Resolution

[GerryS]

I'm confused re resolution, can somebody explain to me why a 12mp FX will give a better image when printed at 40" than a 24mp dx will? If indeed it will?

 

[Phil V]

I'm confused re resolution, can somebody explain to me why a 12mp FX will give a better image when printed at 40" than a 24mp dx will? If indeed it will?

It might - it might not.
Resolution is a complex subject, I can guarantee that if this thread runs for a week, there'll be science discussed that 99% of photographers will not understand.

For a preview - go and read the 'white balance quandary' thread, a simple question, quite a simple answer - and then the 'knowledgeable' try to be 'helpful'

Your answer in short though is that the larger photosites will gather more photons, therefore giving a more accurate rendition of the image focussed on them. Of course it also includes equality of technology, and the fact that a larger sensor can make the most of a lens mtf performance, and there'll be a smattering of signal to noise ratio, and probably many other things I never thought about.

 

[GerryS]

Thanks Phil, that all makes sense. To be a bit more specific, my contention is that a 2015 24mp D7200 will give equal if not better quality 40" prints than a 2008 released 12mp D700.

I'm not set in stone on my opinion and am prepared to be proved wrong but I can't see how a 7 year old lower res design would better a brand new higher res design?

 

[StephenM]

I think I'm about to muddy the waters...

The thread title is "Resolution" but the query seems to be aimed at print quality. As there is more to print quality than simply resolution, what's the exact question?

 

[simonblue]

I stopped worrying about resolution at the moment, i have Fuji 16mp which i find is very good for the photos i do,plus i am to committed to the Fuji system to change,(the financial lost) to even think about an upgrade to an greater resolution,by swapping systems

 

[Phil V]

Thanks Phil, that all makes sense. To be a bit more specific, my contention is that a 2015 24mp D7200 will give equal if not better quality 40" prints than a 2008 released 12mp D700.

I'm not set in stone on my opinion and am prepared to be proved wrong but I can't see how a 7 year old lower res design would better a brand new higher res design?

You're probably wrong, I appreciate that Sony sensors are progressing faster (I know more about Canon) but a 7dII can't hold a candle to a 5dII (similar age differences), but the 5dII does have a higher pixel count too.

But you're better waiting for a Nikon expert to show up, I reckon @Pookeyhead or @HoppyUK will probably have examples, but again, you'll have to compare apples with apples, are you talking ultimate resolution from a low ISO image with a very good lens? Or high ISO under crap light, or both tested with a mediocre lens (where the lens resolution would probably be the defining factor.

 

[chris malcolm]

The answer to the image quality question could depend on the extent to which the shooting conditions favoured the full frame. If the shot was taken at low ISO with lenses of equal quality and did not need extreme post processing I'd expect the 24MP to be better.

 

[GerryS]

I think I'm about to muddy the waters...

The thread title is "Resolution" but the query seems to be aimed at print quality. As there is more to print quality than simply resolution, what's the exact question?

Ostensibly it was resolution, but with a bias towards print quality. Lets run with print quality as it's possibly (???) easier to answer.

Lets assume good light and conditions, lowest native ISO (100 for D7200, 200 for D700) with a good quality lens, all other things being equal.

 

[HoppyUK]

What I think you're mostly referring to is 'sharpness' which has two components - resolution (the fineness of detail) and contrast (how clearly those details are shown). It is not just about pixel count (resolution) and actually it is contrast that conveys most visual impression of sharpness. Resolution vs contrast are the two axes of a lens MTF graph (modulation transfer function) which are those coloured squiggly graphs that lens manufacturers publish.

Fact is, that when lens resolution goes up, image contrast always goes down - because the lens has to work harder*. Smaller formats require greater magnification for a given size output, so they demand more resolution from the lens, and correspondingly image contrast is lower. Format size is key, bigger is better, and pixel count really only enters into it at very big output sizes. That's why FF/FX is sharper than APS-C/DX, which is sharper than M4/3, which is sharper than 1-inch cameras etc etc, and throwing extra pixels at it won't change that.

*It's like a car that will accelerate from 0-60 in six seconds, but takes much longer to get from 60-120.

 

[Pookeyhead]

I'm confused re resolution, can somebody explain to me why a 12mp FX will give a better image when printed at 40" than a 24mp dx will? If indeed it will?

