[falconeye]

Yes. But, it seems to me - 300-400 pair of lines and only at black-and-white technical film.
Add special development and the best scanning.

I would say about 150-200 lp/mm - it's more close to reality.

Correct me, if I'm not right.

The resolution of a lens (prior to the digital age) was measured on the optical bench directly measuring MTF with an analog device using a sine-stripe pattern of a given spatial frequency. The limiting resolution was defined to be the MTF5 value (sometimes, MTF10). MTF5 or MTF10 (read 5% or 10% remaining contrast) seems not much but it is enough to digitally restore the original contrast (aka sharpening).

In the analog age, no differentiation between lines/mm (LPM) and line pairs/mm (lp/mm) was made. It is made now to make it clear that 1 lp/mm = 2 pixels/mm.

So, 300 lp/mm produce 311 megapixel data on a 24x36 sensor) and you're right, outresolves most B&W film which normally doesn't resolve beyond 150 lp/mm. But that's the film's limit, not the lens' limit.

I used ADOX CMS 20 plus ADOTECH which remains grain-free up to ~ 400 lp/mm. It is the world's highest resolving film and still unparalleled by any digital camera except some German line scan cameras. And as you can see, the DA*300 even seems to outresolve the ADOX grain.

[gazonk]

And as you can see, the DA*300 even seems to outresolve the ADOX grain.

Wow, impressive! Anyone tried the Q with the DA*300 yet...?

[ManuH]

This gives you a feeling what we are still missing with current digital sensors ... (read, we're going to see 100MP FF cameras and it would still not resolve all detail from my B&W film sample

Excellent demonstration. I'm quite tired of reading the "internet wisdom" saying that 16 is too much, 18 is too much, 24 is too much because we'll need new lenses. Processing power is the big issue with more megapixels, on the camera and on the computer. Specially on the camera. But if the accompanying technology is there (SDXC faster write speed, faster CPU, etc) then I'm all for more megapixels. And of course the other IQ parameters (noise,DR, etc) should be as good or better than lower mp sensors.

[philbaum]

benjikan is right. High and low megapixel counts are relative to what we are used to.

Let me try to give you a more absolute feeling of where we stand ...

1. Left: an extreme crop of the center where pixels from a hypothetical 24 MP sensor are clearly visible. Just to show how large such pixels are.
2. Right: the same crop, but from a real shot with the DA*300 lens onto film. I used a special film emulsion with very fine grain and the crop was taken via a microscope because no film scanner resolves this much.

This gives you a feeling what we are still missing with current digital sensors ... (read, we're going to see 100MP FF cameras and it would still not resolve all detail from my B&W film sample -- e.g., you see that the circular lines have a flat rather than a rounded end). It's also a reason why I said that a Pentax Q adapted to a DA*300 will be an interesting experiment

Correct me if i'm wrong, but what Falk proved was exactly what was in the title: Sensors are the limiting factor, not the lenses. To prove that conclusion, Falk had to use a microscope attached to a K20 camera. (Brings a whole new meaning to the term "pixel-peeping") But he didn't prove anything with regard to what resolution we need on a practical basis.

Look at what's happening to the pictures we take. Resolution is getting so good with the K20 and K5 and similar competitive brand cameras, that women want special processing so that not every pore is showing in their faces. I know some audubon members who seem to feel that their bird pictures are not sharp enough if the individual branches of each feather are not readily seen (these images almost seem freakish at times). The details so exposed on these birds are often in excess of what we can normally see with our unaided human vision.

For satellites and other military use, research work, etc., one can see uses for ultra high resolution (whatever that means) images, but for a candid street picture or an ID picture for someone's passport or credit card, do we need that same level of resolution?

Buy whatever you want, but i think we're reaching a point where the money poured into ever higher resolution sensors could have been better spent in improving cameras in other ways. For example, i would rather have the following features rather than resolution above 16 megapixels:

a. more reliability (my humble PS Canon with 7 megapixels has never had its firmware updated or needed to be sent to a shop)

b. quiet shutters like the K5, i hate the snap of most cameras

c. better focusing at low light

d. cheaper prices so if they get stolen or dropped, its not that big of a deal. Cheaper also means i don't need to pay for insurance for it.

e. No 1/180s synchro limitation.

