Black and White ONLY Digital Camera

[Boatman]

There have been plenty of discussions regarding converting our color digital images into black and white images.

Are we losing anything when we do this conversion? If we had access to 12-20MP cameras that only captured black and white luminance, would the resulting images be better than the converted color images? (I'm guessing that they would be.)

I understand that we would have to go back to using colored filters and such, since we would not be able to make the channel adjustments we can do now. It would be a whole different paradigm. But would it be better? Has a digital black and white only camera ever been made?

[Donald]

Homer - I think we've had a discussion about this before. One has not been made on any commercial bases, for the simple reason that there's not a commercial market for it.

I don't know the science behind it all, but so far as I am aware, we are not losing anything by deciding to not include some of the data (the colour information) in our final image.

[herbert]

If you convert an RGB image to BW you can control how much red, green and blue are used in the final image and you can do this locally. To replicate this during capture on a B&W camera you would have to use colour filters on your camera lens, you only get one chance at that and the filters are global. Thus you have more control if you shoot RGB and post process digitally.

In the old days there was no option. You can throw away options if you want but you will produce better images from RGB.

On the other hand removing the Bayer filter over the sensor may increase image sharpness and marginally low-light performance since there is less between the sensor and the outside world. However I have not seen a side-by-side comparison. I think the difference would not justify the loss of flexibilty that RGB provides.

Alex

[William W]

Slightly skew but only just off the topic -

RE - Shooting B&W:

If the intention is for the end product to be B&W then, mainly because many people I work with are not familiar with shooting B&W film, I suggest to Students that a few inexpensive Contrast Filters are a good idea to use ON THE SHOOT, through which to view the scene.

Just up to the eye is worthwhile, but through the viewfinder is actually better.
- a RED, a YELLOW, a GREEN and a YELLOW GREEN – to “see” the tonal range and tonal differences that the final B&W shall (or could) have.

This preview is useful for:
changing the Lighting;
changing the Perspective;
changing the the Composition
- to best suit the final (envisaged) Black and White Image

All these elements cannot be changed in the digital darkroom, later.

***

RE - using Colour Correction Filters with Digital:

There are some (rare) circumstances when Colour Correction Filters are useful to use, even when shooting raw with digital and with the intention to manipulate the image later in Post Production.

These circumstances are where the illuminating light source is heavy in one of the colours and or the spectrum of the illuminating light source far from continuous throughout the visible spectrum. Mercury Vapour Lamps are one example.

In this example lighting scenario, the use of Colour Correction Filters can protect over excitement of ONE set of colour receptor cells, in respect of the other two, whilst maintaining a correct exposure within the Dynamic Range of ALL the sensors.

Whether or not the effort is worth it is dependent upon the critical nature of the work the, but once a set of receptors is over excited, that’s an irretrievable situation even if shooting raw.

WW

[Dave Humphries]

There have been plenty of discussions regarding converting our color digital images into black and white images.

Are we losing anything when we do this conversion? If we had access to 12-20MP cameras that only captured black and white luminance, would the resulting images be better than the converted color images? (I'm guessing that they would be.)

I understand that we would have to go back to using colored filters and such, since we would not be able to make the channel adjustments we can do now. It would be a whole different paradigm. But would it be better? Has a digital black and white only camera ever been made?

I can see what Homer is getting at here - if we could have about four times* the resolution, wouldn't it be worth it?

Donald has now placed his order btw

* if you're wondering why 4 times and not 3, it is because most sensors use the bayer matrix of one colour pixel being derived from 1 x Red, 2 x Green and 1 x Blue 'sub' pixels. Read more ...

I am sure there are monochrome sensors, made for medical and scientific imaging, but I suspect the prices are rather high due to low volume of sales.

However, other things to consider;

Are our lenses good enough for that level of sensor?
Could we afford the lens performance necessary?
What would be the point in having that much resolution? - for web use, everyone has to down-size anyway, discarding the gains, plus 'average printing' effectively does the same

Cheers,

[krispix]

I can see what Homer is getting at here - if we could have about four times* the resolution, wouldn't it be worth it?

Donald has now placed his order btw

* if you're wondering why 4 times and not 3, it is because most sensors use the bayer matrix of one colour pixel being derived from 1 x Red, 2 x Green and 1 x Blue 'sub' pixels. Read more ...

I am sure there are monochrome sensors, made for medical and scientific imaging, but I suspect the prices are rather high due to low volume of sales.

However, other things to consider;

Are our lenses good enough for that level of sensor?
Could we afford the lens performance necessary?
What would be the point in having that much resolution? - for web use, everyone has to down-size anyway, discarding the gains, plus 'average printing' effectively does the same

Cheers,

Hi Dave,

I think you're missing a trick here.

