xxMP? No, multiply it by 3 like Sigma

[EchoingWhisper]

I thought that Sigma's Foveon sensor is a good idea, but the design didn't really took off. After seeing the 200MP Hassy, I thought of an idea. You could actually make a Foveon Bayer hybrid. If you move the sensor upwards one pixel and sidewards one pixel, you could actually have 3 different colors at one area. But then this could only be used on tripod. Then do this again by moving the sensor 1 1/2 pixel you get 12 times the resolution!

 

[Garbz]

Nope if you move it a full pixel the end result would be the same data recorded just slightly off to one side.

If you move it by HALF a pixel your idea has merit.

/EDIT: By the way has anyone ever told you that you think too much

 

[Nikon_Josh]

I thought that Sigma's Foveon sensor is a good idea, but the design didn't really took off. After seeing the 200MP Hassy, I thought of an idea. You could actually make a Foveon Bayer hybrid. If you move the sensor upwards one pixel and sidewards one pixel, you could actually have 3 different colors at one area. But then this could only be used on tripod. Then do this again by moving the sensor 1 1/2 pixel you get 12 times the resolution!

How do you cope with the D5100 at only 16MP is my question? It seems you are chasing pixels, so you need to sell your Nikon gear and save up for that 200MP Hassy!

 

[KmH]

The Foveon sensor takes advantage of the fact red, green, and blue light penetrate silicon to different depths.

File:Absorption-X3.png - Wikipedia, the free encyclopedia

No blue light reaches any of the green or red pixels. No green light reaches any of the red pixels. All the pixels 'see' red light.

So if you shoot a pure all blue scene, only 1/3 of the advertised Foveon pixels will record any luminosity data. The green and red pixels won't 'see' any light at all.

 

[EchoingWhisper]

The Foveon sensor takes advantage of the fact red, green, and blue light penetrate silicon to different depths.

File:Absorption-X3.png - Wikipedia, the free encyclopedia

No blue light reaches any of the green or red pixels. No green light reaches any of the red pixels. All the pixels 'see' red light.

So if you shoot a pure all blue scene, only 1/3 of the advertised Foveon pixels will record any luminosity data. The green and red pixels won't 'see' any light at all.

I think you misunderstood my point. If you moved the green pixel up, then the red pixel would take its place. Then if you move the sensor back then sideways, the blue pixel would be at that place. Combine those three pixel and you'll get 3 times less noise.

 

[EchoingWhisper]

How do you cope with the D5100 at only 16MP is my question? It seems you are chasing pixels, so you need to sell your Nikon gear and save up for that 200MP Hassy!

I'm just dreaming, lemmeee dream!

 

[EchoingWhisper]

Nope if you move it a full pixel the end result would be the same data recorded just slightly off to one side.

If you move it by HALF a pixel your idea has merit.

/EDIT: By the way has anyone ever told you that you think too much

Yes. I think too much!

 

[Garbz]

I think you misunderstood my point. If you moved the green pixel up, then the red pixel would take its place. Then if you move the sensor back then sideways, the blue pixel would be at that place. Combine those three pixel and you'll get 3 times less noise.

I think you misunderstood Keith's point. There's no concept of red green or blue pixels on a foveon sensor. If you move the sensor backwards blue still stays blue because the key here is the photon penetration depth into the silicon and not the focal point on the sensor. Move the sensor back 1 micron, and the 450nm photon just travels 1 micron further before it gets absorbed.

If you're after statistical reduction of noise there's no need to move the sensor or to interpolate to do it. Just take 3 shots and take an average reading.

 

[EchoingWhisper]

I think you misunderstood Keith's point. There's no concept of red green or blue pixels on a foveon sensor. If you move the sensor backwards blue still stays blue because the key here is the photon penetration depth into the silicon and not the focal point on the sensor. Move the sensor back 1 micron, and the 450nm photon just travels 1 micron further before it gets absorbed.

If you're after statistical reduction of noise there's no need to move the sensor or to interpolate to do it. Just take 3 shots and take an average reading.

We might get better clarity because we don't need to interpolate. My idea is to simulate the Foveon sensor by moving the entire Bayer sensor to simulate it. I'll try make an illustration.

 

[o hey tyler]

It's intriguing to me that you have a camera that you seem to never take photos with. Instead, you post about sensors of other cameras and theoretical "what if" scenarios rather than posting photos to receive critique or giving critique on posted photos.

You might want to check these out if your interest is truly in the mechanics of sensors and not so much actual photography:

Physics Help and Math Help - Physics Forums

DxOMark - Index

Just offering up some suggestions to connect with more people that have similar interests.

 

[Garbz]

Oh you want to do this to a Beyer sensor, got it. But more importantly just buy a D800. It can record 10mpx with no interpolation

 

[EchoingWhisper]

Oh you want to do this to a Beyer sensor, got it. But more importantly just buy a D800. It can record 10mpx with no interpolation

With 3 times less efficiency.

 

[Garbz]

That's entirely debatable. What you're suggesting is taking 3 photos in quick successions and doing all manner of wonderful post processing. Efficiency in this case is horrendously low not only because of the extra equipment needed to move the sensor (not trivial to do accurately at the precision we're talking about), but also the time taken to take 3 photos, read them out, and finally run them through a new form of post processing.