Would yo buy a B&W only 16 BIT M9 ?

[Sonnar Brian]

Well, if Leica does come out with this camera, I will procure one. I have a nice project that it would be perfect for. But I could not justify the entire NRE for producing one. But I am going to buy a Monochrome Digital camera for VNIR work. I'd love it to be an M9 as the APO lenses are available and you can see through the viewfinder with an 88a filter on it.

 

[biggambi]

I think this would be a significant statement by Leica, and it would set them apart from the competition in a very unique manner. I can not imagine a more satisfying system on a personal perspective, then a color M9 and a B&W M9. A thing worth striving for, wouldn't you say? Just like before, one camera body optimized with B&W film and the other with color film.

 

[kdemas]

I'd most likely buy one. The loss of the Bayer filter and the subsequent file quality gains would be pretty spectacular in my opinion.

What makes me laugh about this thread is that I discussed this with a very high (as in high up, not stoned) Leica official (he brought it up) well before the M8 was a reality. It seems this still percolates along in someones head 🙂

Kent

 

[biggambi]

I'd most likely buy one. The loss of the Bayer filter and the subsequent file quality gains would be pretty spectacular in my opinion.

What makes me laugh about this thread is that I discussed this with a very high (as in high up, not stoned) Leica official (he brought it up) well before the M8 was a reality. It seems this still percolates along in someones head 🙂

Kent

This is the most promising thing I have heard. Thank you for sharing it with us.

 

[Sonnar Brian]

There is a market in the technical field for a hand-held camera that takes interchangeable lenses that is Monochrome Visible+Near Infrared, 18MPixel, 16-bits per pixel. I asked another group would they buy such a camera for $10,000. We'd buy three of them for work, and some APO lenses. My last two camera procurements were over $20K each. The technical market tends to have reasonable budgets for lab equipment.

 

[jaapv]

I'd most likely buy one. The loss of the Bayer filter and the subsequent file quality gains would be pretty spectacular in my opinion.

In this thread it is taken as a given that the file quality would be significantly better for a Bayer-filterless sensor. I cannot see the reasons for this.

1. The Bayer filter and subsequent interpolation do indeed pool pixels, but that is only in the colour channels. The luminosity channel has the full resolution, regardless of the presence of a Bayer filter, as all pixels are recorded individually. So the result would be identical to removing colour information in post-processing, except for point 2.
2. The only gain we get is an increase in sensitivity, as no light is blocked by the filter with the attendant reduction of noise.
3. The results of a B&W image from the luminosity channel are less than optimal. The response of a sensor is clearly different from the response of panchromatic film.
4. That means there would have to be filters in front of the sensor to regain "panchromatic" sensor response, negating the gain of the removal of the Bayer filter.
5. When converting to B&W from a colour sensor, the information from the colour channels is essential to create a good B&W image. That flexibility would be lost.

So I have a strong suspicion that the results of a Bayer-filterless sensor for B&W would be worse instead of better.


Can anybody point to a credible publication that proves the opposite of my supposition?

 

[Sonnar Brian]

You do not need a Mosaic Filter to perform a gain flattening function for the spectral response of the sensor vs film, if you wanted to mimic it. I do not see why you would want to, when the response of the newer Silicon Sensor with improved blue response is fairly good across the spectral range. You would not need to do interpolation, and keep full spatial resolution.

http://www.kodak.com/global/plugins...ISS/datasheet/fullframe/KAF-6303ELongSpec.pdf

Spectral Response is on Page 13.

Example Film.

http://www.kodak.com/eknec/documents/13/0900688a802b0913/EN_ti0912.pdf

http://www.kodak.com/eknec/documents/1e/0900688a802b091e/ti1172.pdf

If anything, film is too sensitive in the Blue region, much more than the human eye, and goes too deep into the UV range. So which one does a better job of seeing Monochrome than the eye does...

In any event, The full spectral range of the Silicon sensor without any Spectral Gain Flattening would be better for the Technical Market, and I doubt a film user could tell the difference.

Besides, the Leica Monochrome Microscope Cameras do not use spectral gain flattening filters, and therefore Leica believes that they are not required.