Unless you're shooting in half-tone (literally B&W) you will still need a panchromatic sensor which is sensitive to all the visible spectrum. If you have a sensor which is only sensitive to blue light (say) all the red and green light will be lost and end up black, similarly any of the other colours. Just because your final result is monochrome you still need to be able to record all the colours in order to get the proper monochromatic rendering.

I hope this makes sense - I didn't want go on for page after page on the theory of image recording.

[FrankMi]

Hi Homer, lets say that you could shoot B&W digital images (as there was no mention of simply shooting B&W film), wouldn't you also need a B&W monitor as well to achieve your results???

[Donald]

For some reason my response has shown up twice and refuses all attempts to remove the second entry. Any Mods out there with the power to edit?

I am that man!

Job done.

[stg]

Hi Dave,

Unless you're shooting in half-tone (literally B&W) you will still need a panchromatic sensor which is sensitive to all the visible spectrum. If you have a sensor which is only sensitive to blue light (say) all the red and green light will be lost and end up black, similarly any of the other colours. Just because your final result is monochrome you still need to be able to record all the colours in order to get the proper monochromatic rendering.

Chris, very informative. I have a question though. With a Bayer filter on, each pixel catches photons of a particular frequency/wavelength only. With the filter removed, each pixel would gather all photons sent there by the the lens. In that case we'd be in need for pixels with much higher capacities or am I getting it wrong?

[RazorTM]

Homer - I think we've had a discussion about this before. One has not been made on any commercial bases, for the simple reason that there's not a commercial market for it.

I don't know the science behind it all, but so far as I am aware, we are not losing anything by deciding to not include some of the data (the colour information) in our final image.

http://www.luminous-landscape.com/re...dak-760m.shtml

[Boatman]

The piece on Luminous-Landscape is very interesting. Thank you for sharing that Razor. The article backs up my thinking that a monochrome digital camera would be really special while it also reinforces the fact that the industry is not moving in this direction and is not likely to do so.

The M760 Petere Myers writes about was shipped in 2002 so the technology in it would be at least ten years old now. Imagine what could be built today with current technology and parts!

[Dave Humphries]

Hi Dave,

I think you're missing a trick here.

Unless you're shooting in half-tone (literally B&W) you will still need a panchromatic sensor which is sensitive to all the visible spectrum. If you have a sensor which is only sensitive to blue light (say) all the red and green light will be lost and end up black, similarly any of the other colours. Just because your final result is monochrome you still need to be able to record all the colours in order to get the proper monochromatic rendering.

I hope this makes sense - I didn't want go on for page after page on the theory of image recording.

I don't think I am - but perhaps I wasn't clear enough or used the wrong terms.

I was assuming that the pixels under the bayer microfilters are panchromatic - basically luminance light sensors, so if you have a sensor without the bayer filter, you get 4 times the resolution - of course, it ain't that simple, as all the image processing on chip expects there to be the filters in place and will co-site the images losing the rsolution in the process. But you get the idea?

[Glenn NK]

I won't get into the technical feasibility of a B/W only sensor because it's well beyond my technical knowledge. But for the sake of argument, let's suppose that it was feasible and would produce superior results for B/W.

A quick count on another forum which has a high ratio of professionals yielded the following:

In the last 64 colour images submitted, there were eight B/W images for a ratio of one B/W for every eight colour images. While this could possibly make a B/W sensor camera a viable investment for a manufacturer, the ratio of photographers that do B/W imaging exclusively may well be less than 1 in 100 or so. For all the others that do both, this means an extra body - another investment.

How many users would invest the money in a limited production (more costly) B/W only sensor camera?

As intriguing as it is, I'd be surprised if a manufacturer pursued this concept. In fact, I suspect that they've all already looked into it and have passed.

Glenn

[krispix]

I don't think I am - but perhaps I wasn't clear enough or used the wrong terms.

I was assuming that the pixels under the bayer microfilters are panchromatic - basically luminance light sensors, so if you have a sensor without the bayer filter, you get 4 times the resolution - of course, it ain't that simple, as all the image processing on chip expects there to be the filters in place and will co-site the images losing the rsolution in the process. But you get the idea?

Hmmmm!

I suppose the removal of the Bayer filter will, in itself improve resolution, by taking another light-scattering object out of the equation. But I still have doubts that it's quite that simple. The problem is that sensors are less sensitive to green light than they are to red and blue which is why the Bayer pattern mosiac has 3 green for every 2 blue and 2 red filters. If you remove the filter the sensor would be more sensitive to red and blue light and greens would be degraded. In a B&W rendering the grass would be very dark while the sky and the No.9 London bus would stand out well. So, although you're seeking to render in monochrome you still need to record the panchromatic spectrum in terms of light intensity.
In the old days of film. B&W negative stock was still filtered to mimic the light reception levels of the human eye and cut out the non-visible spectrum. So, although it was monochromatic in output it was still panchromatic in sensitivity, and I can't see a way round it. You need the filtration and you need the panchromatic sensitivity.
Of course, if you're looking for improved resolution you could spend loads of money and get a camera with a Fovean sensor as these have no filter and the separation is done on each individual RGB layer. I'm told these do deliver, but you would still need that panchromatic separation.