We're so used to resolution getting higher and higher with every new model, that we don't question it most of the time. But when i saw Falk using a microscope to ferret out information on a lens' maximum resoluion, it struck me as almost comical as to how specialized our cameras are getting, and not in a good way.

In writing this, i find myself reaching the conclusion that in one or two years, once APS sensors are available in mirrorless cameras, with the features i want, and the lenses i want, I will readily give up the pursuit of the next greatest dslrs with their 100 megpixels. And it isn't just the cost of the camera, its upgrading the computer systems and storage systems to deal with the every larger data files.

[falconeye]

But he didn't prove anything with regard to what resolution we need on a practical basis.
...
But when i saw Falk using a microscope to ferret out information on a lens' maximum resoluion, it struck me as almost comical

I agree.

I didn't intend to say anything about what makes sense or not.

BTW, the microscope was attached to film. The K20D was only there to share information on the internet. There is nothing "new" in what I did. I just illustrated what lenses can do and probably can do since 100 years.

Some artists actually made use of this giant resolution by doing B&W fine art printed in giant size to hang on walls. Such photos exist since long before the digital age.


So while I generally agree with what you said, philbaum, I don't exclude the possibility that new artistic ways of expression will emerge which will make good use of the resolution. Who knows.

I already see the "fly-into" gigapixel panos pop up everywhere.

One application I foresee are digital image walls in our homes. A 200" x 100" photo wall with 100 dpi will need 200 MP, at 240 dpi will need 1 gigapixels. If I sit in front of my photo wall eating my breakfast, I don't want to see pixels or blur. I want to see the beach as if I were sitting just there it. A virtual wall-sized window. A TV would just be an overlay rectangle on top of such a wall. Printed OLEDs could make this possible and affordable when and if research makes the corresponding breakthrus.

So, unlike many, I see a need for 100MP+ cameras but I don't see a need for 3D cameras as long as I don't see a technical solution for the presentation side of things.

[philbaum]

I agree.

I didn't intend to say anything about what makes sense or not.
...

So while I generally agree with what you said, philbaum, I don't exclude the possibility that new artistic ways of expression will emerge which will make good use of the resolution. Who knows.

One application I foresee are digital image walls in our homes. A 200" x 100" photo wall with 100 dpi will need 200 MP, at 240 dpi will need 1 gigapixels. If I sit in front of my photo wall eating my breakfast, I don't want to see pixels or blur. I want to see the beach as if I were sitting just there it. A virtual wall-sized window. A TV would just be an overlay rectangle on top of such a wall. Printed OLEDs could make this possible and affordable when and if research makes the corresponding breakthrus.

So, unlike many, I see a need for 100MP+ cameras but I don't see a need for 3D cameras as long as I don't see a technical solution for the presentation side of things.

Falk
I've always appreciated your prodigious technical analysis that you've provided on various subjects. And you are precise in your logic path.

I find today's DSLR camera marketplace to be a fascinating study. Market leaders Nikon and Canon seem to be loathe to introduce product change in any way that would undercut their current lucrative product lines (i think that explains why they are so late into the mirrorless ILC products). Opposed to them, you find Sony, Panasonic, Olympus, and Fuji pushing any innovative changes which will increase their market share. Its classic free market capitalism at its best.

Although we haven't seen it yet, i am coming around to the conclusion that others have voiced long before me, that the MILC are going to reduce the dinosaur population in a significant way. But yes, Falk is also right - we eventually will have 100 megapixel dslrs, just not in my hands

Cheers, Phil

[ManuH]

we eventually will have 100 megapixel dslrs, just not in my hands

You may not care when processors will run at 20Ghz on 32 cores, lower end SDXC cards will have 2T and Hard disks 500TB. These figures may look big but I remember my first Smart Media card was 8Mb (Mb not Gb). And of course it costed more than any current 16Gb SDHC card.

At that time when 100mp sensor will be a commodity I'm sure we will see long forum debates when the 150mp will be announced and how our lenses will not be able to resolve that "50% more resolution" (maths still wrong), how CPU have to be upgraded etc etc.

[falconeye]

You may not care when processors will run at 20Ghz on 32 cores

That seems to underline my point.

However, one should not take such things for granted.

Technology drives applications and vice versa. This circle can break on either side.

Moore's law is broken for CPU clock speeds which didn't increase the last ten years. And we did not (yet) fly to Mars.

This is why it does matter what 100MP+ would be good for.