 

[jaapv]

Ok, so you gain about one stop in sensitivity, which is nice for noise. You gain nothing in resolution, as you are in the equivalent of the luminosity channel of a standard sensor and the interpolation argument is neither here nor there. You lose the creative possibilities of colour information you would normally have in B&W conversion. So why bother?

 

[Sonnar Brian]

No color-aliasing because of the interpolation. And you pick up 12 rows and columns by getting rid of the interpolation algorithm. If you want to do a measurement of the intensity hitting the pixels, you don't want some interpolated value. I've been through this a long time ago with Kodak. They made the camera for me, we bought it.

Again- my interest for this camera on a professional basis is for scientific and technical applications. I'll take three. I'd also like it for near-IR work. For that, dumping the IR absorbing glass works wonders. On an M9, it's probably about 10 stops or so?

I think the difference here is the processing applied to an image. Is it more artistic, or more technical. It is a different world, requiring different salient features.

And- I have had Spectral Gain Flattening filters custom made for optical systems, mostly for optical amplifiers.

 

[kdemas]

I am in awe of the technical know-how of many of you and I do not even pretend to have the level of engineering insight that has been displayed in this thread. That said, I have heard repeatedly, from those in the camera biz and from photographers that used mono cameras in the past, that there would be huge gains in file quality with a B&W only application in something like an M8 (M9 now).

Take the example of the Kodak DCS 760m, a momochrome camera based on a Nikon chassis. Here's a ciip from a review in Luminous Landscape (just an example of what I have heard anecdotally):

Without an anti aliasing filter and no Bayer color matrix, the resolution of a 6 mega pixel monochrome camera is astonishing. In monochrome, 6 mega pixels effectively does what it takes 12-24 mega pixels with a color matrix.

Full article on this quirky camera can be seen here.

Anyway... that short quip reflects what the individual from Leica had told me so I find this whole discussion very interesting. Who knows, I'm just a marketing guy and a sometimes photographer 🙂 I can say that if such a gain were possible I'd be interested as most of my M work is B&W.

Kent

PS- I used to work in the semiconductor industry... you'd think I'd know more about the sensors!!

 

[Sonnar Brian]

I guess in the photographic world, spectral gain flattening filters are functionally like color correction filters...How come everybody just doesn't call them spectral gain flattening filters...

Spectral Response of the Human Eye.

http://everything2.com/title/relative+spectral+sensitivity+of+the+human+eye

If you look at the spectral response of the Kodak Sensor Array and truncate the Infrared portion with an IR cut filter, it is closer to the response of the Human Eye than the spectral response for the panchromatic films shown.

So- all you B&W film lovers, if you have to go digital- this will lesson the pain.

IR users- if you thought the M8 was good, dump the IR absorbing glass on the M9 and be amazed.

(kdemas, thanks.)

Fun Shots with the ancient Digital IR camera.

DCS200ir w 55/2.8 Micro-Nikkor, R60 filter. Hand-Held.

I hope my M8 is still working in 2030. Even if I'm not...

 

[biggambi]

Here's two cents worth.
1 cent: Elimination of the Bayer filter will increase luminance resolution, with the CCD sensor. This results partly from the not having direct luminance samples from each pixel and partly from the anti-aliasing filters that are needed in cameras to control color artifact generation. This array cut's luminance resolution by about 70%, this is a significant degradation. As an example, a 24mp sensor with a Bayer filter will resolve closer to an 18mp sensor without a Bayer filter. You will see more refined detail. (You eliminate this with layered photodiode sensing, but you gain other problems)
So, you do gain both sensitivity, as Brian has pointed out, and resolution.
2 cents: You actually attain an image with significantly less manipulation, much like film. It is a cleaner system vs the color sensor driven system. Again, this means a difference in the visible results.

Kindest Regards,

 

[JoeV]

Ok, so you gain about one stop in sensitivity, which is nice for noise. You gain nothing in resolution, as you are in the equivalent of the luminosity channel of a standard sensor and the interpolation argument is neither here nor there. You lose the creative possibilities of colour information you would normally have in B&W conversion. So why bother?

As for losing creative possibilities, what about good, old fashioned filters in front of the lens, like film photographers who shoot in B/W still do?

 

[jaapv]

Here's two cents worth.
1 cent: Elimination of the Bayer filter will increase luminance resolution, with the CCD sensor. This results partly from the not having direct luminance samples from each pixel and partly from the anti-aliasing filters that are needed in cameras to control color artifact generation. This array cut's luminance resolution by about 70%, this is a significant degradation. As an example, a 24mp sensor with a Bayer filter will resolve closer to an 18mp sensor without a Bayer filter. You will see more refined detail. (You eliminate this with layered photodiode sensing, but you gain other problems)
So, you do gain both sensitivity, as Brian has pointed out, and resolution.
2 cents: You actually attain an image with significantly less manipulation, much like film. It is a cleaner system vs the color sensor driven system. Again, this means a difference in the visible results.

Kindest Regards,

Ehhh..The M8/9 does not have an AAfilter...advantage gone....

All comparisons I found on the Web were with relatively low res sensors (5-6Mp) with thick AA filters. It is highly unlikely there will be any resolution advantage to M8/9 sensors
Removing the Bayer filter would maybe give a marginal gain in acuity, similar to removing a colour filter from a lens, but that would be offset by the need for a clear protective filter in front of the sensor, if only to allow the user to clean it, and by the need to use yellow, red etc filters in front of the lens.

The only plus I can see would be about double the light striking the sensor, i.e. a one stop gain in sensitivity and S/N ratio.

The minus would be the loss of colour capability and creative use of colour to get a perfect gradation.
The price would be 50-100% more that a standard camera.
To be honest my estimate of world-wide sales would be less than 100.

 

[biggambi]

Ehhh..The M8/9 does not have an AAfilter...advantage gone....

I am well aware that the M8 & M9 does not have an AA filter, I have included this for reference to those cameras that do, hence the example given with the 24mp japanese wonder machines, etc.

The first stated factor does pertain directly to the M8 & M9. I did not give an example as to the percentage difference between B&W versions of these systems as I have no means to test them. There will be a difference and it will be measurable. Unless you are going to assert there is no Bayer filter.

As does the second stated point regarding the systems algorithms. All will effect the final image integrity, and the B&W system benefits from avoiding these pitfalls.

I can go deeper into detail but it would be extensive, and difficult as I am really not looking to write a paper on it. So, I would rather that you seek out journal articles or a text book in the engineering of these systems. If you are interested, maybe I can run a search on a scientific search engine. Or, email you something. This is not my area in physics, but I have some basic knowledge in it.

So, there are two direct key factors that would change within the M8 & M9, and there are three key factors that differentiate it from other digital systems.

I hope this makes the point clearer.

Kindest Regards,

 

[biggambi]

Brian,

You post one more picture of what my M8 cannot do, and you may have a burglary. We have got to see this thing gets built. Your killing me.

M.

 

[jaapv]

Imo, if one wants a Near-IR high resolution camera, as Brian states, I thnk that taking an M9 and replacing the IR filter by a plain filter, as some companies offer, would come rather close. I'll happily grant you a higher acuity for a non-Bayer camera, but I would like to see comparisons, as my doubts for an advantage in real-life photography remain. After all, 95% of all B&W photographs do not use more that a 5 Mp resolution in print , and for web use maybe 2,so any gains would only be useful for scientific use and those very few photographers that actually need them. The same goes for the M9. The undoubted quality advantages it has over the M8 do not show up for a considerable number of users. One could certainly say the same for the current 20Mp + high-end cameras of other makers.
For myself, I can only say that any image quality gain over the M9 would be invisible to me, both on the web and in print.

 

[user237428934]

Brian,

You post one more picture of what my M8 cannot do, and you may have a burglary. We have got to see this thing gets built. Your killing me.

M.

Are you referring to #118? Yes indeed. My (and your) M8 can not take such unsharp images.

 

[Sonnar Brian]

You try using your M8 hand-held for 1.25:1 Macro images and see how well you do.

One of the reasons I would buy a VNIR M9 is for the improved resolution. However, using the DCS200ir for work, I could do some quick-and-dirty experiments that proved a project would work. Funds would be more than adequate for procuring two VNIR M9's if they were produced, budgeting $10,000 per camera. So maybe these images look unsharp to the untrained eye, but the camera has proved to be a good investment.

This camera was made in 1993. I wonder how many M8's will still be in use in 2030. Maybe mine will.

 

[antiquark]

You guys are talking like sharpness actually matters! 🙂