M Monochrom: is "significantly higher resolution" junk science?

[sojournerphoto]

Very interesting. But just one pseudo-question: are you suggesting (that they're suggesting) that it is the sensor having a higher sensitivity towards the blue end of the [visible] spectrum? A yellow filter would absorb some blue, but let pass most everything else (in the visible spectrum). A light yellow filter has a filter factor of about 1 (or 0.5), and a medium yellow filter of 2.

In any case, it is usually the least light "hungry" of the colored filters, so I'm guessing they're trying to simplify the issue and not getting into technical stuff that most people now don't even care or even knew existed.






This is the one thing that is actually confusing to me: does the sensor have a de facto "absolutely panchromatic" (doubtful) response which can reliably (within reason) be expected to respond with the use of colored filters as one would in the "film world"? If not, one seems to be "locked in" as you would, say, having a camera perpetually loaded with Ilford Pan-F. Making the case for using a color sensor in this case. If you're into that sort of thing. Otherwise, I doubt any of this would be any more revelant than clicking "Desaturate" in Photoshop -- or simply change from RGB or CMYK mode to Greyscale.

As far as the resolution "issue"...I think a lot is being lost in translation between the engineering department and the marketing department. Marketing is not usually thought of as the "source of facts".

Gabriel

1. I agree - I think that Leica are simplifying, and may also be equating 'the film look' with 'the look of film with a yellow filter'. I don't know the spectral response curve of the sensor, but it will undoubtedly have its own interaction with filters. If you look at the curves for a few (panchromatic) films they can be very different.

2. The base sensor spectral response will be locked in and sensors have a linear response to intensity, hence including Silver Efex in order to adjust tonality (curves). Silver Efex won't be able to replicate differing spectral responses when provided with grayscale data of course.

I'm really pleased that Leica have made this, but there are a few things to work through if you want to buy and use it I suspect.

Marketing departments are not a reliable source of factual information I fear.

Mike

 

[sojournerphoto]

I don't know what spectral sensitivity they'll end up with, but once it's decided upon, you are stuck with it. And it's going to be roughly panchromatic (because not doing that would be silly for this kind of camera). If you want something different, you'll have to use color filters. So yes, the camera will always be loaded with Ilford Pan-F, or in this case, Truesense/Leica M-M.

You can look at the test images and see they don't look like trad B&W shot through a red or blue filter, or even a green one. So you can tell it's reasonably panchromatic already.

However, most of the B&W films I use, which are all panchromatic, have different spectral sensitivities. Tri-X is different from Pan-F is different from T-Max (as you state). They all respond reliably to different color filters. Sure a yellow filter on Tri-X is slightly different from a yellow filter on T-Max, just like those two films are different without filters. But I still have a good idea how a yellow filter will affect each of the films, and in use, it all works out just fine. The M-M should work pretty much the same. I expect the only real difference it will have is possibly greater UV or IR sensitivity than your standard film (say, Tri-X or HP5+).

If it has red sensitivity extended into the IR it might look a bit like Rollei 80S, and you could dig out the old IRUV cut filters to turn it into 'ordinary pan film and a nice cyan filter to give Adox CHS25.

I'll get me coat...

Mike

 

[bwcolor]

Falk Lumo has just posted a reasonably careful comparison of D800 and D800E resolution.

Comparison using a zoom lens? Doesn't that disqualify him from being described as "careful"?

 

[kuzano]

M9-Mono and some rabid thoughts....

M9-Mono and some rabid thoughts....

I have not been following the M-Mono development, and may have no right to post on the subject. However, I am registered on the forum, I have an internet connection, and in front of me I have a screen, keyboard and mouse, and I am logged in. For all intents that makes me as much of an authority on any subject posted on the forum as many I see posting.

Therefore, let me expound on some of my lesser lucid thoughts on the M9-mono subject. (less lucid thoughts occur between when the prescription runs out and I go to the pharmacy and refill it).

