[RonHendriks1966]

If that is the most convincing argument you can make for the use of a square sensor, photography is in a bad state.

Well I think you forgot one important thing. A camera is there to earn money for the photographer (hobby aside) handling it. I think there is more money made in and with Instagram then many other fields of photography.

http://www.stuff.co.nz/business/money/77536020/how-much-cash-could-you-make-with-your-instagram

https://www.businessinsider.nl/how-much-money-you-can-earn-on-instagram-2017...onal=true&r=US

---

Last edited by RonHendriks1966; 07-23-2017 at 02:02 AM.

[Digitalis]

I agree that output uniformity may be perceptually more important for monochrome than for color for the reasons you mention, I disagree that it's hard to do.

For 35mm sensors this might be the case, but for stitched medium format sensors - it's a nightmare.

lf anything, it's easier to map/shift or interpolate over stuck/dead/hot pixels on the monochrome sensor because immediately adjacent pixels are all of the same color band.

Colour band? Luminance is the issue here, Luminance which is difficult to manipulate electronically since our eyes can pick up very small differences* in adjacent luminance values. Stuck pixels can be mapped out, the problem with this is when there are many of them, bayer sensors have a degree of redundancy and interpolation which is often used in software as a failsafe, with a monochrome sensor there is zero redundancy and no interpolation to hide artifacts behind.

correcting for sensitivity variations is extremely easy. I once designed a high resolution film scanner that used a monochrome CCD camera and we used both dark field and white field calibrations to correct for each pixel's idiosyncrasies in black level and sensitivity.

How much time did you have for that project? How many pixel element rows were on that CCD? Can you imagine doing calibrations for 10,000 sensors?

Compensating for sensitivity variations in an imaging sensor might be easy, when you have all the time in the world to do it. How practical is the method of using darkfield/ white field calibration in a mass production run? Time is money, the more time spent in QC the unit price of the sensor goes up. Also the defect rate** has to be taken into account, if your method is effective in correcting monochrome sensors however it increases time on QC and adds yet constraint upon which sensors pass or fail. High failure rates, and increased QC will force the unit price goes up. This will make such a sensor particularly undesirable for camera makers due to the per unit cost.

Finally, there's the issue that the monochrome version of the sensor will be more uniform than it's CFA cousin because the color sensor's pixels are subject to sensitivity variations caused by pixel-to-pixel manufacturing variations in the CFA filter material. The point is that getting uniform output was not that hard.

I agree - with a properly manufactured Monochrome sensor from a single wafer it is, however with medium format many sensors are still stitched, this introduces many homogeneity issues with sensors that have to be painstakingly calibrated and scrutinized above what Bayer equipped imaging sensors are subject to.

* with training and experience with working in monochrome luminosity and tonal aberrations become particularly easy to spot.
** There are always defective sensors on a wafer.

[photoptimist]

I'll continue to dream too, even for the square sensor. I must admit one of my most enjoyed features on the K-1 is shooting in 1:1 crop mode again. Probably dates me to my favorite camera, the C330 tlr, which I actually preferred over my 6x7.

Please feel free to correct if I'm wrong (this is a learning experience) but I thought sensors by nature are monochromatic - each pixel measuring & recording intensity of (or lack of) light and then a filtration system, either Bayer or Foveon are used to either transmit or block specific wavelengths achieving perception of color.
I also remember reading that sensors by nature are already full spectrum, so adding a hot mirror inside or externally shouldn't make a difference - especially in monochrome.
If this is so, then the hardware is already there and writing the coding would be the majority of the development.

And the technology is already on the market, (Red, Leica, PhaseOne) plus a few others who manufacture mostly for microscope imaging.
Hopefully, a Ricoh rep looks at your thread.
I'll dream for a FF monochrome k-mount and can make due with the converted K-01 for IR.
But if someone is looking at this I'll dream big - a full frame, full spectrum monochrome with 1:1 crop ratio.

Thanks again for starting this thread, there has been a wonderful amount of knowledge shared as well as dreams.