This is an awkward one, but I'd say the D7200 will yield a file more able to print 40" across than a D700, simply because the D700 is limited by it's lower resolution. However... the D700 can yield prints bigger than 20" easily, despite being half the resolution.
Take a full frame camera with slightly higher resolution however, such as the D3 at 16MP, and the gap closes more than you would imagine. Sensor size is a factor here. Bigger sensors yield higher quality images simply because the lens doesn't have to work so hard. If both sensors were 16MP, the full frame one would be demonstrably better.


Explained simply, this is how it works.

• All lenses have a limit to the size of object or detail that they can resolve. This is known as the circle of confusion. For most digital SLR lenses let's assume for the same of this argument that it's around 0.03mm.

Assuming you use the same lens on a APS-C crop sensor camera like a Nikon D7000 and a full frame camera like the D800, it will still resolve detail down to a minimum of 0.03mm regardless of which camera it is on as it's a fixed property of the lens.

• However, 0.03mm is smaller in proportion to the area of a full frame sensor than it is a crop sensor.

• 0.03mm is 0.12% of the total image width of a APS-C sensor (23.5mm x 16mm)

• 0.03mm is 0.08% of the total image width of a full frame sensor (36mm x 24mm)

In other words, the blurriness caused by lens defects is 33% less (or smaller) on a full frame camera compared to a crop sensor camera, regardless of it's resolution.

Again assuming we could use exactly the same lens on a 5 x 4 inch camera (I know you can't befoe anyone points this out), the percentage of image width taken up by the circle of confusion would be 0.02%

This is a 84% decrease in circle of confusion size compared to the whole image area compared to a APS-C crop sensor.


Apparent sharpness can therefore be said to be a product of sensor size.

How visible the aliasing (pixels) will be in print can therefore be said to be a product of the image resolution.

A combination of large sensor and high resolution is best, but a 16MP image on a small sensor camera will be visibly less sharp than a 16MP image from a larger sensor camera despite the aliasing being identical.

Identical sized prints from files for comparison

Sharpness at single pixel level.

16MP D800 images are not taken in DX crop mode, but are resized FX images. All I've done is made the pixels bigger. A 16MP FX sensor would look pretty much the same as the resized D800 image.

The resolution is identical, yet sharpness is greater from the FX image. Fact.

This is why the push for greater and greater resolution from sensors is pointless now. With the D800 we've hit a limit set by the lenses (for 35mm it's actually around 24MP)... not the sensor. If the D4X has greater than 36MP when it arrives (if it arrives) it will be utterly stupid and Nikon just pandering to people like you who feel greater pixels mean better images. If you want more sharpness now, you either optimise lens design more (as MFT has done), or move up to medium format digital. There's no more to be done. More pixels have b****r all to do with it.


However, because the D700 is so much lower in resolution than the D7200, then aliasing comes into play, and the much more visible pixel structure of the lower resolution D700 becomes a factor.

I'd say, despite the D700 being full frame, the lower 12MP resolution would make pixel aliasing very visible at 40", whereas the D7200 would hole up well. At smaller print sizes the D700 would win, but at 40" the visible pixelation would be an issue at close viewing distances. I think the 12MP of the D700 pushes this concept of sensor size a little too far.

 

[ChrisR]

At normal viewing distances (rather than "close viewing distances") is there likely to be any visible difference? I remember @DuncanDisorderly including a print from his X10 (12mp) alongside 5D3 prints in one of his RPS panels (probably smaller than 40" TBF) and said it was very hard to tell which was from the smaller camera.

 

[GerryS]

This is an awkward one, but I'd say the D7200 will yield a file more able to print 40" across than a D700, simply because the D700 is limited by it's lower resolution. However... the D700 can yield prints bigger than 20" easily, despite being half the resolution.
Take a full frame camera with slightly higher resolution however, such as the D3 at 16MP, and the gap closes more than you would imagine. Sensor size is a factor here. Bigger sensors yield higher quality images simply because the lens doesn't have to work so hard. If both sensors were 16MP, the full frame one would be demonstrably better.


Explained simply, this is how it works.

• All lenses have a limit to the size of object or detail that they can resolve. This is known as the circle of confusion. For most digital SLR lenses let's assume for the same of this argument that it's around 0.03mm.

Assuming you use the same lens on a APS-C crop sensor camera like a Nikon D7000 and a full frame camera like the D800, it will still resolve detail down to a minimum of 0.03mm regardless of which camera it is on as it's a fixed property of the lens.

• However, 0.03mm is smaller in proportion to the area of a full frame sensor than it is a crop sensor.

