Would a multi-MP-Sensor be technically feasable or is it purest Sci-Fi?

[pedro]

As a non-anglo it's difficult to explain my idea. Hope I kinda match the terms.
Moreover, I've got absolutely no technical knowledge in that field.

There was a discussion somewhere on multi-crop bodies: one could chose 1.3, 1.6. or FF according to job needs.
Taking this a step further from the view point of the sensor I come up with the idea, if there is a possibility to build a multi-MP-sensor. Or let's say: Is it possible to activate individual sensor areas using pixel clusters? e.g. if one needs extreme low light abilities for noise free high ISO pictures he'd choose the 12 MP mode on a 28 MP sensor which would deliever more pixel pitch and more surface for photons (did I get that right?) , or 18 or whatever. So a sensor could be adjusted to certain photographic tasks...Hope my thoughts will get across...Regards from Switzerland. Pedro

[rcormack]

I Beleive Nikons D3S does this, using its DX sized 1.5x crop, and the full frame FX. It also allows DX crop lenses to be mounted to it, reducing the mp, but giving it the crop factor. Rumours have gone around that this may make its way to the 1Dv with FF and asp-h

[richy]

hmm I think you may mean 'pixel binning' where you have say a 20mp sensor that can group blocks of say 4 pixels together to reduce noise and make a 5mp image thats very clean compared to the 20mp image.
Yes it is possible The amount of benefit gained is debatable as the 20mp image would have much smaller pixels so in a larger print size it may end up looking better anyway even with the noise and whilst there would be more deviation between pixels those pixels would be a quarter of the size and therefore have less impact than you may think,

[bvukich]

Phase one does it on it's P40+ & P65+ digital MF backs.

http://www.phaseone.com/Digital-Backs/P65/P65-Technologies.aspx

[K3nt]

Technically yes and as mentioned is being done in several cameras. However, as always with this type of setup it is a compromise. You won't get the best of both worlds but with todays technology maybe 80% performance out of each if you did a combined APS-H and FF sensor for example.
But I can certainly see a market for this type of device so when someone comes up with a device that can do this and preserve image quality across the board, I might even get one. It'd be great to take portraits and other wide shots with a FF sensor and then use a 2x extender with a 400mm lens and 1.6 crop factor to take pics of the moon. 

[John Smith]

I Beleive Nikons D3S does this, using its DX sized 1.5x crop, and the full frame FX. It also allows DX crop lenses to be mounted to it, reducing the mp, but giving it the crop factor. Rumours have gone around that this may make its way to the 1Dv with FF and asp-h

This gives the photographer nothing, except the ability to mount DX lenses on FX bodies.  If one uses an Xmm DX lens on the D3S, the result would be the same as mount an Xmm FX lens, and cropping to keep the center 1/1.5^2 of the pixels.

[OK, there are differences such as in lens IQ, and a different focus mechanism might be used due to the mirror being locked up. The point is, there is no free lunch here.]

[pedro]

Thanks everyone for your posts. @bvukich: sounds great, what phase one provides. but full implementation in future dslr's would be quite expensive as I can see. Thanks again and regards. Pedro

[CR Backup Admin]

Thanks everyone for your posts. @bvukich: sounds great, what phase one provides. but full implementation in future dslr's would be quite expensive as I can see. Thanks again and regards. Pedro

Its always possible that the next new generation of Digic Processors will support pixel binning.  Its one of those things that Canon and Nikon have so far not implemented.   Its just something that has not yet been deemed necessary for the relatively low pixel count cameras.  When we hit 40 or more MP, it might be needed.

[dilbert]

Its always possible that the next new generation of Digic Processors will support pixel binning.  Its one of those things that Canon and Nikon have so far not implemented.   Its just something that has not yet been deemed necessary for the relatively low pixel count cameras.  When we hit 40 or more MP, it might be needed.

But is it necessarily better or just a way to provide lesser MP photos with the benefit of better sensors?

It would be interesting to know how the ratio of the size of the pixel vs the area between pixels has changed over time.

If the gap-less microlenses over each pixel gave you 100% coverage of the sensor and no light was lost, what do you gain by binning multiple small pixels vs having larger ones?

Take the 40MP tech. and build a 10MP sensor of the same size. That'd rock - but it might not sell.

[TexPhoto]

But what is the advantage over cropping after the fact?  I can take a FF image and crop it to the dimentions of a 1.3 1.6, or any other "crop" right now.  I could crop at 1.4, and more that to side.

I kind of understand Nikon's crop with DX lenses, but it's stupid as the DX lens coves more than the 1.5 crop area of the FF sensor.  Thus if the camera just let you take FF photos with a DX lens, you could always just crop them yourself, possibly getting more MP, or even images that were square.  And some DX lenses, especially zooms are known to cover the entire FF sensor, requiring no crop.

[CR Backup Admin]

But is it necessarily better or just a way to provide lesser MP photos with the benefit of better sensors?

It would be interesting to know how the ratio of the size of the pixel vs the area between pixels has changed over time.

If the gap-less microlenses over each pixel gave you 100% coverage of the sensor and no light was lost, what do you gain by binning multiple small pixels vs having larger ones?

Take the 40MP tech. and build a 10MP sensor of the same size. That'd rock - but it might not sell.

Its hard to rate a non-existent feature, but, if you read the Phase One claims, their camera is 4 times more light sensitive when the binning is turned on.  If it can do that, than in very low light, I'd take the 10mp and 4X improvement in light sensitivity.

[EYEONE]

But what is the advantage over cropping after the fact?  I can take a FF image and crop it to the dimentions of a 1.3 1.6, or any other "crop" right now.  I could crop at 1.4, and more that to side.

I kind of understand Nikon's crop with DX lenses, but it's stupid as the DX lens coves more than the 1.5 crop area of the FF sensor.  Thus if the camera just let you take FF photos with a DX lens, you could always just crop them yourself, possibly getting more MP, or even images that were square.  And some DX lenses, especially zooms are known to cover the entire FF sensor, requiring no crop.

Pixel binning is different than cropping. We are not talking about cropping images like the D3 camera does.
Take a 40mp MF sensor. If you apply pixel binning and group adjacent pixels into 2x2 pixel squares that act as one pixel then you have a 10mp MF sensor. This allows for very large pixels and a low density sensor that is better for low light.

However if you use just the 10mp in the center of the sensor then you are getting a crop factor which is what you are talking about.

[pedro]

"Its hard to rate a non-existent feature, but, if you read the Phase One claims, their camera is 4 times more light sensitive when the binning is turned on.  If it can do that, than in very low light, I'd take the 10mp and 4X improvement in light sensitivity."

plus one.

@scalesusa: I would take one too!

Well, let's stretch the dream a bit: How much would that mean in cash? Let's hope that the overnext Generation of Canon's will implement this technique...that is 3 to 6 years...from now;-)

[CR Backup Admin]

Well, let's stretch the dream a bit: How much would that mean in cash? Let's hope that the overnext Generation of Canon's will implement this technique...that is 3 to 6 years...from now;-)

I rather doubt the 4X light sensitivity claims, but you would get better signal to noise ratio, which might give a slight improvement.

It should not cost anything, its just in the processor design, a one time cost.