Technically, the silicon sensor pixels are panchromatic in that light of all wavelengths from near-infrared to near UV induce a signal although the sensor is more sensitive middle wavelengths (somewhere in the yellows and oranges). Photons enter the silicon and excite electrons. The electrons themselves have no color information. The readout circuits count the number of electrons and infer a color signal based on where those electrons came from (e.g., all electrons from a "red" pixel are assumed to have been created by red light).

Bayer uses a filtration system in front of the silicon chip. Each pixel has a color filter in front it and the software then assumes that what ever signal comes from that pixel must have been of the color implied by the filter.

Foveon has no color filters but, instead, relies on an unusual physical property of silicon by which blue photons tend to be absorbed near the upper surface of the sensor, green photons tend to get a little further into the silicon before being absorbed, and red photons tend to travel the furthest. Each Foveon pixel is actually a stack of 3 photodiodes with the top one being the most blue sensitive, the middle one tended to get more green photons than blue or red, and the deepest on getting most of the red photons. But the color separation is not very good (a fair number of red photons are also absorbed by the blue and green diodes and some blue and green photons make it deeper into the supposedly red layer of the sensor. Some clever math enables the camera's CPU to statistically estimate the intensities of red, green, and blue signal at each location (that statistical process explains Foveon's poor high-ISO performance).

But the color system of these sensors is not perfect by any means. If you shine pure blue light on a Bayer or Foveon sensor, you'll see a strong signal from the "blue" read-out but also some signal from the green and red ones, too. That's caused by cross-talk in the color filter or Foveon photodiodes. And if you close the shutter, put on the lens cap, and take a very long exposure in hot environment, you'll see a strong color-speckled output from the sensor. That color is not really there at all. It's the camera's software mistakenly assuming that any electrons measured in R, G, and B pixels or photodiodes must be coming from R, G, or B photons.

[Sluggo]

Fleshing out why I voted the unpopular/unrealistic "no, but it should be available" option.... This thread has made me feel mistier than any in recent memory - nostalgic for days when I shot Acros and PanF+ and debated the merits of different developers with my friends. I was never a great B/W photog but certainly had the bug for a while. A dedicated mono DSLR from Pentax, if affordable compared to the niche competitors, could be tempting; but only that. I don't really think I would buy. When I want monochrome, I'm pretty happy with the quality I can get by conversions from the current crop of color sensors, and especially like being able to simulate different filters after the fact. I recognize after reading through the posts here that it could be even better, and would love to see what others could do with such equipment if it became more widely and inexpensively available.

[zapp]

Just scratch of the Bayer filter array yourself Scratching the Color Filter Array Layer Off a DSLR Sensor for Sharper B&W Photos

You can buy this as a service, it is called sensor shaving - but without any warranty incase your camera is damaged

[Digitalis]

Technically, the silicon sensor pixels are panchromatic in that light of all wavelengths from near-infrared to near UV induce a signal although the sensor is more sensitive middle wavelengths

Sensors are panchromatic: correct. Heightened sensitivity in the "middle" yellow/orange range: No.



Of all the physical properties of the imaging sensor, the EPI* layer has a big impact on spectral sensitivity. A sensor with a thicker EPI layer will have extended IR sensitivity - a sensor with a thinner EPI layer will have diminished sensitivity in the IR range.The thickness of the sensor surface passivation layer has an impact upon UV sensitivity. Manufacturers intentionally make their sensors as insensitive to UV and IR already, but as things stand we still need to use Hot mirror filters, which are rather expensive.


*Electron penetration layer.

[photoptimist]

Sensors are panchromatic: correct. Heightened sensitivity in the "middle" yellow/orange range: No.



Of all the physical properties of the imaging sensor, the EPI* layer has a big impact on spectral sensitivity. A sensor with a thicker EPI layer will have extended IR sensitivity - a sensor with a thinner EPI layer will have diminished sensitivity in the IR range.The thickness of the sensor surface passivation layer has an impact upon UV sensitivity. Manufacturers intentionally make their sensors as insensitive to UV and IR already, but as things stand we still need to use Hot mirror filters, which are rather expensive.


*Electron penetration layer.

Good points! Peak sensitivity does depend on design but it also depends on how you measure "sensitivity". On an amps-per-watt basis, your statement is true. On an electrons-per-photon basis, my statement is true.