• 0.03mm is 0.12% of the total image width of a APS-C sensor (23.5mm x 16mm)

• 0.03mm is 0.08% of the total image width of a full frame sensor (36mm x 24mm)

In other words, the blurriness caused by lens defects is 33% less (or smaller) on a full frame camera compared to a crop sensor camera, regardless of it's resolution.

Again assuming we could use exactly the same lens on a 5 x 4 inch camera (I know you can't befoe anyone points this out), the percentage of image width taken up by the circle of confusion would be 0.02%

This is a 84% decrease in circle of confusion size compared to the whole image area compared to a APS-C crop sensor.


Apparent sharpness can therefore be said to be a product of sensor size.

How visible the aliasing (pixels) will be in print can therefore be said to be a product of the image resolution.

A combination of large sensor and high resolution is best, but a 16MP image on a small sensor camera will be visibly less sharp than a 16MP image from a larger sensor camera despite the aliasing being identical.

Identical sized prints from files for comparison

Sharpness at single pixel level.

16MP D800 images are not taken in DX crop mode, but are resized FX images. All I've done is made the pixels bigger. A 16MP FX sensor would look pretty much the same as the resized D800 image.

The resolution is identical, yet sharpness is greater from the FX image. Fact.

This is why the push for greater and greater resolution from sensors is pointless now. With the D800 we've hit a limit set by the lenses (for 35mm it's actually around 24MP)... not the sensor. If the D4X has greater than 36MP when it arrives (if it arrives) it will be utterly stupid and Nikon just pandering to people like you who feel greater pixels mean better images. If you want more sharpness now, you either optimise lens design more (as MFT has done), or move up to medium format digital. There's no more to be done. More pixels have b****r all to do with it.


However, because the D700 is so much lower in resolution than the D7200, then aliasing comes into play, and the much more visible pixel structure of the lower resolution D700 becomes a factor.

I'd say, despite the D700 being full frame, the lower 12MP resolution would make pixel aliasing very visible at 40", whereas the D7200 would hole up well. At smaller print sizes the D700 would win, but at 40" the visible pixelation would be an issue at close viewing distances. I think the 12MP of the D700 pushes this concept of sensor size a little too far.

@Pookeyhead Thanks so much for taking the time to explain that David, it makes perfect sense and was very easy to follow the process through. It's also helped me some considerable way with my D750 vs D810 dilemma. It was a very interesting read and I'm very grateful, thank you.

 

[HoppyUK]

<snip>

This is why the push for greater and greater resolution from sensors is pointless now. With the D800 we've hit a limit set by the lenses (for 35mm it's actually around 24MP)... not the sensor. If the D4X has greater than 36MP when it arrives (if it arrives) it will be utterly stupid and Nikon just pandering to people like you who feel greater pixels mean better images. If you want more sharpness now, you either optimise lens design more (as MFT has done), or move up to medium format digital. There's no more to be done. More pixels have b****r all to do with it.

<snip>

Excellent post David

But you keep saying that modern cameras are now lens limited and it's simply not true. You can't specifiy any kind of lens limit with resolution/pixels alone, without also stating the % MTF contrast at that level. Even quite modest lenses can resolve very fine detail, way above your 24mp ceiling (on full-frame, which equates to only 10mp on APS-C), just at a lower contrast level - but you can still see, you just have to look more carefully. Perceived sharpness reduces gradually, so at what point does it stop? It's a meaningless statement without that, and depending on the user and application, very different standards apply.

At one end, there are the very modest demands of on-screen and on-line viewing, and at the other there are keen landscapers striving for best quality 40in prints. They will easily see the difference between 24mp and 36mp and even 50mp. I have made all those comparisons, and the difference is not hard to see (with the same lens). The best new lenses now appearing are also a step above anything I've seen before, well capable serving a 50mp sensor, and more. Another thing, if you put the same lens on a higher resolution sensor, you will get increased sharpness because of the way MTFs cascade.

If you want to set limits, then you must specify the parameters, and also be aware that different standards apply to different uses.

 

[Chrisphoter]

At one end, there are the very modest demands of on-screen and on-line viewing

Very modest in the days when 72 dpi was the standard for computer monitors, not so very modest these days with Retina displays and 5k monitors?

 

[sk66]

Too many variables IMO...
How the lens is used makes a difference (aperture). If a fast sharp lens is used wide then you can get much more resolution than even the D7200 demands (in excess of 100MP is theoretically possible).
The max IQ/MTF of the lens makes a difference...many fast primes may be able to deliver 50+MP on APS, but many other lenses may not be able to deliver much more than 13MP (f/8 APS) regardless of the MP's on the sensor.
If the light is poor/ISO higher, then the smaller sensor(smaller pixels) is at a disadvantage in terms of energy collected per pixel. It is also at a disadvantage in terms of total energy collected just because it is smaller.
Larger sensors have an intrinsic advantage... they are larger and therefore they need less enlargement for a given size.
And then there's the printer's resolution and ink bleed (if a factor).

In optimal usage/conditions (and with the right lens) the D7200 can deliver larger/sharper images. But, those situations are relatively few IMO. In more typical conditions they will tend to equalize. And once they equalize the larger sensor will retain it's "larger" advantage from then on.
My personal belief is that very few rarely actually see the benefit/potential of a 24MP APS or a 36-50MP FF sensor (myself included)... But that doesn't really mean you are any worse off than using a sensor of lower resolution.

A 40" wide print is pretty demanding... For critical high quality you would need a system resolution capability of ~4000+ lp/ph (lens/camera/printer) and there aren't a lot of combinations (or lenses) that can do that. For very good quality (average viewing) you would need something closer to ~3000 lp/ph which is more likely.

 

[Pookeyhead]

Excellent post David

But you keep saying that modern cameras are now lens limited and it's simply not true. You can't specifiy any kind of lens limit with resolution/pixels alone,

When I see softness caused by lens limitations before I see aliasing, then it's lens limited. The end. I apply the same logic to film scanning. If I can see aliasing before I can clearly see grain, then I need to scan at a higher res. If I see grain clearly before I see aliasing, then I'm OK.


As usual.. a purely academic study of lens resolution figures measured on a bench is one thing, but out out in the field, just eyeballing it is usually enough.

 

[chris malcolm]

When I see softness caused by lens limitations before I see aliasing, then it's lens limited. The end. I apply the same logic to film scanning. If I can see aliasing before I can clearly see grain, then I need to scan at a higher res. If I see grain clearly before I see aliasing, then I'm OK.


As usual.. a purely academic study of lens resolution figures measured on a bench is one thing, but out out in the field, just eyeballing it is usually enough.

I don't like the Procrustean term "lens limited". My worst lens in general use is an 18-250mm. On my previous 14MP APS-C camera it was clearly a touch softer even at its f8 sharpest than my better lenses (which are sharper at larger apertures). I've never seen aliasing on any of my lenses, unless you mean the staircasing of diagonal lines, which I can see on all my lenses, just softer on the worst. Maybe my Sony cameras have good AA filters. Maybe I need to look more carefully at photographs of fabric and distant railings.

Going by the wisdom of the web I expected to find as I upgraded to 24MP that this 18-250mm lens would be a casualty, unable to do any better with 24MP then 14MP. The sensor would have "out-resolved" the lens, if indeed it wasn't already "out-resolved" at 14MP. What I found to my surprise that on careful scrutiny of carefully selected areas of some images there was a little extra real detail visible on 24MP compared to 14MP, just less and harder to find than with better lenses.

That's why I don't like the term "lens-limited". The detail resolution capabilities of a lens and sensor combination multiply together rather like failure probabilities do. There's no hard limits. There's graded fall offs. Tests of maximum detail resolution are just different ways of placing a stake in the slope to establish a normalised nominal boundary point for lens test comparison purposes. Different tests will find different boundaries.

Generally speaking if I'm shooting jpegs with the 18-250mm and making big prints directly from the the ex-cameras jpegs there's no point in shooting jpegs larger than the half-size 12MP jpeg option of my 24MP camera. Nobody but me would be able to find the few very slight diffetrednces. But I don't print unprocessed jpegs. I always tweak them a bit, and often correct mild misalignments of verticals. Working at 24MP gives better results when processing images, even with such an "out-resolved" lens.

 

[HoppyUK]

I don't like the Procrustean term "lens limited". My worst lens in general use is an 18-250mm. On my previous 14MP APS-C camera it was clearly a touch softer even at its f8 sharpest than my better lenses (which are sharper at larger apertures). I've never seen aliasing on any of my lenses, unless you mean the staircasing of diagonal lines, which I can see on all my lenses, just softer on the worst. Maybe my Sony cameras have good AA filters. Maybe I need to look more carefully at photographs of fabric and distant railings.

Going by the wisdom of the web I expected to find as I upgraded to 24MP that this 18-250mm lens would be a casualty, unable to do any better with 24MP then 14MP. The sensor would have "out-resolved" the lens, if indeed it wasn't already "out-resolved" at 14MP. What I found to my surprise that on careful scrutiny of carefully selected areas of some images there was a little extra real detail visible on 24MP compared to 14MP, just less and harder to find than with better lenses.

That's why I don't like the term "lens-limited". The detail resolution capabilities of a lens and sensor combination multiply together rather like failure probabilities do. There's no hard limits. There's graded fall offs. Tests of maximum detail resolution are just different ways of placing a stake in the slope to establish a normalised nominal boundary point for lens test comparison purposes. Different tests will find different boundaries.

Generally speaking if I'm shooting jpegs with the 18-250mm and making big prints directly from the the ex-cameras jpegs there's no point in shooting jpegs larger than the half-size 12MP jpeg option of my 24MP camera. Nobody but me would be able to find the few very slight diffetrednces. But I don't print unprocessed jpegs. I always tweak them a bit, and often correct mild misalignments of verticals. Working at 24MP gives better results when processing images, even with such an "out-resolved" lens.

Yes. I don't like the term 'lens limited' either - because it's simply untrue and unhelpful. Sharpness is a coin of two halves, both lens and sensor, and if you put a higher resolution sensor behind any lens, you will see detail improvements. And if you put a sharper lens on a lower-res sensor, you will also get significant benefits at all sizes of output long before pixilation becomes an issue because contrast is higher. The third element is sensor size, and this is one thing where David and I agree - larger formats are sharper regardless of resolution because they allow the lens to perform at a much higher contrast level.

David is also implying that the lens tests I do for magazines and websites are 'academic... bench tests'. Well, they're not. Some studio tests for sure but I don't have an optical bench and they're as real world as I can make them. With the reviews I do for one big US website, most of the work is field testing and includes a couple of thousand images where you can see the differences. I will also say that maxing out a 50mp sensor is not easy and in practise even the slightest focusing error or camera-shake will lose the finest detail but that's a technique issue, not camera or lens related or 'limited'.

 

[Pookeyhead]

Yes. I don't like the term 'lens limited' either - because it's simply untrue and unhelpful. Sharpness is a coin of two halves, both lens and sensor, and if you put a higher resolution sensor behind any lens, you will see detail improvements.

I have to disagree with this. It implies there's not a finite amount of resolution from any given lens, and you can just keep on upping the resolution of the sensor, and keep making gains, when it's patently obvious that's not the case. You will quickly reach a limit where the lens has nothing more to give.. hence the limitation is the lens, not the sensor: Lens Limited.

 

[Pookeyhead]

David is also implying that the lens tests I do for magazines and websites are 'academic... bench tests'. Well, they're not. Some studio tests for sure but I don't have an optical bench and they're as real world as I can make them.

Can we see some of the lens tests you've conducted for them that show lenses out resolving high resolution sensors then?

 

[chris malcolm]

I have to disagree with this. It implies there's not a finite amount of resolution from any given lens, and you can just keep on upping the resolution of the sensor, and keep making gains, when it's patently obvious that's not the case. You will quickly reach a limit where the lens has nothing more to give.. hence the limitation is the lens, not the sensor: Lens Limited.

Not a necessary implication. As MP increase the detail resolution from a given lens increases, but by less and less as the MP increase. It also depends on what the resolution test is. Black stripes on white paper will be less sensitive than black twigs against a grey sky. It also depends on what image processing is allowed. Detail which has just become imperceptible on an ex-camera JPEG as MP are increased can be brought back to visibility by increasing contrast and sharpening. When that in turns fails, sophisticated noise reduction software may be able to make it visible again. Finally of course there will come an MP increment where no amount of careful selection of test and detail recovery post processing will recover any extra detail. That is the point at which for all kinds of image and processing the detail has become unarguably lost due to imperfections of the lens: the ulimate limit of the lens.

Let's say for the sake of argument that was at 80MP. Disallowing noise reduction might bring that down to 40MP. Disallowing back lit irregular contrast and using black ink on white paper might reduce that to 20MP. Disallowing differences which only expert pixel peepers can see at very large magnifications might reduce that to 10MP. Disallowing differences which most of the general public won't notice on an A4 print might reduce that to 5MP.

Don't forget that doubling MP is an area measure -- the linear increase in resolution which is what most resolution tests rely on is only an extra 41% when MP are doubled.

Note too that some of the technical definitions of "lens limited" are at the point where more than 50% of the lost detail is due to the lens, the other 50% to the sensor etc., in other words only at the point where lens limitations has become the dominant factor in detail resolution. That's some way before the point at which further increases in MP will show zero extra detail.

