations still maintaining a fairly high resolution in the process. Seems Nikon could do something like that with their relatively large Z mount.
 
Member's gallery
DRwyoming said:
As I read Eric's post the idea would be to project the round image circle from the lens onto a round sensor but then crop in camera(or even in post) to various aspect ratios. IOW, with no change in camera position you could crop to a horizontal 3:2 AR or a vertical 3:2 AR if the sensor was built to capture the entire projected image circle. Basically any rectangle that could be captured horizontally (landscape format) could also be captured vertically (portrait format) without changing the camera position or requiring vertical grips or L brackets and the like and with no change in captured image resolution.
Click to expand...

Yes - or a larger rectangular sensor that could take advantage of the image circle. For example, the current sensor is 8256 x 5504. But an optimal 5:4 image would not be 6880 x 5504 (37.9 megapixels) as it is today, but rather the largest 5:4 size that could fit in the image circle which is closer to 7750 x 6200 (48
megapixels). Likewise a 1:1 aspect ratio today is 5504 x 5504 (28 megapixels), but the largest possible 1:1 image within the image circle produced by the lens is closer to 7016 x 7016 (49 megapixels).
 
MikePapple said:
My 32" 4K monitor is 3840x2160 (16:9). I tend to crop and re-size to this. When I make presentations, they are now all 16:9.
Click to expand...

I guess it depends a lot on the image itself as well as the use you have for the image. Some shots seem to beg for a vertical presentation, while others seem better suited to horizontal. Sometimes it is how the patterns, framing elements, and diagonals in the image relate to the diagonals and edge lines of a certain aspect ratio. Other times you know it's going to be displayed on a certain screen or in a certain frame or maybe a publication wants a certain size. It's all good in my view as long as the attributes of the image are in the driver's seat.
 
You must log in or register to reply here.
Back
Top