[pcrichmond]

Fleshing out why I voted the unpopular/unrealistic "no, but it should be available" option.... This thread has made me feel mistier than any in recent memory - nostalgic for days when I shot Acros and PanF+ and debated the merits of different developers with my friends. I was never a great B/W photog but certainly had the bug for a while. A dedicated mono DSLR from Pentax, if affordable compared to the niche competitors, could be tempting; but only that. I don't really think I would buy. When I want monochrome, I'm pretty happy with the quality I can get by conversions from the current crop of color sensors, and especially like being able to simulate different filters after the fact. I recognize after reading through the posts here that it could be even better, and would love to see what others could do with such equipment if it became more widely and inexpensively available.

I'm glad the poll had this option. Though I am a die hard monochrome lover it is nice for others to reflect on how they see.
I had many friends in the film days who wouldn't give up the versatility of color and made fantastic monochrome prints from color negatives, so understand how one would enjoy working in both and have a nostalgia for the old B/W film too.

Even though I would definitely buy a monochrome camera I can't say I've been disappointed in any conversions I've done in post.
If I saw the image in B/W before I shot, then the cross over later fell right into place.
The detail and recorded information the newer digital cameras offer make either process possible.

[Digitalis]

On an electrons-per-photon basis, my statement is true.

Only as long as the sun is involved.

[photoptimist]

Only as long as the sun is involved.

The differences between silicon's sensitivity as measured in amps/watt versus electrons/photon (quantum efficiency) have nothing to do with the sun or the light source. It is a direct consequence of the energy per photon. Blue and UV light packs much more energy per photon than red or infrared light but the quantum nature of semiconductors means that one absorbed photon excites one electron regardless of how energetic the photon was. Put another way, a watt of deep red light has almost twice as many photons per second as a watt of deep blue light which seems to induce a greater output (higher amps/watt sensitivity) even if red photons have a much lower chance of exciting the silicon (low quantum efficiency).

(Perhaps you are thinking of all the calculations of light intensity as measured in lux , lumens, candelas, etc. which are tied to a human vision color spectrum model such that watt-for-watt green light has about 10X the lumens of blue light).

[Bob 256]

You can buy this as a service, it is called sensor shaving - but without any warranty incase your camera is damaged

While the Bayer layer can be removed, it's a tedious process and can lead to damage of the sensor. More importantly, the camera software still thinks the Bayer layer still exists, so it won't treat all the pixels the same (as would be the case in a true monochrome camera). That leads to a number of problems in creating a monochrome image from the data. You'd be better off just using a digital monochrome simulation filter (which can't do true IR), and leaving the sensor alone.

If the sensor is monochrome and matched to a camera that interprets the sensor output as that of an unfiltered sensor (no Bayer or anti-aliasing layers), you gain resolution and the ability to capture UV (near visible unless special lenses are used), visible, and IR. Spectral sensitivity can be tailored by using filters over the lens or in front of the sensor (it can be used to capture a color or false-color image by using multiple exposures with different filters). That was one of the reasons, B & W emulsions were used in astronomy for so many years and you still find astronomers using monochrome sensors for that reason (another reason, as pointed out, is the problem with point images falling on a Bayer array - doesn't work well at all).

[Bophoto]

A good thing about Pentax software is that if you are raw and monochrome set, you have monochrome in the software as well.
Easy.

[elicius]

I just saw that one of the huwai makes mobile phones has a monochrome sensor…

If they can get there hands on b/w sensors why could not a DSLR company…

[photoptimist]

I just saw that one of the huwai makes mobile phones has a monochrome sensor…

If they can get there hands on b/w sensors why could not a DSLR company…

They can but it's not like sensor makers have a warehouse full of both color and monochrome sensors and camera makers can simply buy either off the shelf.

Monochrome sensors are a special order that requires an up-front payments to cover engineering and factory resources (and hassle) to switch between the two types and a commitment to buy enough monochrome sensors.

All it takes is money.

[goatsNdonkey]

I wish there was a monochrome only Pentax dlsr, only I know that I wouldn't be able to afford one if it came out. I wish that I could shift my K10D to directly take black-and-white in-camera jpegs, though I wonder if they would be as good as the conversions I get now. Still it would be fun to shoot monochrome directly, now and then.

Here's a conversion, cross-posted earlier today: