Sony to unveil a 50 MP new A7 body at Photokina?

[jrista]

The DXO info perfectly supports my claim. Jumping to a 50mp sensor, your never going to get close to that resolution with most of your lenses.

You didn't link to anything that supports this claim. For however significant you consider the jump to 50 MP in the first place, that gain is not going to be muted or blocked by better lenses. Glass is not (yet) the limiting factor, at least not at the upper end.

I did link it previously and only the zeiss APO 135mm was able to get the maximum 36mp out of the camera. A lowly 24-70mm wasn't even close to getting the details out of the d800. You should check the link.

To really get 50mp worth of detail, don't expect anything less than the most expensive glass to achieve it. Sure, I understand there will be gains but in my original post here, I stated most lenses will never get close to resolving the full 50mp in that sensor. Just like most lenses won't get even close to resolving the majority of the 36mp in the current d800.

First...no lens could ever get close to resolving the actual megapixel count of any sensor unless your talking about diffraction limited performance at extremely wide apertures...like, f/2 and faster wide. It's an asymptotic relationship...theoretically, it's actually impossible for an undersampled lens to ever actually resolve the full megapixel count of a sensor.

That's why DXO's numbers are such a joke. To claim that ANY lens could actually resolve "36mp out of 36mp" period is just flat out impossible. Not even the Zeiss could do that.

I would also be willing to bet that if you could slap a Canon 24-70 II onto the D800, that it would resolve a hell of a lot more detail than the Nikon 24-70. The Canon lens scores the same as the Nikon lens (both score 28), despite the fact that the 5D III has 62% fewer pixels than the D800. The Canon lens actually resolves 18mpix according to DXO, vs. the Nikon lens which resolves 16mpix.

It is a very common occurrence on DXO for Canon lenses to score the same as Nikon lenses, when used on the 22.3mp 5D III vs. the 36.3mp D800. In the majority of those cases, Canon lenses score higher on the MPix scale, despite the lower pixel counts. (The real flaw in DXO's lens scoring is their massive overweighting of "Transmission"...which completely imbalances resolving power tests in favor of benchmarking how much vignetting is occurring (an issue that merely requires a single click to fix in post). That's in line with the core theory of how lenses and sensors work together to resolve...one doesn't simply outresolve the other and that's that.

 

[deleteme]

That would make me buy an a7. Really? 50mp? This is madness...

Then you would have to use a Sony. :-\

 

[RLPhoto]

The DXO info perfectly supports my claim. Jumping to a 50mp sensor, your never going to get close to that resolution with most of your lenses.

You didn't link to anything that supports this claim. For however significant you consider the jump to 50 MP in the first place, that gain is not going to be muted or blocked by better lenses. Glass is not (yet) the limiting factor, at least not at the upper end.

I did link it previously and only the zeiss APO 135mm was able to get the maximum 36mp out of the camera. A lowly 24-70mm wasn't even close to getting the details out of the d800. You should check the link.

To really get 50mp worth of detail, don't expect anything less than the most expensive glass to achieve it. Sure, I understand there will be gains but in my original post here, I stated most lenses will never get close to resolving the full 50mp in that sensor. Just like most lenses won't get even close to resolving the majority of the 36mp in the current d800.

First...no lens could ever get close to resolving the actual megapixel count of any sensor unless your talking about diffraction limited performance at extremely wide apertures...like, f/2 and faster wide. It's an asymptotic relationship...theoretically, it's actually impossible for an undersampled lens to ever actually resolve the full megapixel count of a sensor.

That's why DXO's numbers are such a joke. To claim that ANY lens could actually resolve "36mp out of 36mp" period is just flat out impossible. Not even the Zeiss could do that.

I would also be willing to bet that if you could slap a Canon 24-70 II onto the D800, that it would resolve a hell of a lot more detail than the Nikon 24-70. The Canon lens scores the same as the Nikon lens (both score 28), despite the fact that the 5D III has 62% fewer pixels than the D800. The Canon lens actually resolves 18mpix according to DXO, vs. the Nikon lens which resolves 16mpix.

It is a very common occurrence on DXO for Canon lenses to score the same as Nikon lenses, when used on the 22.3mp 5D III vs. the 36.3mp D800. In the majority of those cases, Canon lenses score higher on the MPix scale, despite the lower pixel counts. (The real flaw in DXO's lens scoring is their massive overweighting of "Transmission"...which completely imbalances resolving power tests in favor of benchmarking how much vignetting is occurring (an issue that merely requires a single click to fix in post). That's in line with the core theory of how lenses and sensors work together to resolve...one doesn't simply outresolve the other and that's that.

You made my original point exactly. The nikkor 24-70mm would be a poor performer to get 50mp of detail while a better lens like the Canon would get closer, and the Zeiss would get the very closest.

And again, the 16-35s, 24-70,70-200s are very common good lenses but even they wouldn't be the lenses that could get the most out of a 50mp sensor. To do so, you'd need those primes and if that's the case, the actual details rendered from a7r 36mp----->50mp a7x wouldnt be stark/negligible unless you had an otus or 135mm apo.

Now if all lenses performed at the level of the otus and we all had them already, great no complaints from me about more resolution.

 

[ULFULFSEN]

Now if all lenses performed at the level of the otus and we all had them already, great no complaints from me about more resolution.

So afterall there is a market, thought not a big one as for Rebels.

Zeiss lenses are not so expensive when you compare it to other Medium Format glas.
Nobody said this 50MP camera is for everyone.

You sure don´t need it for images you put on a website of Facebook.

 

[moreorless]

The issue for me wouldn't be that you would see no improvement with 50 MP but that I'd question how much demand there would be for such a camera on the Sony FE mount.

You could argue that it might appeal to those adapting Canon DSLR lenses I spose, something like the new 24-70mm would probably get much of the benefit but really is that a big market?

The FE system thus far seems to have prioritised size over performance for all its wideangle options and whilst the 70-200mm f/4 seems a bit better in that reguard is anyone going to choose the FE system for say wildlife shooting with its AF performance?

 

[dtaylor]

I did link it previously

The page I saw did not in any way support what you are claiming.

To really get 50mp worth of detail, don't expect anything less than the most expensive glass to achieve it. Sure, I understand there will be gains but in my original post here, I stated most lenses will never get close to resolving the full 50mp in that sensor.

I think the #1 problem is that the way you are describing things is fundamentally wrong. MP is not a measure of resolved detail, but a measure of sensels on a chip. If you put a Coke bottle in front of a 50 MP chip, you still get 50 MP. There's no such thing as "real" or "fake" 50 MP or getting "50 MP worth" of detail.

If you are looking at DxO "MP equivalent" scores for lenses they are wrong to present the data that way, and (yet again) observably wrong even if you try to translate their incorrect usage of terminology into something valid.

lpmm for a given target MTF point is a measure of resolved detail.

A 50 MP sensor + just about any lens will yield an observable system gain in lpmm over most or all of the MTF curve. The lpmm resolved will be higher with better glass of course.

Now I would be the first to say that we are into diminishing returns and unless you are printing really, really big there's not going to be much to gain by jumping to 50 MP. But I would not say that 50 MP is invisible or completely worthless with everything except a handful of super expensive lenses.

And I'm not just saying that from theory. Observation trumps theory, and you can readily observe IQ gains on 18-24 MP APS-C sensors with glass that's already on the market. Those lenses will yield similar gains on 40-50 MP FF sensors. I'm not as familiar with Nikon glass, but there is a lot of Canon glass which would work well on a higher resolution FF sensor.

 

[RLPhoto]

I did link it previously

The page I saw did not in any way support what you are claiming.

To really get 50mp worth of detail, don't expect anything less than the most expensive glass to achieve it. Sure, I understand there will be gains but in my original post here, I stated most lenses will never get close to resolving the full 50mp in that sensor.

I think the #1 problem is that the way you are describing things is fundamentally wrong. MP is not a measure of resolved detail, but a measure of sensels on a chip. If you put a Coke bottle in front of a 50 MP chip, you still get 50 MP. There's no such thing as "real" or "fake" 50 MP or getting "50 MP worth" of detail.

If you are looking at DxO "MP equivalent" scores for lenses they are wrong to present the data that way, and (yet again) observably wrong even if you try to translate their incorrect usage of terminology into something valid.

lpmm for a given target MTF point is a measure of resolved detail.

A 50 MP sensor + just about any lens will yield an observable system gain in lpmm over most or all of the MTF curve. The lpmm resolved will be higher with better glass of course.

Now I would be the first to say that we are into diminishing returns and unless you are printing really, really big there's not going to be much to gain by jumping to 50 MP. But I would not say that 50 MP is invisible or completely worthless with everything except a handful of super expensive lenses.

And I'm not just saying that from theory. Observation trumps theory, and you can readily observe IQ gains on 18-24 MP APS-C sensors with glass that's already on the market. Those lenses will yield similar gains on 40-50 MP FF sensors. I'm not as familiar with Nikon glass, but there is a lot of Canon glass which would work well on a higher resolution FF sensor.

The page supports it. You will never get 50mp or even close to it with most lenses. The jump from 36mp to 50mp will be slight with most lenses and stark with supreme lenses. Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass. That's why cropping FF 22mp vs aps 18mp the results are nearly identical. On paper they shouldn't be, but on the other thread, negligible.

 

[surapon]

Dear Friends.
This Attachment is the Old News in August 24, 2010 , about Canon New 120 MP sensor = 4 years ago.
BUT, Why, Now 2014, The Best Sensor that on the Market by Canon = only 22.3 MP---- ??, And Top of the Line of Canon = 1Dx only = 18.1 MP..
May Be ( ??) ,Past 10 years, Only 1.23 Billion users of Facebook( Like ME ) have post the Photos on Face book need only 2 MP Cell Phone Camera's photos of Foods that they eat at morning/ Breakfast , to show their friend---NOT 50 MP. Sony Camera, Now.

http://www.dpreview.com/articles/2773376832/canon120mpsensor

Enjoy
Surapon

http://www.theguardian.com/technology/2014/feb/04/facebook-10-years-mark-zuckerberg

Thank you Mr. Surapon for attempting to lighten the discussion a bit. I always appreciate your contributions and the optimism you bring to this forum. BTW, is Biscuitville still in operation down there? They made the best breakfast!

Ha, Ha, Ha, Dear friend Mr. Skatol
You are welcome, Sir--------For all High tech Discussion, We will let our Teachers and all of my friends take care. But We try to Light it Up, and Just For Super FUN.----Human life are too short.
Buy the Way, Yes Biscuitville are one of the best Breakfast , But $ 5 US Dollars is to much for me, At Bojangles = 3.50 US Dollars with Big Great Cup of coffee too----My Wife id so tight for High cost of Breakfast----Ha, Ha, ha. Nice to talk to you, Sir.
Have a great weekend.
Surapon

 

[jrista]

The page supports it. You will never get 50mp or even close to it with most lenses. The jump from 36mp to 50mp will be slight with most lenses and stark with supreme lenses. Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass. That's why cropping FF 22mp vs aps 18mp the results are nearly identical. On paper they shouldn't be, but on the other thread, negligible.

The point I was trying to make before was that you can't get 50mp with any existing lens, and probably wouldn't with any lens created within the next decade. The same goes for 36mp, 24mp, 18mp. You cannot actually resolve those resolutions with ANY lens, even the best of the best of the best. Because output resolution has an asymptotic relationship with the least resolving component of the system. To even get close to 50mp, you would probably need to be shooting a lens like the Otus at f/1.4 (assuming it's diffraction limited at that aperture...if not, then you would need a lens even better than the Otus).

Assuming you don't have a crappy lens, then you can realize improvements by moving to a higher resolution sensor. Every time you do, the nyquist limit drops. That allows more information to be resolved usefully. It might be resolved at lower contrast, but until your down near the Rayleigh limit, you can still do stuff with lower contrast detail (it's more work to enhance it, but it can be done.) If you jump from 18mp to 50mp, even with just an "ok" lens, your going to see a huge difference. The frequency of detail that might have been resolved crisply at 18mp will now probably look a little soft...however, your going to be resolving a level of detail the 18mp couldn't even see at all. The same would go for the difference between 36mp and 50mp...although the amount of smaller details you resolve wouldn't be as significant, and would require more work to enhance.

For any given lens, going from 24mp to 36mp, then to 50mp, WILL realize an improvement. The improvement might start out at around 18% for the jump from 24-36mp, then drop to 11-12% for the jump from 36-50mp. It might be a mere 5% in a jump from 50mp to 70mp. Beyond that, your probably within the margin of error...however, then your in the realm of oversampling. That has a whole 'nother set of benefits.

 

[RLPhoto]

The page supports it. You will never get 50mp or even close to it with most lenses. The jump from 36mp to 50mp will be slight with most lenses and stark with supreme lenses. Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass. That's why cropping FF 22mp vs aps 18mp the results are nearly identical. On paper they shouldn't be, but on the other thread, negligible.

The point I was trying to make before was that you can't get 50mp with any existing lens, and probably wouldn't with any lens created within the next decade. The same goes for 36mp, 24mp, 18mp. You cannot actually resolve those resolutions with ANY lens, even the best of the best of the best. Because output resolution has an asymptotic relationship with the least resolving component of the system. To even get close to 50mp, you would probably need to be shooting a lens like the Otus at f/1.4 (assuming it's diffraction limited at that aperture...if not, then you would need a lens even better than the Otus).

Assuming you don't have a crappy lens, then you can realize improvements by moving to a higher resolution sensor. Every time you do, the nyquist limit drops. That allows more information to be resolved usefully. It might be resolved at lower contrast, but until your down near the Rayleigh limit, you can still do stuff with lower contrast detail (it's more work to enhance it, but it can be done.) If you jump from 18mp to 50mp, even with just an "ok" lens, your going to see a huge difference. The frequency of detail that might have been resolved crisply at 18mp will now probably look a little soft...however, your going to be resolving a level of detail the 18mp couldn't even see at all. The same would go for the difference between 36mp and 50mp...although the amount of smaller details you resolve wouldn't be as significant, and would require more work to enhance.

For any given lens, going from 24mp to 36mp, then to 50mp, WILL realize an improvement. The improvement might start out at around 18% for the jump from 24-36mp, then drop to 11-12% for the jump from 36-50mp. It might be a mere 5% in a jump from 50mp to 70mp. Beyond that, your probably within the margin of error...however, then your in the realm of oversampling. That has a whole 'nother set of benefits.

Your absolutely correct and that's what I've been saying. To get near the 50mp resolution, you would now need supreme lenses to see a major difference. 36--->50mp would be a small jump with most lenses(negligible)and like you mentioned, huge if with an Otus.

Back to my original statement, forget about getting the most out of those 50mp with the current selection of native lenses in the A7 mount. Added on top with the small pixel size, extra camera shake.

 

[jrista]

The page supports it. You will never get 50mp or even close to it with most lenses. The jump from 36mp to 50mp will be slight with most lenses and stark with supreme lenses. Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass. That's why cropping FF 22mp vs aps 18mp the results are nearly identical. On paper they shouldn't be, but on the other thread, negligible.

The point I was trying to make before was that you can't get 50mp with any existing lens, and probably wouldn't with any lens created within the next decade. The same goes for 36mp, 24mp, 18mp. You cannot actually resolve those resolutions with ANY lens, even the best of the best of the best. Because output resolution has an asymptotic relationship with the least resolving component of the system. To even get close to 50mp, you would probably need to be shooting a lens like the Otus at f/1.4 (assuming it's diffraction limited at that aperture...if not, then you would need a lens even better than the Otus).

Assuming you don't have a crappy lens, then you can realize improvements by moving to a higher resolution sensor. Every time you do, the nyquist limit drops. That allows more information to be resolved usefully. It might be resolved at lower contrast, but until your down near the Rayleigh limit, you can still do stuff with lower contrast detail (it's more work to enhance it, but it can be done.) If you jump from 18mp to 50mp, even with just an "ok" lens, your going to see a huge difference. The frequency of detail that might have been resolved crisply at 18mp will now probably look a little soft...however, your going to be resolving a level of detail the 18mp couldn't even see at all. The same would go for the difference between 36mp and 50mp...although the amount of smaller details you resolve wouldn't be as significant, and would require more work to enhance.

For any given lens, going from 24mp to 36mp, then to 50mp, WILL realize an improvement. The improvement might start out at around 18% for the jump from 24-36mp, then drop to 11-12% for the jump from 36-50mp. It might be a mere 5% in a jump from 50mp to 70mp. Beyond that, your probably within the margin of error...however, then your in the realm of oversampling. That has a whole 'nother set of benefits.

Your absolutely correct and that's what I've been saying. To get near the 50mp resolution, you would now need supreme lenses to see a major difference. 36--->50mp would be a small jump with most lenses(negligible)and like you mentioned, huge if with an Otus.

Back to my original statement, forget about getting the most out of those 50mp with the current selection of native lenses in the A7 mount. Added on top with the small pixel size, extra camera shake.

Ah, ok. I guess I misunderstood something along the line, when you said you believed DXO when they stated the Zeiss 135 could actually get 36mp out of the D800. Good to know we agree.

I don't know much about native lenses for the A7 mount...however, I would really be interested in seeing how some of Canon's newer lenses work with that sensor when adapted to the A7 mount. I bet the 24-70/2.8 II is phenomenal.

