Pro Class compact and Unique Lenses Breakthrough 8MP 16 BIT video-sensor, canon R mount camera

[Asher Kelman]

This new camera

​

Canon Announces EOS C70 4K Digital Cinema Camera and Mount Adapter EF-EOS R 0.71x

What might this mean to you?

Well me for it perhaps means giving up MP count in bird photography for amazing color and noise-free shadows.

I would consider taking a video across a landscape and then stitching that as a highly detailed richly colored panorama!

Moreover, with the R to Eos adapter one gets a 0.7 factor of increased, speed booster effect, if one so wishes!

My guess is that the 2 extra BITS of data overcomes the lesser MP prowess of this sensor.

What do you think?

Also it suggests that when Canon delivers it “One” Pro Class flagship version on the R Mirrorless camera, it could also be 16 BIT!

That should be spectacular!

Asher

 

[Jerome Marot]

There is very little to be gained with a 16 bits sensor over a 14 bits sensor in practice. Due to several noise sources in the signal (lens flare, electronics but also the quantum nature of light itself), the signal/noise ratio is only so much.

There are losses when using a video codec for photography. If you need short bursts of pictures taken at blazing speed to extract a single photographs, there are devices which will do that on a lossless format.

I think that this device is rather a proof that the cameras have gotten as good as it gets. We are approaching limits deriving from the fundamental nature of light itself.

 

[Asher Kelman]

There is very little to be gained with a 16 bits sensor over a 14 bits sensor in practice. Due to several noise sources in the signal (lens flare, electronics but also the quantum nature of light itself), the signal/noise ratio is only so much.

There are losses when using a video codec for photography. If you need short bursts of pictures taken at blazing speed to extract a single photographs, there are devices which will do that on a lossless format.

I think that this device is rather a proof that the cameras have gotten as good as it gets. We are approaching limits deriving from the fundamental nature of light itself.

What about their special dual path for reading each pixel.

Asher

 

[Jerome Marot]

What about their special dual path for reading each pixel.

I don't know, but there is only so much which can be done if the problem is with fundamental physical limitations in the signal itself. And is that any different that correlated double sampling, which has been around for some time now?

 

[Asher Kelman]

I don't know, but there is only so much which can be done if the problem is with fundamental physical limitations in the signal itself. And is that any different that correlated double sampling, which has been around for some time now?

I think this could be the fruit of the now ~10 year Stanford University project and separate patents by Canon in which the pixel array consists of separate CMOS cameras, each independently controlled with an electronic shutter on the rate of filling with photons. This could allow early shut off of most brightly illuminated pixels and the dimmest areas illuminated can keep counting voltage increase until the number of photons counted is accurate for that “camera”. But here they have two pipelines. The first gets enough data to determine color an then it switches off and then the second continues until a sufficient number of photons are collected to get accurate numbers above noise.

This could be a version of the layered photo cells that the Stanford group started a commercial company. That CMOS takes measurements at something like 100,000 frames a second if I remember right.

They don’t need to say “Foveon-like CMOS”, if that’s what they have, as Canon has its own patents for layered photocells that each give color data. Or they could of course be using a Bayer array as usual. But the special difference here is the two separate logical pathways for color and brightness.

Asher