[ManuH]

Moore's law is broken for CPU clock speeds which didn't increase the last ten years. And we did not (yet) fly to Mars.

We could fly to Mars, I believe it's more a matter of money in this case. But why go there physically anyway?

CPU clock speeds did improve a bit during the last five years, and I agree there seems to be a wall here. OTOH processing power DID improve, may not be at the same rate as Moore's law but the i7-3400 is noticeably faster than the Q6600 of 4 years ago. And more overclockable too BTW.

I know this is nitpicking but regarding MP I'm sure the camera manufacturer balance this with the processing power they have on hand. They already have the technology to produce 100MP sensors, at least the pixel pitch.

This is why it does matter what 100MP+ would be good for.

For marketing people I guess

As for computers they are already able to process 100MP+ pictures. I did a stitching of 400MP the other day and it was quite manageable on a modern i7 with 8Gb RAM. And 2TB drives are now less than 100$.

The Nikon V1 Expeed 3 process 10MP @60fps. In theory the same processor could do 100MP @6fps!

64Gb SDXC cards are now available with 128Gb not too far away.

So 100MP sensors are probably not that far away either.

[gazonk]

CPU clock speeds did improve a bit during the last five years, and I agree there seems to be a wall here. OTOH processing power DID improve, may not be at the same rate as Moore's law but the i7-3400 is noticeably faster than the Q6600 of 4 years ago. And more overclockable too BTW.

Newer processors get faster all the time. The P4s won the GHz race, but they were extremely slow power-hogging monsters with their ridiculously long pipelines. So just looking at GHz can be quite misleading.

Still I agree that there has been some stalling. But there are several possible ways to new processor technology breakthroughs, for instance possible graphene-based microprocessors.

[feishui]

Well, maybe your computer and camera can handle those gigantic pictures, but like a barrel, you have so many boards and one short board would be fatal. For example, I browse photos 95% of the time on a laptop screen. My guess is 80% of the people look at their photos in a similar way. I just don't see how many more pixels can be put in a laptop screen, and even if more can be put, can we see the difference with our naked eyes?

I am not saying it is not important to increase the pixels in a camera sensor, but this will be not as interesting as optimizing the way people look at their photos, like 3D or retina projection

We could fly to Mars, I believe it's more a matter of money in this case. But why go there physically anyway?

CPU clock speeds did improve a bit during the last five years, and I agree there seems to be a wall here. OTOH processing power DID improve, may not be at the same rate as Moore's law but the i7-3400 is noticeably faster than the Q6600 of 4 years ago. And more overclockable too BTW.

I know this is nitpicking but regarding MP I'm sure the camera manufacturer balance this with the processing power they have on hand. They already have the technology to produce 100MP sensors, at least the pixel pitch.



For marketing people I guess

As for computers they are already able to process 100MP+ pictures. I did a stitching of 400MP the other day and it was quite manageable on a modern i7 with 8Gb RAM. And 2TB drives are now less than 100$.

The Nikon V1 Expeed 3 process 10MP @60fps. In theory the same processor could do 100MP @6fps!

64Gb SDXC cards are now available with 128Gb not too far away.

So 100MP sensors are probably not that far away either.

[jogiba]

64Gb SDXC cards are now available with 128Gb not too far away.

http://sandisk.com/products/dslr/sandisk-extreme-pro-sdhcsdxc-uhs-i-memory-cards



Lexar 128GB Professional 133x SDXC Card LSD128CRBNA133 B&H Photo



Delkin Devices 64GB Elite SDHC 633 UHS-I DDSDELITE633-64GB B&H

[Aristophanes]

Still I agree that there has been some stalling. But there are several possible ways to new processor technology breakthroughs, for instance possible graphene-based microprocessors.

The software has to catch up. This has been the bottleneck. 20 guys can design and litho the most amazing processor, but it takes 10,000 programmers to make products for the market and support.

[gazonk]

The software has to catch up. This has been the bottleneck. 20 guys can design and litho the most amazing processor, but it takes 10,000 programmers to make products for the market and support.

The problem is that the programmers get lazy as the processors speed up, MS Word actually feels (and I think it is!) slower today than it did on my 1992 mac. On the other hand, that mac almost grinded to a halt if I tried to view a 0.5mp image on it, so some things have improved

[froeschle]

DxOMark - Sony A77, measurements and review of the world-first 24 MP APS-C camera