Monochrome sensor conversion... seeking advise

D Cox wrote:

bclaff wrote:

Meuh wrote:

Any reason why you wouldn't buy a Sigma Merrill and use that in B/W selecting the blue channel only?

Photons of different energy (wavelengths) penetrate the silicon to a different depth.
Using only the blue (top) plane would not capture all the visible wavelength; only predominantly blue. You'd miss the green and red wavelengths that penetrate deeper.

The top layer is miscalled "blue". It is panchromatic with a blue bias. There is no filter in front of it to remove the red and green, but some of the green and red photons do pass through it.

I was able to take three photos through Kodak colour separation filters, convert them all to monochrome as "100% blue", and recombine them in Photoshop to give a normal full colour image.

Interesting, because where you said "some" above I would have thought "most".
So you're saying that something with a strong red component still registers significantly in the top plane?
I have some old test data. I'll have to look at hat.

Regards,

khunpapa wrote:

bclaff wrote:

Meuh wrote:

Any reason why you wouldn't buy a Sigma Merrill and use that in B/W selecting the blue channel only?

Photons of different energy (wavelengths) penetrate the silicon to a different depth.
Using only the blue (top) plane would not capture all the visible wavelength; only predominantly blue. You'd miss the green and red wavelengths that penetrate deeper.

You misunderstand this . Practically, that's Sigma's early statement's fault.

The top layer is only 1 uM thick (thin?). It absorbs all R G B photon. But because it's so thin & the P-N is doped to be higher band gap, the G & R (& some B) photon pass the top layer P-N junction almost without any interaction.

That's what's behind Sigma's new Quatro sensor. The top layer determines the resolution and color. The middle green and low red has 4 times larger P-N area to capture the remained photon, thus not only provides more accurate color information but also increases the QE considerably (1 stop, or 100%).

From one of the very early documents: Foveon Technology and the Changing Landscape of Digital Cameras we see absorption and penetration vary considerably versus depth (note the logarithmic scale).

What saves you is the wide color response of the top ("blue") plane.

However, the sensitivity as you head toward red drops quite a bit.
So yes, you can do it, but I suspect that effective Quantum Efficiency (QE), especially for red, is pretty low compare to a sensor that has had the Color Filter Array (CFA) removed.

Regards,

bclaff wrote:om one of the very early documents: Foveon Technology and the Changing Landscape of Digital Cameras we see absorption and penetration vary considerably versus depth (note the logarithmic scale).

What saves you is the wide color response of the top ("blue") plane.

However, the sensitivity as you head toward red drops quite a bit.
So yes, you can do it, but I suspect that effective Quantum Efficiency (QE), especially for red, is pretty low compare to a sensor that has had the Color Filter Array (CFA) removed.

Regards,

-- hide signature --

Bill ( Your trusted source for independent sensor data at http://www.PhotonsToPhotos.net )

BTW do you know how to model spectral response of sensor with arbitrary number of layers and thicknesses, and for different angle of incidence?

Some introduction.

We are designing and making imaging devices for wide range of applications and industries.

Also as a small portion of what we do we do infrared and full spectrum conversion on cameras.

About your goal.

IF you will succeed in removal of micro-lenses your camera will continue to readout image as combination or RGB pixels. RAW file is NOT a direct readout of per pixel information! So all the task is USELESS if you are  not changing camera firmware to record per pixel information.

Every sensor has QE curve of the sensor and spectral parameters (transmission) of Red, Green and Blue lenses. Camera firmware corrects/adjust sensor signal based on sensors QE curve and transmission of R G B pixels as well as transmission overlaps. This is happening BEOFRE information gets into your RAW file.

Say single red pixel in your out of camera file will NEVER correspond to the single red pixel under red colored on-sensor lens.

About removal of lenses:

Forget about mechanical removal. Those youtube videos are useless and they WILL render your sensor inoperable.

We did try chemical dissolution of polymer lenses. To some degree it was ok. Some sensors had lenses from cross-linked polymer. Forget about dissolving of those.

Another method was evaporation of micro-lens array by target focused femtosecond laser. This worked best. However you would need to take into consideration you would need to seal sensor again. In inert gas atmosphere.

In our case with experimental laser evaporation sensor died on 8th day of life after procedure.

Put is this way.

If it is not a kind of sport for you (where results do not matter) them play with it!

If you are result oriented then forget about it.

It would be cheaper to by Leica Monochrome caemra (remember about camera firmware?)

Try and you will see.

P.S.

There was remark about how QE (quantum efficiency) of sensor changes with removing IR Cut filters. QE of a sensor is an inherit property or particular model of photo-detector. Basically said if your single photo-detector is bean stroked by 10 photons and photo-detector converted to 10 corresponding pulses (electrons) then QE=1 Removal of filters from in-front of the sensor will not change properties of the sensor (unless you screwed it up :).

Sensor will be ruined as per my opinion.

Which solvent works best to remove the CFA from a Sony sensor, like that in the Nex-6 or Nex-5n?

Nobody asked if there's "wise" to do so. There are people who are willing to risk destroying a camera, who are we to set goals for them. 
Of course, it's not much better to plug in native BW sensor, if someone's willing to provide them  with one...
There are people (e.g. Iliah Borg) who removed CFAs with solvents with acceptable results.

I'm thinking about doing this procedure to my Nikon d60, it's not an incredibly good camera or anything but parts are cheap enough to do it and not care that much. Also, it's a CCD and not a CMOS so, no microlens I'm trying this week or the coming weekend, I'll post some photos of the result and hopefully some test shots...

Chrisbavon wrote:

I'm thinking about doing this procedure to my Nikon d60, ...  it's a CCD and not a CMOS so, no microlens ...

FWIW, just because it's CCD doesn't mean that it doesn't have microlenses.

bclaff wrote:

Chrisbavon wrote:

I'm thinking about doing this procedure to my Nikon d60, ... it's a CCD and not a CMOS so, no microlens ...

FWIW, just because it's CCD doesn't mean that it doesn't have microlenses.

I know, and also know this one doesn't doesn't have microlenses...

Chrisbavon wrote:

bclaff wrote:

Chrisbavon wrote:

I'm thinking about doing this procedure to my Nikon d60, ... it's a CCD and not a CMOS so, no microlens ...

FWIW, just because it's CCD doesn't mean that it doesn't have microlenses.

I know, and also know this one doesn't doesn't have microlenses...

Can you cite a source for how you know this?