Canon DSLR Rumors

The CMOS Sensor Squared [CR2]

Get Squared
This is a leak from an unnamed camera manufacturer about square sensors.

We have heard in the past that the 1Ds Mark IV would have a square CMOS sensor. Most people dismissed it as poppycock.

This rumor may appear to be in a different format that previous rumors, however this is how it was sent to me and I was told to post it as such.

by Dean Francis

Camera image sensors commonly use rectangular formats (3:2, 5:4, 16:9 aspect ratios). However, to obtain the highest image quality, the square format (1:1 aspect ratio) should be used. The advantages are surprisingly extensive as outlined below:

MAXIMIZED FIELD OF VIEW
The square format covers 100% of the maximum field of view. The maximum FOV is the largest area that can be covered by any four sided sensor (see diagram). A rectangular format (3:2) uses only 89% of the maximum FOV. Simply stated, the wider the rectangle, the smaller the photograph. The square format is about 11% larger – a considerable amount of image data.

PERIPHERAL LOSS MINIMIZED
Generally speaking, image quality suffers along the outer third edge of lenses. That’s where any circle of confusion (COC) issues become readily apparent, such as blurring, chromatic aberration, distortion, vignetting, etc. A rectangular format actually pushes outward into that area of the lens. 18% of the image (4 corners & 2 sides) are clearly degraded. However, the square format eliminates degradation (all 4 sides) and minimizes the remainder by pushing it farther into the corners.

MICRON GAIN IN RESOLUTION
Photo sites benefit dramatically when their size is increased. The square format has a surface area 12% larger than a rectangular format (3:2). Photo sites can be manufactured 12% larger, a gain of +1.12 per micron. The 12% increase also expands sensel (super pixel) variations of resolution and enables 16 bit RAW capture when coupled with improved binning algorithms.

CROP LOSS SOLVED
Cropping a landscape oriented photograph (3:2) down to portrait discards about 60% of the image data (see diagram). With a square format, only 20% is lost. Landscape vs. portrait orientation can be determined later based upon output. Once cropped, only 3% of the peripheral loss region remains (extreme corners).

ADVANCEMENTS
Photographers no longer have to rotate the camera and the secondary portrait grip becomes a thing of the past. Also, the quantity of materials is lessened (about 10%) by shortening the toe, foot and heal of the bottom plate. Weight is thereby decreased, allowing for significant changes and reallocation of the source/component materials. Additional battery capacity can be maintained and/or relocated to a secondary or supplemental location. The sensor is also cooled more evenly (about 5%) by equidistant dissipation of heat through the mount.

SUMMARY
The list of improvements for the square format is extensive, benefitting camera functionality in all photographic and video applications. It’s also highly cost effective from a materials/manufacturing standpoint, desirable from the consumer’s viewpoint and fully marketable as a standardized format.   Dean E. Francis

cr

187 responses to “The CMOS Sensor Squared [CR2]”

  1. Isn’t plenty of glass out-resolved by the sensor anyway?

    And purple fringing and such issues may become less with lower pixel density?

    And eventually all surface area isn’t useful but taken up with some other electronics, don’t know, but if it is and said electronic take the size it does no matter what the size of the pixels are then more pixels = more electronic = less total area available for pixels = worse in any case.

    Though noise levels seem so low nowadays anyway.

  2. why no circle sensors

    you would use the whole fov in every direction
    you can change image orientation without cropping more and more imageinformation (well just once to get the rectangular picture out of the circle for prints/web)
    the sensor could be build with a new kind of pattern maybe a hexagonal one
    this would also solve some moiree problems maybe

  3. Larger mirror: think about how it’ll swing up. Flange distance on EF mount would no longer be sufficient. Would they go to the trouble of bringing in a new mount?

  4. Maybe you can read the pixels in a circular fashion from the edges and inward and write them down as you go on a circular piece of plastic.

    But that would be pretty sweet, if they could make it fast enough, put a small censor on a rotating pin which spins covering the whole light area and samples the light in that area as it goes/passes =P. It just need to spin fast enough so nothing is able to move quicker before it have made a full lap ;D

    Would save sensor area! ;D

  5. The viewfinder would obviously be electronical and show the current crop if any at 100% coverage.

Leave a Reply

Your email address will not be published. Required fields are marked *