As I understand it the Leica company is coming out with an M9 that only shoots B&W and will be asking a thousand dollars more for it than the original M9. I gather that at this time, the $1000 bump is estimated?

Now, in my past experience as a computer consultant, I recall a time when Intel was producing the 486 CPU running at 25 Mh. (probably 20 years ago). It turned out that many of these processors were coming off the assembly line with malfunctioned floating point calculators. However, Intel soon discovered that they could change the "instruction set" in the processor and bypass the FPU. The resulting CPU became known as the i486sx processor and could be sold at a slightly lower price.

You all know the saying, "If the world gives you lemons, make lemonade". Well, intel is a master of this practice, and may well have been the source of the saying itself.

There was a ready market for less expensive computer at that time, because the going price for a 486 desktop computer with a 40Mb hard disk and 4 Mb of RAM was $2000.

Well, there are some interesting coincidences to the M9-mono tale. And again, without indepth study, I bring you my thoughts:

1) The M9 is experiencing some sensor failures in the field
2) Leica appears to be having some difficulty in sourcing sensors
3) Leica users are frustrated at the delays in M9 repair because of the "sensor shortage"
4) Leica usera are vocal about the problem
5) Leica may be also having problems producing new inventory of M9
6) Suddenly there is talk of a BW M9, thereby creating a new market
7) This new market may be taking the heat off Leica somewhat on the sensor shortage
8) The new M9-mono is going to be slightly more expensive than the color M9
9) Slightly in M9 pricing is measured at or near $1000.

Now, again short of medication, here is where my mind starts to make all sorts of interesting connections.

1) Is it possible that Leica now owns a large supply of sensors in which the ability to produce color has failed?
2) Is it possible that having been dealt this "lemon" situation, it has been determined that these sensors are adequate, if not better for monochrome use?
3) Is it possible that these "failed" sensors generated a huge "brain fart" in the marketing department as they turned the problem to "lemonade"?
4) This would be consistent with marketing style thinking, as engineers could never turn a negative into a positive. They would generate a "cost overrun" request and start over. Marketing people work with what's dealt to them and find ways to make it profitable.

Lastly, is it even remotely possible that Leica would rise to this level and "stick it" to a loyal fan base by raising the price on a failed product.......NAH!!!

 

[doolittle]

How does the dynamic range of the MM compare to an M9 or to various black and white films?

 

[Harry Lime]

So how much of a jump in dynamic range can we realistically expect?

The new base iOS of 320 suggests one stop, but with the color filters removed may we see more?

To me this is far more important than sharpness. 18mp is plenty, in color or b/w

 

[semilog]

So how much of a jump in dynamic range can we realistically expect?

The new base iOS of 320 suggests one stop, but with the color filters removed may we see more?

To me this is far more important than sharpness. 18mp is plenty, in color or b/w

Removing the color array should have little effect on DR. Just more base sensitivity.

In fact the achievable DR is likely to be worse on a monochrome sensor because there are tricks that can be used during the demosaic process to extract additional DR from the different color channels on an RGB array -- when one channel is saturated you can make inferences about luminance from the other channels. With a monochrome sensor that cannot be so easily done. About the best you can do is look at the shoulders of the saturated region and use a Gaussian or related fit to make an inference about what's going on in the region that's saturated.

On the other hand the monochrome sensor will likely have lower shadow noise under strongly biased (incandescent, fluorescent) lighting where one or more of the RGB channels is not being utilized to its potential.

There will be no universal answers here. The relative DR and SNR of the monochrome versus RGB sensors will depend on subject, lighting, and the ISO at which the cameras are operated.

My guess is that the RGB sensors will continue to do relatively better than expected because so much R&D is being focused on algorithm development for JPEG engines and RAW developers. Vastly more effort than is going into signal processing for monochrome sensors*.

*For pictorial photography. For machine vision, astronomy, microscopy, etc. there is obviously a huge amount of prior art dedicated to dealing with monochrome images.

 

[Tim Gray]

At the same time, DR might be a little better because noise from one noisy channel (B) won't be mixed in with the other channels during demosaicing, because you don't demosaic. But as semilog, said, in practice, I'd be surprised if it's a significant change in any direction.

 

[135format]

As I understand it the Leica company is coming out with an M9 that only shoots B&W and will be asking a thousand dollars more for it than the original M9. I gather that at this time, the $1000 bump is estimated?

Don't worry that $1000 dollars is just to cover the cost (and then some) of the free software you get with camera. 😉

The R&D on this camera must have been soooooo much less than the M9 so that explains it being the same price plus the $1000 for software. 🙄

 

[ampguy]

Jaapv, as you noted those images show no advantages in their reduced size.

A note on moire and lack of AA filters - depending on what you photograph, you may never see this. I have about a quarter million clicks on AA less digital cameras, and only a handful of images that show moire. The patterns that express it are rarely present in nature (landscapes), but are present in close-ups of man-made patterns. Most Leica/RF lens don't focus close enough to show it regularly. I wonder if Nikon D800e users will though.

So what's with the 16 bits of dynamic range? Does anyone have a medium (mechanical or electronic) that displays the theoretical increase? Will dxomark's tests show it?

I like the idea of the camera, wish it were full spectrum. But also admit that my EPL1 photos with weak AA filter, and use of the color for b/w filtering also take as good or better monochrome images than the M8/M9 did.

It would have been a bigger hit if Leica came out with this before the M9 color. Or even before the M8.

 

[Tim Gray]

Besides the obvious monochrome bit, I thought it had the same 'upgrades' that the M9-P did: same top plate, lack of markings (which must cost Leica *so* much money), and the sapphire cover glass. Whether or not the M9-P warrants the price increase over the regular M9 is another discussion, but if you accept that as fact, then it seems reasonable that a more limited production camera based on the M9-P with more software included costs the same the same as the M9-P.

 

[daveleo]

colour doesn't exist. It is a human perception, an interpolation from elctromagnetic waves. Sensors just measure electromagnetic waves and bayer filters filter wave bands not colour. Only we just assume that because the filters are made of material that we perceive as being Red, Blue or Green that the sensors are seeing colour. In reality they are measuring elctromagnetic waves of the wavelengths that we convert to colour in our visual perceptual system. So no sensors see in colour. They see in luminosity which is an accumulation of all the wavelengths that hit them.

Thank you enormously for that posting.

 

[ampguy]

Please explain

Please explain

Tim,

Would it be possible for them to have put different "films" or spectral sensitivity profiles in user selected firmware selections, even if requiring "re-booting" the sensor?

I wonder when the first M-M user is going to go this route and open it up a bit -

http://www.lifepixel.com/tutorials/infrared-diy-tutorials/olympus-e-pl1-ir

I don't know what spectral sensitivity they'll end up with, but once it's decided upon, you are stuck with it. And it's going to be roughly panchromatic (because not doing that would be silly for this kind of camera). If you want something different, you'll have to use color filters. So yes, the camera will always be loaded with Ilford Pan-F, or in this case, Truesense/Leica M-M.

You can look at the test images and see they don't look like trad B&W shot through a red or blue filter, or even a green one. So you can tell it's reasonably panchromatic already.

However, most of the B&W films I use, which are all panchromatic, have different spectral sensitivities. Tri-X is different from Pan-F is different from T-Max (as you state). They all respond reliably to different color filters. Sure a yellow filter on Tri-X is slightly different from a yellow filter on T-Max, just like those two films are different without filters. But I still have a good idea how a yellow filter will affect each of the films, and in use, it all works out just fine. The M-M should work pretty much the same. I expect the only real difference it will have is possibly greater UV or IR sensitivity than your standard film (say, Tri-X or HP5+).

