What about a BW-M8?

[luigi bertolotti]

Advertisement (gone after registration)

I wonder if a black and white edition would sell as good as the white edition?

 

.

 

LOL... could be.. provided that the black and white edition would sport a richly colored finishing...

[danhefner]

This is an excellent idea.

 

All the comments about firmware/software aside, the issue here is the B&W-specific sensor. It would open the camera up to a level of sharpness that would be incredible.

 

If Leica were to do this (and if the result were as good as I think it would be), I'd very seriously consider buying one.

 

I would hope that the company would tweak the firmware to optimize for B&W images, especially in JPEG images, if they implement this.

 

Oh, PLEASE, if Leica were to go forward on this front, don't make a special orange edition of the camera. :-)

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[jonoslack]

Hi there

A couple of years ago I would have been panting for this camera . . . .but these days I like the ability to control the different colour channels when converting to black and white.

 

The extra resolution from a monochrome sensor was very seductive . . .when sensors were 6mp, but I don't often run out of resolution with my M8. I think I'd rather Leica concentrated on a Bayer M9 with a ff sensor and some more resolution.

[nzav]

In the German department of the LUF, we talked about the idea of a digital M just having a BW sensor built in.

So it could have nearly double the solution of a colour sensor by having the same size of sensor and of the pixel, so sharpness and clearness would be exampleless

The contrast and tonal range you could manage with yellow. orange, red or green filters, like we do or did with film, which is more effectiv than the post work in PS with a colour picture.

With this M-Leica it would be possible to make the highest quality BW pictures of all digital cameras by showing the strength of the M lenses really.

So much M-users use their Leicas just with a BW film, I think more that with any other type of camera, so I think, there would be enough costumers buying a BW-M-digital, including me.

 

Hi eckart,

 

I f I understand your post correctly, you are assuming the removal of the bayer filter array would somehow increase the M8's resolution (pixel count). This is, unfortunately, not the way it works.

 

The M8's 10.3mp will remain 10.3mp regardless of whether the bayer array is in place or removed. The sensor does not see in color; it only counts the photons that strike the photosites (pixels) and reports the light level achieved to the camera's CPU. The bayer array is indexed so the on-board processor knows which photosite has what color (R, G, or filter over it. The processor then notes the amount of light coming throught the filter (remember, the sensor does not see color, only light levels) and compares it to the surrounding photosites. The on-board computer then interpolates the assumed color of that photosite from the surrounding photosites data. This happens for every single active photosite on the sensor.

 

Once the interpolation is accomplished, the photosites become pixels with a color attribute. The actual color of a red-filtered photosite is only going to be a red pixel if the software determines from the surrounding photosites that it is red. This is a crude description of how color is derived from a piece of silicon that only responds to light levels, but I hope it clarifies why you would not gain resolution by eliminating the bayer filter.

[mjh]

removing the bayer layer would save 2/3 stop in light

In the case of the M8/M8.2, the efficiency of the red-, green-, and blue-sensitive pixels is 21, 40, and 32 percent, respectively, or 31 percent on average. Removing the RGB filters would increase the sensitivity by 1.69 EV – one stop more than your rather conservative estimate suggests.

Edited July 28, 2009 by mjh

[eckart]

I f I understand your post correctly, you are assuming the removal of the bayer filter array would somehow increase the M8's resolution (pixel count). This is, unfortunately, not the way it works....

 

No, Sir, I don't just want to remove the bayer filter from the existing M8 sensor, I want a digital M with a monochrome sensor and this could have, by using the same sensor and pixel size, a much better resolution.

I think you can compare it in a way to a the difference of a channel mixed BW-picture from a color negative to a highest resolution BW film picture. With the first possibility you can have a really good result, but you will never reach the original BW stuff, cause they are especially made for that.

M-photographers use a camera with restrictions compared to a DSLR concerning to the range of possibly lenses, because of the special benefits of this system and last not least,

for more than fifty years, for it's own and the highest quality, which this type can bring us back

So if there is a chance for this camera it must be a M.

Edited July 28, 2009 by eckart

[mjh]

Advertisement (gone after registration)

The M8's 10.3mp will remain 10.3mp regardless of whether the bayer array is in place or removed.

That much is obvious, but the point is that a b&w sensor without RGB filters arranged in the Bayer pattern would have the effective resolution (i.e. the ability to resolve fine detail) of an RGB sensor with about twice as many pixels. There is a trade-off between color resolution and spatial resolution, and if you give up color altogether, this pays off in a substantial increase in spatial resolution.

[Photoskeptic]

In the German department of the LUF, we talked about the idea of a digital M just having a BW sensor built in.

So it could have nearly double the solution of a colour sensor by having the same size of sensor and of the pixel, so sharpness and clearness would be exampleless

The contrast and tonal range you could manage with yellow. orange, red or green filters, like we do or did with film, which is more effectiv than the post work in PS with a colour picture.

With this M-Leica it would be possible to make the highest quality BW pictures of all digital cameras by showing the strength of the M lenses really.

