Canon announces 120MP SLR, 8K (35Mp) 60fps cine cam.

[Bart_van_der_Wolf]

That optimal pixel size changes with technology was an integral premise of the Stanford paper that I referenced so there was no need to reiterate that fact. Changes in lens design and sensor technology would be necessary for a 120 MP 135 sensor to be practical.

Hi Bill,

I'm not sure how much influence the (untested) addition of micro-lenses has, on the model's optimal outcome. Since the model mentions 35% fill factor, there should be a significant enough positive effect on the number of converted Photons, thus boosting the DR levels (if well depth / capacitance is maximized per photosite). It also looks like well depth, despite smaller photosite pitch, has increased in recent years from something like 1500 e- per square micron, to some 2500 to 3280 e- per square micron (if the data at sensorgen is accurate).

Cheers,
Bart

[BJL]

The fact that this camera is using an APS-H sensor (which Canon stopped putting into new cameras after 2009) tells us that it's almost certainly a technology development announcement, not a real product announcement.

There is no reason to build this sensor as APS-H for production release.

I agree that Canon will never again put a 30x20mm sensor (which Canon bizzarely calls APS-H) into a "consumer" camera, but I can see a couple of reasons to make such sensors and more special-purpose cameras that use them.  Both are based on the fact that this is about the largest size that can be made without the on-wafer stitching needed to make sensors 36x24mm and bigger. (All steppers with feature size small enough to make pixels this small are limited to a 33x26mm field size, and a mm or two is lost along the edges to non-image recording stuff.)  Avoiding stitching reduces costs, and probably makes it easier to get the pixel size down, by avoiding the very precise alignment needed when the wafer is moved part-way through the on-wafer stitching process.

Firstly, it can make sense for prototypes and demonstrations to potential (industrial/scientific) customers: to develop, test, and demonstrate a new higher pixel count technology. The subsequent announcement of a 250MP 36x24mm sensor goes with that.

Secondly, there are other potential markets for Canon sensors besides EOS DSLRs!  Canon has previously provided some sensors for astronomy, and there are fields like machine vision, mapping and surveillance where the users might care very little about using exactly the Leica-ordained 36x24mm frame size: total detail resolved, as roughly indicated by pixel count, can be more important in some technical applications.


But if some people wish to believe that no sensor should be developed and no product should be announced that is not of immediate interest to "our sort of photography", please carry on.  I instead appreciate any information about longer term research and development plans and directions that the companies are willing to share with me.

[Phil Indeblanc]

I too think its more marketing than a actual product that will be shipping in the next 18 months.

I just wish they would get together with Sigma and push the Foveon sensor tech to the forefront. It looks to like a great way to boost major quality and pixel count.