 

[RLPhoto]

The page supports it. You will never get 50mp or even close to it with most lenses. The jump from 36mp to 50mp will be slight with most lenses and stark with supreme lenses. Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass. That's why cropping FF 22mp vs aps 18mp the results are nearly identical. On paper they shouldn't be, but on the other thread, negligible.

The point I was trying to make before was that you can't get 50mp with any existing lens, and probably wouldn't with any lens created within the next decade. The same goes for 36mp, 24mp, 18mp. You cannot actually resolve those resolutions with ANY lens, even the best of the best of the best. Because output resolution has an asymptotic relationship with the least resolving component of the system. To even get close to 50mp, you would probably need to be shooting a lens like the Otus at f/1.4 (assuming it's diffraction limited at that aperture...if not, then you would need a lens even better than the Otus).

Assuming you don't have a crappy lens, then you can realize improvements by moving to a higher resolution sensor. Every time you do, the nyquist limit drops. That allows more information to be resolved usefully. It might be resolved at lower contrast, but until your down near the Rayleigh limit, you can still do stuff with lower contrast detail (it's more work to enhance it, but it can be done.) If you jump from 18mp to 50mp, even with just an "ok" lens, your going to see a huge difference. The frequency of detail that might have been resolved crisply at 18mp will now probably look a little soft...however, your going to be resolving a level of detail the 18mp couldn't even see at all. The same would go for the difference between 36mp and 50mp...although the amount of smaller details you resolve wouldn't be as significant, and would require more work to enhance.

For any given lens, going from 24mp to 36mp, then to 50mp, WILL realize an improvement. The improvement might start out at around 18% for the jump from 24-36mp, then drop to 11-12% for the jump from 36-50mp. It might be a mere 5% in a jump from 50mp to 70mp. Beyond that, your probably within the margin of error...however, then your in the realm of oversampling. That has a whole 'nother set of benefits.

Your absolutely correct and that's what I've been saying. To get near the 50mp resolution, you would now need supreme lenses to see a major difference. 36--->50mp would be a small jump with most lenses(negligible)and like you mentioned, huge if with an Otus.

Back to my original statement, forget about getting the most out of those 50mp with the current selection of native lenses in the A7 mount. Added on top with the small pixel size, extra camera shake.

Ah, ok. I guess I misunderstood something along the line, when you said you believed DXO when they stated the Zeiss 135 could actually get 36mp out of the D800. Good to know we agree.

I don't know much about native lenses for the A7 mount...however, I would really be interested in seeing how some of Canon's newer lenses work with that sensor when adapted to the A7 mount. I bet the 24-70/2.8 II is phenomenal.

I said I trusted DXO lens scores to a degree and other reviewers reflected my experience with the lenses tested. IE: Abbotts 135mm APO tests.

 

[dtaylor]

The page supports it.

: The page listed 3 lenses recommended for the D800. That is all. It said nothing to support what you are claiming. Unless you have another link this point is done.

You will never get 50mp or even close to it with most lenses.

Stop using terminology incorrectly. You will get 50 MP with the lens cap on.

Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass.

Not what anyone else reports with APS-C

 

[RLPhoto]

The page supports it.

: The page listed 3 lenses recommended for the D800. That is all. It said nothing to support what you are claiming. Unless you have another link this point is done.

You will never get 50mp or even close to it with most lenses.

Stop using terminology incorrectly. You will get 50 MP with the lens cap on.

Even with aps cams, the difference is the same, indistinguishable with most glass and stark with supreme glass.

Not what anyone else reports with APS-C

Great because we all want 50mp of detail in our lens cap shots. : Please review the info again because Jrista explained what DXO is referring about actual resolution received.

 

[ULFULFSEN]

I want a 50MP camera and the 55mm and 85mm Zeiss Otus as kit lenses.

And why not 50MP?

Some peeps who are serious about big prints will sure love a small camera with MF quality. They will pay the price for the best glass. The more options the better.

 

[psolberg]

I want a 50MP camera and the 55mm and 85mm Zeiss Otus as kit lenses.

And why not 50MP?

Some peeps who are serious about big prints will sure love a small camera with MF quality. They will pay the price for the best glass. The more options the better.

+1. More options the better. As you say. I really can't understand the hate crowd. It's like they want to live in a world where everybody shoots the same boring 20MP with the same boring options. They probably want everybody to drive a gray toyota prius

I say bring it on 50-100MP. Surely we could have cameras which do the down sample internally and surely adobe will make it standard to down sample your images prior to developing it at your target resolution and space requirements. I know I will always prefer the latitude of editing over the compromise. Otherwise I'd be shooting 6MP because it was "good enough". Well, "good enough" is ok for "good enough" goals. I want more

 

[psolberg]

Sorry, that's totally false.

Go try it yourself. There are A7 / A7R test scene samples at multiple sites.

Furthermore, if you need to upsample, starting with more data means the upsampling process is more accurate, as it has to fabricate LESS artificial data than s 24mp or 20mp or 18mp file.

That's true in so far as there's more real data (lpmm at a given MTF) to begin with.

Hang out on a print forum sometime. For all the bickering we do over pixels, they do 10 times more over paper quality, ink quality, dMax, L*, metamerism, bronzing, ink emulsions, etc. It matters to the people at the lab who do all the printing for...millions of people who order prints from labs.

Any of the current Epson Professional series printers/ink sets on a paper like Hot Press Bright...you're looking at the very top end of what can be laid down on paper today. I won't order laser photo paper prints (Frontier or Noritsu) when I can print or order Epson.

You yourself know it isn't just about looking at two images strait out of camera. The editing latitude differences matter as well, the response to things like NR or white balance or exposure changes that better data with richer information that gives us the ability to reduce noise more effectively with less effort, that all matters. Maybe not to everyone, maybe not even the majority (at the moment), but to a lot of people...to an increasing number of people, IMO.

There's not a significant difference between an A7 and an A7R in these respects.

Again, to separate the print issues from actual RAW fidelity. You call it pixel peeping, I call it latitude of post processing and oversampling. Per your point, remember the A7R shoots really compromised RAWS. It is not the same data from the Nikon's more refined pipeline. I wouldn't really shoot with an A7R personally if I had a D810/800 because sony's RAW file choices crippled the A7R.
http://diglloyd.com/blog/2014/20140214_1-SonyA7-artifacts-star-trails.html
http://diglloyd.com/blog/2014/20140212_2-SonyA7-RawDigger-posterization.html

you can find more on the subject online off course.

So while the A7R is a great tool indeed, let's just say it doesn't speak for the state of the art 36+MP sensor class.

I've shot with 20MP canon bodies, then moved to 36MP Nikon bodies, and would certainly consider 50-100MP bodies from whatever company can deliver as much in the RAW file as they can (not the A7R way). Surely sony learned and this will carry forward. The point being not the brand which is a secondary aspect. The point being, other than storage, super sampling and higher resolution is always going to trump lower resolutions regardless of media output for my choices. This is why I'll always prefer 36MP to 22 because there is just no benefit to me in the 22MP file, and far too many drawbacks. This is why I'd prefer 50 to 36MP for the same reason and beyond. Now if I covered sports and say, ended up with 2000 images per hour, that equation would change. And off course these high detail cameras aren't aimed at everybody. I recognize that.

Ultimately as I've said it, if you don't value high MP output you're in two camps:
1) you don't need the latitude or would benefit from over sampling. Other aspects rule more.
2) you could use the gains, but are just holding on to what you have because you want/have to.

#1 will always be the case. #2, IMHO, it is inevitable. I don't see canon/sony/nikon ever making another 20s MP full frame again save for the people in camp #1, which I admit is a market. Their landscape/studio game will be 30 or 40 minimum to be competitive. At the density of even APSC, we'd be in the high 40s to 50s territory by now. In many ways, it is not just inevitable, but overdue.

Many of the fears is storage and processing. However the emergence of high resolution photography will provide the tools. As 36-50-100MP become "standard" surely adobe and others will incorporate oversample in their workflow. As indicated by myself and others, color accuracy, aliasing, bayer interpolation artifacts, all are directly tied to resolution with improvements on the MP count making a positive impact. Therefore it stands to reason PS and LR could very much offer you the option to take the 50MP raw, super sample it down to your workflow target resolution, yielding superior results to low resolution capture. You can then choose to discard, or archive the original RAW and work on the derived file.

This super sample step would go BEYOND mere resizing post raw to RGB conversion. RAW->RGB to low MP is what people currently do to demonstrate benefits of 36 vs 22 and surely will of 50 vs 36 but this is not the best way. Ideally the raw processor would create your say 36MP file from the 50-100MP RAW without having an intermediate RGB image that is then downsized using the various interpolation methods. You'll always work and in your case print from the target resolution of your choice, but you just have a better version of it than if you didn't. Again, all inevitable.