So much M-users use their Leicas just with a BW film, I think more that with any other type of camera, so I think, there would be enough costumers buying a BW-M-digital, including me.

 

Now you're talking!

[farnz]

In the case of the M8/M8.2, the efficiency of the red-, green-, and blue-sensitive pixels is 21, 40, and 32 percent, respectively, or 31 percent on average. Removing the RGB filters would increase the sensitivity by 1.69 EV – one stop more than your rather conservative estimate suggests.

Michael,

 

Are you sure your figures are correct? Something seems wrong here because the most sensitive pixels are the green ones but the Bayer matrix already ensures that there are twice as many green pixels as blue or red so the sensor's output appears to be heavily weighted towards green.

 

Pete.

[mjh]

Are you sure your figures are correct?

Copied verbatim from Kodak’s spec sheet.

 

Something seems wrong here because the most sensitive pixels are the green ones but the Bayer matrix already ensures that there are twice as many green pixels as blue or red so the sensor's output appears to be heavily weighted towards green.

It is generally true that the green filters are the most sensitive ones. That’s because their transmission curve covers a relatively broad part of the spectrum, similar to the sensitivity of green-sensitive cells in the human eye. That’s why there is always more noise in the red and blue channels than in the green channel. (Incidentally this is also the reason why IR contamination, smearing, and purple fringing show up as purple/magenta rather than some other color.)

[tomasis7]

That much is obvious, but the point is that a b&w sensor without RGB filters arranged in the Bayer pattern would have the effective resolution (i.e. the ability to resolve fine detail) of an RGB sensor with about twice as many pixels. There is a trade-off between color resolution and spatial resolution, and if you give up color altogether, this pays off in a substantial increase in spatial resolution.

 

sounds good.. Mono M with cheap 1,33x 20mp sensor

[noah_addis]

There was a time when I would have thought this was a great idea. Now that I'm used to digital though, one of the great advantages is that you can shoot and get B&W and color from the same photograph.

 

I love B&W and want most of my work presented in B&W. But the fact is that sometimes clients want color. And with film there was always a compromise in trying to shoot both, since obviously color film could be converted to B&W but at the expense of the true look of a B&W film like Tri-X.

 

With digital, if I want to shoot B&W for a project but then a client wants to license a photo for use in color, I can do that if I choose to.

 

Personally, if I wanted to be locked into B&W, I'd just shoot film (and I'd be happy to do it!). The main reason I don't shoot film anymore is that the reality of the market is that I can sell more work in color and the M8 gives me the option of both very high quality B&W as well as color.

 

So I suspect many pros won't be into a B&W only camera, even though in theory it could provide high-quality results.

[tgray]

The way I see it, you'd gain the advantages of more effective resolution, better sensitivity, and the ability to use traditional filters. I'd go for it. Maybe

 

I don't think the fixed wavelength sensitivity would be that big of an issue. If you didn't like it, you'd tweak it a bit with filters. But I'm sure most would be perfectly fine with it.

 

While I grant that one of the advantages of digital is you get color and B&W at the same time if you want it, it's also one of the disadvantages.

[adan]

Reality checks:

 

First, I routinely capture 1-pixel-narrow details with the M8, thanks to the absent AA filtering (but with the Bayer pattern firmly in place). In both B&W and color. If I'm already getting 1-pixel-narrow details, how can a monochrome sensor of the same pixel pitch/count do any better? There is no such thing as half or 2/3rds of a pixel.

 

Color resolution is reduced somewhat by the Bayer pattern, but not luminance resolution (B&W brightness). I.E., a lppm test chart printed in contrasting colors will moire faster than one printed in B&W - but I don't shoot that many colored resolution charts.

 

Second, Kodak made several monochrome digital camera variants over the years. Kodak's total sales of monochrome digital cameras were under 2000 units over a decade.

 

This review is from 2002 Kodak 760m Review - I saw the original one developed with Leaf and the Associated Press in 1992 (along with their first color digital press camera - both 1.5 Mpixels & $25,000).

 

The original Kodak monochrome camera was "sharper" for the same resolution because Kodak did use a strong AA filter in the color camera, and left it out in the monochrome version. Kodak and the AP claimed monochrome 1.5 Mpixel shots could be enlarged to 11 x 14 prints - although that meant 90 ppi, so I guess their standards were not high.

 

The monochrome cameras do have higher ISOs - note, in the LL review linked above, that the base (lowest) ISO of the DCM760(m) was 400. The base ISO for the full-color DCM760 was 80. So assuming Kodak has used the same color filtration strength over the years (and Bayer was a Kodak engineer), an unfiltered monochrome-only 10.3Mpixel M8(m) should have ISOs of 800, 1600, 3200, 6400, and 12800.

 

So - I doubt that a monochrome M8 would deliver more than a 2-5% gain in base-ISO fine-detail resolution, and that only if the color structure of the subject was moire-prone.

 

But an M8(m) shot at ISO 800 would deliver a lot more detail undisturbed by noise than a color M8 shot at ISO 640.