 

[rxmd]

Would it be possible for them to have put different "films" or spectral sensitivity profiles in user selected firmware selections, even if requiring "re-booting" the sensor?

No. Spectral sensitivity is 100% hardware. This sensor is more sensitive to blue, which gives it a bit of a "flat" tonal curve.

Base ISO 320 is a problem, though. Losing the ability to shoot at low sensitivities is pretty bad if you want to work outside and don't want to use ND filters all the time.

 

[Tim Gray]

rxmd is spot on. The only way you could change it on the fly is a filter in front of or behind/in the lens. Behind/in the lens introduces aberrations if the lens design doesn't account for it.

Base ISO of 320 doesn't bother *me* as I shoot film - my fastest shutter speed is currently 1/1000. If I had an M-M with base ISO of 320, I'd also be gaining 1/4000 s, which would be equiv of 1/1000 ISO 80. I never really shoot anything slower than ISO 100 film anyway. Then again, on film I don't really care about overexposing by a stop or two from error, so maybe would be more problematic 😀

I can see how it might annoy those who are on an M9 right now and want to shoot at f/1.4 or something. But really, it's only a stop's worth of difference from the M9.

 

[ampguy]

Thanks

Thanks

So it is not yet known exactly what filtering over the CCD will ship yet? Or is the proposed blue level "tuning" for the JPG output only?

I found 160 high in the M8 as a base, and in the China M-M photo samples, I find the 10K iso unusabe, though "artful", I think it's usability is up to 2500/5K or in Canon/Nikon/Olympus terms, about 3200.


good info. on spectral sensitivity is here: http://www.scribd.com/doc/87464984/109/Spectral-sensitivity-of-digital-cameras

(around pages 211-212)

No. Spectral sensitivity is 100% hardware. This sensor is more sensitive to blue, which gives it a bit of a "flat" tonal curve.

Base ISO 320 is a problem, though. Losing the ability to shoot at low sensitivities is pretty bad if you want to work outside and don't want to use ND filters all the time.

 

[silverbullet]

In the worst case there are two different filters at the same time in certain situations:

ND + color filter. A nice gift from the times of the M8......

 

[jaapv]

No. Spectral sensitivity is 100% hardware. This sensor is more sensitive to blue, which gives it a bit of a "flat" tonal curve.

Base ISO 320 is a problem, though. Losing the ability to shoot at low sensitivities is pretty bad if you want to work outside and don't want to use ND filters all the time.

The spectral sensitivity at the moment is more biased to green according to very knowledgable beta testers. Not a bad thing at all, as that is the optimum frequency on most lenses. Maybe Leica should leave it at that. I quite see your point, but Leica was talking about firmware tweaks.

 

[jaapv]

If you find 10.000 ISO unusable you are simply wrong. The files are totally clean up to 2500, virtually noiseless up to 5000 and very good at 10.000 with noise that is subdued and much more tolerable than the digital noise we are used to. And the sensor is biased towards green.

So it is not yet known exactly what filtering over the CCD will ship yet? Or is the proposed blue level "tuning" for the JPG output only?

I found 160 high in the M8 as a base, and in the China M-M photo samples, I find the 10K iso unusabe, though "artful", I think it's usability is up to 2500/5K or in Canon/Nikon/Olympus terms, about 3200.


good info. on spectral sensitivity is here: http://www.scribd.com/doc/87464984/109/Spectral-sensitivity-of-digital-cameras

(around pages 211-212)

 

[semilog]

The spectral sensitivity at the moment is more biased to green according to very knowledgable beta testers. Not a bad thing at all, as that is the optimum frequency on most lenses. Maybe Leica should leave it at that. I quite see your point, but Leica was talking about firmware tweaks.

That's typical of good CCD sensors and pretty typical of most "panchromatic" films, as well. On a monochrome sensor there's pretty much nothing that you can do in firmware to alter the spectral sensitivity profile. Screw-on filters it's gotta be.