Ultimately I'm not trying to convince people to give up their 20+MP gear. I moved out of that, and I know many will once canon catches up and then, surely everybody will build a shrine around some gear and sing praises to all the things I've mentioned. What I care about the trend, not any specific model or brand. There is a lot more than shooting B+W geometric patterns and whatever DXO computers spit out as a number. So when people get worked out about what their 24-70 can do with a pattern in a high vs low sensor resolution, they are essentially missing the forest for the trees. Some will realize this, some won't. And that's ok. everybody should do what works for them. ;D

 

[ChristopherMarkPerez]

Back in the day when I was trying to sort this out for myself, I had a long conversation with a physics professor. I was just an engineer, after-all, and I wanted to make sure I fully understood the underlying physics, both practical and theoretic. He too had been looking at the question for years.

What it comes down to is this: While the 1/R formula that is commonly found (one example is found on Fuji's website, of all places) that attempts to describe the relationship between light capture media and optics is fundamentally flawed and is easily proven wrong through simple testing.

The problem is that the formula that is the basis for nearly every single one of these kinds of conversation/discussions/flame-wars is wrong. The issue is in the mis-application of particle physics as applied to light. The formula is simple, and it's "feels" right, but it's not. Light (in terms of resolution) acts more as a wave-front than that of a particle. A much more complex equation would be needed to properly explain the relationship between optics and sensors.

To simplify, the professor said that from his perspective, corrected math fully supports his claim that at normal working apertures (f-wide-open through f/11) and commercially available components (ie: anything coming from current camera manufacturers), film or digital sensors are the limiting factor and set the limits of image resolution.

Said simply, you can use any correctly manufactured commonly available optic on any sensor currently made and that lens will have more than adequate performance for your system. Which leaves us with the straight-forward exercise of calculating real world resolution by looking only at the sensor.

If any of what I just said was wrong you'd not be able to build the kinds of computer parts that enable this very discussion. Talk to any mask builder at Intel or AMD and you'll fully understand what I mean by this.

The point I was trying to make before was that you can't get 50mp with any existing lens, and probably wouldn't with any lens created within the next decade. The same goes for 36mp, 24mp, 18mp. You cannot actually resolve those resolutions with ANY lens, even the best of the best of the best. Because output resolution has an asymptotic relationship with the least resolving component of the system...

 

[moreorless]

I want a 50MP camera and the 55mm and 85mm Zeiss Otus as kit lenses.

And why not 50MP?

Some peeps who are serious about big prints will sure love a small camera with MF quality. They will pay the price for the best glass. The more options the better.

This is I'd say likely to be the problem with Sony pushing megapixels with the FE system, the lenses needed to get the most out of such sensors(especially I'd guess wideangles) will need to be large.

The A7s seems to make more sense to be as its much less resolution hungry.

 

[jrista]

Back in the day when I was trying to sort this out for myself, I had a long conversation with a physics professor. I was just an engineer, after-all, and I wanted to make sure I fully understood the underlying physics, both practical and theoretic. He too had been looking at the question for years.

What it comes down to is this: While the 1/R formula that is commonly found (one example is found on Fuji's website, of all places) that attempts to describe the relationship between light capture media and optics is fundamentally flawed and is easily proven wrong through simple testing.

The problem is that the formula that is the basis for nearly every single one of these kinds of conversation/discussions/flame-wars is wrong. The issue is in the mis-application of particle physics as applied to light. The formula is simple, and it's "feels" right, but it's not. Light (in terms of resolution) acts more as a wave-front than that of a particle. A much more complex equation would be needed to properly explain the relationship between optics and sensors.

To simplify, the professor said that from his perspective, corrected math fully supports his claim that at normal working apertures (f-wide-open through f/11) and commercially available components (ie: anything coming from current camera manufacturers), film or digital sensors are the limiting factor and set the limits of image resolution.

Said simply, you can use any correctly manufactured commonly available optic on any sensor currently made and that lens will have more than adequate performance for your system. Which leaves us with the straight-forward exercise of calculating real world resolution by looking only at the sensor.

If any of what I just said was wrong you'd not be able to build the kinds of computer parts that enable this very discussion. Talk to any mask builder at Intel or AMD and you'll fully understand what I mean by this.

The point I was trying to make before was that you can't get 50mp with any existing lens, and probably wouldn't with any lens created within the next decade. The same goes for 36mp, 24mp, 18mp. You cannot actually resolve those resolutions with ANY lens, even the best of the best of the best. Because output resolution has an asymptotic relationship with the least resolving component of the system...