I'm going to lie down in a darkened room now, because my brain is beginning to hurt.

PS. Thanks Donald for sorting the double post.

[Steaphany]

Monochrome digital photography is not only available, it is how some work every day.

If you look to the high end astronomical photography cameras, they use imaging chips with no on-chip Bayer mask. Each photosite will capture the full spectrum that the chip is capable of, infrared, visible, and near ultraviolet:

Click image to enlarge

Unlike a camera intended for general photography, astronomical cameras must operate with very well known performance characteristics. The reason for not putting a Bayer mask on the chip itself is to not limit imager performance. There is no Hot Mirror to block Infrared. There is no Bayer Antialiasing "soft focus" filter. Noise levels are controlled by thermo-regulated chilling of the imager by multistage Peltier Thermoelectric modules with the hot side either air or fluid cooled. The imager is maintained at a constant temperature to within 0.1 °C over exposures ranging from fractions of a second to hours. The imager is even characterized to the number of electrons each photosite can contain and based on the filtering and known quantum efficiency, you can calculate how many photons would saturate a photosite. Digital color astronomical photography is conducted by using a filter wheel with a set of spectrum selective filters. Even at this level, the images are not in "color", but in post processing the various filtered exposures are assigned to specific color primary channels to effectively colorize the image. Since every photosite is used to capture multiple exposures to "create" a color image, there is no Bayer interpolation and the resolution and clarity is maximized. Astronomical post processing even has developed manipulations which further reduce noise and even increase resolution by combining multiple sets of exposures.

Click image to enlarge
M81 and M82 by Pavel Cagas
Camera Moravian Instruments G4-16000

Moravian Instruments from the Czech Republic manufacturers the G4-16000 camera used to shoot the above image. The camera has a unique mount which facilitates not only attachment to a telescope, you can also mount SLR lenses:

Click image to enlarge
Camera Moravian Instruments G4-16000 with Canon EOS bayonet lens

For terrestrial photography, a camera such as the Moravian Instruments G4-16000 could be used, easily allowing direct monochrome photography or enabling a level of control impossible with a conventional SLR. Just be aware that imagers in the class of the Kodak KAF-16803 CCD have a extremely high light sensitivity. ( ISO equivalents are rarely quoted for astronomical cameras )

If you want to learn more about Moravian Instruments, here is their web site:

http://www.gxccd.com/cat?id=130&lang=409

[stg]

Each photosite will capture the full spectrum that the chip is capable of, infrared, visible, and near ultraviolet... The imager is even characterized to the number of electrons each photosite can contain and based on the filtering and known quantum efficiency, you can calculate how many photons would saturate a photosite.

Steaphany, the sensors of these cameras have photosites with higher capacities than the sensors of conventional dSLRs?

[Steaphany]

Steaphany, the sensors of these cameras have photosites with higher capacities than the sensors of conventional dSLRs?

Steve,

Honestly, I do not know for sure, especially since general photography camera manufacturers never provide the details which are common to specifications of astronomical cameras. Nikon and Canon never list how many electrons a photosite can contain. Another factor that can impact this is the number of bits in the ADC used to convert photosite measurements into digital values. I've typically seen SLR's with 12 bits of ADC resolution which places a limit to distinguishing subtle differences in photosite to photosite measurements. By contrast, the Moravian Instruments G4-16000 employs a 16 bit ADC and when considering three exposures of Red, Green, and Blue, each RAW pixel would be represented by 48 bits. ( HDR imaging native to the camera ) Based on the specifications, each ADC step corresponds to 1.6 electrons when the photosites are set to 1x1 binning ( You can actually group photosites, from 1 to 4 in the X or Y axis, if you need greater dynamic range ). At a quantum efficiency of 40%, for 700 nm Red light, that implies each ADC increment corresponds to 4 photons. ( Go ask Nikon for that performance specification )

The Handbook of Astronomical Image Processing has a whole chapter dedicated to characterizing a camera so that you can derive the performance specifications.

Something that also needs to be noted with astronomical cameras, they are not self contained. Unlike even the most advanced general photography SLR, an astronomical camera must be used while connected to a computer to control and store the exposures and you also need an external power source. These aren't the fastest at shooting multiple images either, as the Moravian Instruments G4-16000 requires 22 seconds to offload and store a single 16MP image. The fastest shutter speed is 1/5th Second, so freezing motion, well what you'd normally get with terrestrial photography, would be a challenge.

I personally feel a Moravian Instruments G4-16000 would be an interesting camera to work with shooting general purpose photography for anyone who really seeks control. ( I want one )