 

Alternatively, although this would require a whole new sensor design, a monochrome M8 with 41 Mpixels in the same area as the current M8 sensor would deliver the SAME noise levels at the SAME ISOs as the current M8 does in color. Roughly. 4x the sensitivity, but for pixels 1/4 the area (and thus each receiving 1/4 the light).

 

Would I buy the 41Mpixel, high-res monochrome version? - No.

 

Would I buy the 10.3Mpixel, high-ISO monochrome version? - Maybe, but not as a sole camera.

 

Realistically - I'll wait for an M9 full-frame, and see what the new sensor and new (S2-based?) processor do for noise. Then revisit the issue.

Edited July 29, 2009 by adan

[tgray]

I think people were comparing a B&W sensor to color sensors that use AA filters. As you stated a B&W sensor wouldn't need to worry about the AA filter OR moire. With color, you either get the fuziness of an AA filter or moire. Take your pick. Most of the market chose to use an AA filter, though as you point out, Leica went the other way.

 

Regardless, I think the sensitivity gain is the really attractive one (to me). I agree though, very hard to market.

[eckart]

In my opinion the amount of sold Kodak Monos says nothing about today's selling chances of a Monochrome-M.

First, this cameras have been regarding to their possibilities to earn money for a professional incredibly expensive. So if somebody gave away this amount of bucks, he wanted colours.

The BW-M would be more of a FineArt-camera and as a Leica for sure not cheap, but compared to the situation in the end of the 90th or beginning of 2000 in digital solutions a much better price.

Second, a lot of M-photographers in the world do not want colours, they prefer a perfect B&W

Regarding the technical potence of this lady, I am not a electronic or an optic genius, but if Michael says it would have a nearly double effectiv resolution and a 1.69 f-stop better sensitivity, it is easy to believe, in this things he normally tells the truth.

[tomasis7]

Clean high iso like 12500 could be selling point. Lack of AA is even better. A few stops better dynamics than color one if one use software manipulation. Smaller files out of processing engine? because less info in BW..

 

so many nice ideas here in the thread...

[horosu]

I, for one, would be terribly interested in such a camera. Many photographers use B&W exclusively, so I guess there is a "niche" market here..

[perb]

I'm afraid if you require it to be an M camera for M lenses, then you will have to talk to Leica about it.

 

However, you can already today buy a special design B&W camera for astronomy or other technical purpose. They could be based on this 16MPx chip from Kodak:

 

http://www.ccd.com/pdf/ccd_16000.pdf

 

Kodak CCD's are very popular with astro cameras. Such a camera would typically be cooled for low noise and have full 16 bit A/D conversion. You wouldn't carry it around your neck, but photographing from a tripod and tethered to a laptop would be fine. I suppose using R lenses with such a camera would give great results.

 

You could even do colour photography with a motorised filter wheel

 

Regards

Per

[Advertisement]

Hallo,

Look here: Leica M

[nzav]

That much is obvious, but the point is that a b&w sensor without RGB filters arranged in the Bayer pattern would have the effective resolution (i.e. the ability to resolve fine detail) of an RGB sensor with about twice as many pixels. There is a trade-off between color resolution and spatial resolution, and if you give up color altogether, this pays off in a substantial increase in spatial resolution.

 

Explain please how you are increasing the resolution of a 10.3mp sensor simply by removing the Bayer filter array... Each pixel of the M8’s 10.3mp is represented by full color data points in the final image file. When you print your image, you are printing the full 10.3mp stretched across your print media, unless you crop or interpolate up to a higher pixel count. Removing the filter array will only produce a 10.3mp monochrome sensor with the same resolution of 3916 x 2634 pixels for DNG (10.31 MP), or 3936 x 2630 pixels (10.35 MP) for JPG. The only way to increase resolution of a given sensor is to artificially increase the number of pixels by interpolation. Where are you finding the extra 10.3mp to effectively double the resolution?

 

Every photosite (pixel) records only the monochrome light levels striking it and inserts them into the Raw image file; there is no color at this stage yet. All sensors are monochromatic in nature and require software to mathematically calculate color from the additional data that identifies what color filter is over which photosite (pixel). By comparing it to the surrounding photosites, the demosaicing software obtains the best guess about the actual photosite's color that it can manage. The more surrounding photosites sampled, the more accurate the final color attributed to the selected photosite, and now it becomes a pixel.

 

Whether converted in-camera, or later in C1 Pro, or ACR, etc., the captured monochrome 10.3mp Raw image file has color information added to each photosite during the Raw conversion process. The reason you can easily (well… sometimes) change white balance in Raw files is because you have no set color for every photosite – they a strictly levels of light as recorded. You can bias the color in any direction as needed. It is also the reason why the blue channel most often suffers from noise; it must be amplified to compensate for excess red component in common incandescent lighting. All amplified signals suffer from noise when amplified enough.

 

Perhaps you are suggesting an increase in dynamic range resulting from the filter array removal? In any case, you would be losing far more than gaining (except for the small ISO boost) by going this route. There are several special-purposed B&W senor equipped cameras that are attenuated to specific light waves such as UV or IR, but we are not discussing those, or are we?