Your professor is right, to a degree...however, I'd be curious to know exactly what he said, as he is also incomplete. First, he is correct that light behaves as a wavefront. The particle-nature of light is useful for describing in geometric terms how optics behave...but in reality, we don't work with individual particles of light. We work with a continuous, complex wavefront of light that produces a three dimensional structure, a "light cone" for lack of a better term, within the lens that ultimately resolves, or focuses, at the sensor plane. Another critical point about a photonic wavefront is that diffraction is an inherent trait, not something like "light bending around an obstacle." If you want to learn more, you can read about it here:

http://www.telescope-optics.net/wave.htm

As for the resolving power of a camera. I've said it a thousand times on these forums already: Lenses don't outresolve sensors, and neither do sensors outresolve lenses. The two work together to convolve the output, which will have lower resolution than both, with an asymptotic relationship with the least-resolving component of the system.

To approximate how two lenses will perform, you can use the following formula:

sysRes = (1/SQRT((1/(lensRes*2))^2 + sensorPixelPitch^2))/2

Fundamentally, this formula is calculating the system spot size, the convolved result of a single point of light by both the lens and sensor. The rest is simply to convert from spatial resolution in lp/mm and back. So, assuming we have an f/11 lens that means the diffraction-limited performance at MTF50 is 63lp/mm. That is a pretty low resolution...there are actually lots of sensors that have higher spatial resolution than that. However, let's say we take a 22.3mp, 36.3mp, and 50mp sensor. Assume we use all of them with that diffraction-limited lens at f/11. Here are the results of measuring the MTF50 spatial resolution of the resulting images:

22.3mp: 49.45lp/mm
36.3mp: 53.60lp/mm
50mp: 55.97lp/mm

As you can see, even at f/11, as you increase megapixel count (reduce pixel size), the ability of a sensor to resolve detail with a fully diffraction limited lens producing a large spot size is still possible. The differences are not large...probably not visible, and probably only measurable with software.

Now, f/11 is a very limited aperture. It resolves 63lp/mm. If we jump up to f/4, the diffraction-limited resolution jumps to 173lp/mm. Then, we get the following results:

22.3mp: 72.6lp/mm
36.3mp: 88lp/mm
50mp: 99.6lp/mm

If we think about something like the Otus, which is diffraction-limited at even wider apertures, here is what we could resolve at f/2 where we have 346lp/mm spatial resolution:

22.3mp: 78lp/mm
36.3mp: 98lp/mm
50mp: 115lp/mm

As you can see, even though sensors are not resolving as much, in terms of spatial resolution, as lenses, higher and higher resolution sensors are still capable of producing higher resolution results. It doesn't matter if your at a heavily diffraction limited f/11, or at a minimally diffraction limited f/2...you will still resolve more detail with a higher resolution sensor. The sensors will ultimately set the limit...however we are currently far from reaching those limits with high quality lenses, such as the Otus (best example, there are other lenses out there that resolve near the diffraction limit wide open, such as most of Canon's great white supertelephoto lenses.) You can also approach the true diffraction limit of the lens by increasing sensor resolution. You want to actually resolve 63lp/mm from an f/11 lens? Well, you'll need a sensor capable of resolving a few hundred lp/mm to actually be able to get close...which means we'll need sensors capable of resolving hundreds of megapixels.

The diffraction-limited resolution of an f/4 lens is 173lp/mm. A 50mp sensor combined with such an f/4 lens is only resolving about 100lp/mm. That means were barely more than half way to the diffraction limit of the lens. We could still resolve more...with a 75mp sensor, a 100mp sensor, a 150mp sensor.

Regarding the kinds of UV lithography we use for etching sensor parts at 22nm. You have to remember that the wavelengths of light we use to do that are extremely short. Visible light wavelengths stretch from about 380nm near-UV to 850nm near-IR. Deep UV is less than 200nm in wavelength down to around 100nm in wavelength, and EUV is a mere 13.5nm in wavelength. The photomasks used in photolithography are also quite large, and the systems that are used to actually etch silicon with UV light use the best optics on the planet (usually Zeiss) and are perfectly diffraction limited (and at such small wavelengths, a diffraction spot is very small). Large masks result in little diffraction, and ultra small wavelengths smaller than the smallest transistor ensure we can actually create structures that small. We even have techniques that allow us to perform sub-wavelength etching, which is why we will ultimately be able to etch around 7nm and maybe even 4nm transistor sizes with a 13.5nm wavelength.