I'm suggesting that the figures in my above example are not unreasonable. In other words I'm suggesting that for the same lens one man's perfectly reasonable definition of lens limited detail might occur at 5MP, whereas another man's perfectly reasonable but different definition of lens limited might put it at 80MP.

It's rather like the improvements in musical fidelity you can get by doubling your investment in HiFi, a well known example of diminishing returns. For some the limit of technological improvement might be when doubling the cost made no discernible improvement in quality to 90% of listeners. But there might still be a discernible improvement if you quadrupled the cost, or if you changed the listening room from average sitting room to anechoic chamber, or if you chose your listening panel exclusively from those with trained "golden ears".

 

[Pookeyhead]

Not a necessary implication. As MP increase the detail resolution from a given lens increases, but by less and less as the MP increase.

Exactly... it's finite. When that happens, the limiting factor is the lens.

I'm not concerned with a technical definition... some arbitrary number where one decides where "lens limitation" begins.

If I can see softness, diffraction, or any other effects caused optically before I can discern the aliasing, then the limiting factor in that lens/sensor system is the lens. That can simply not be argued with.

The fact is, I can detect such with the D800E and even some fairly expensive lenses.

 

[HoppyUK]

I have to disagree with this. It implies there's not a finite amount of resolution from any given lens, and you can just keep on upping the resolution of the sensor, and keep making gains, when it's patently obvious that's not the case. You will quickly reach a limit where the lens has nothing more to give.. hence the limitation is the lens, not the sensor: Lens Limited.

I'm not saying there is no limit, only that if you want to specify a maximum resolution figure then there must also be a contrast level to go with it, or it's meaningless. Plenty of people would disagree with your 24mp ceiling on full-frame though, however you want to measure it, and the web is full of such comparisons. By the same standard of lens performance then, 24mp on FF equates to only 10mp on APS-C... You can't be serious.

Can we see some of the lens tests you've conducted for them that show lenses out resolving high resolution sensors then?

You can see them when they're published David, same as anyone else.

 

[Pookeyhead]

I'm not saying there is no limit, only that if you want to specify a maximum resolution figure

I don't. I just look at the image. If I can see lens defects, yet still not clearly discern the aliasing of the digital file, then I'll assume the lens is running out of steam before the sensor.

 

[sk66]

Take a detail projected at any give size onto a sensor (it's airy disk). If it perfectly matches the size of a pixel, then two pixels can ideally define the edge with maximum contrast. Now in a second case we have a sensor with smaller pixels and two pixels fit within that airy disk and it now takes four pixels to define the edge...same "detail" with less contrast. And if we can use eight pixels to define that airy disk we get finer resolution of that same detail at an even lower contrast.

In all cases except for the first, the lens is the limiting factor... in the second case we might say the resolution is diffraction limited. But even in the last case there is some minor (potential) benefit to using a higher resolution sensor (heavily offset by a loss of contrast which is very significant in our perception of sharpness/detail).
It's very similar to the position that the D7200 may print better at a large size simply due to the higher resolution, even if the image itself is limited to a lower detail MP count. And it is very similar to the idea that the D8xx can equal the D4 in low light due to oversampling (higher resolution).

But, IMO it is all fairly academic because none of these differences/potential differences will be visible in a case where you can view the image in entirety. The best information I can find indicates that if the print COC is redefined based upon the resolution of the human eye, the max requirement is in the 12MP range for any print size.

 

[HoppyUK]

Would some kind person resize the first 5D2 image for me, so that both are the same size for comparison? Try as I might, I can't get Photobucket to do it!
Thanks

Rest of post deleted. No point without properly sized images. See next two posts below.

Just for info, they were two big crops taken on a Canon 5D2 and 5DSR with a Sigma 35/1.4 Art lens. Maybe it's just as well, because they were images of a £10 note and the Bank of England gets a bit touchy about that kind of thing!

 

[Pookeyhead]

Would some kind person resize the first 5D2 image for me, so that both are the same size for comparison? Try as I might, I can't get Photobucket to do it!
Thanks

Can't download it.



Besides... Many lenses can out resolve a 24MP sensor, and the 35mm ART lens is one of them, as I suspect you know already. However... they're few and far between. Try it with a Canon L series 35mm.. Also... a better test still would be the 5D S image against a D810 image.

DP Review have raws for comparison. The 5DS R image was taken with a Canon 85mm 1.4L lens, and the D810 image with a Nikkor 85mm 1.4G. I've resized the Canon image to 7360 across. (If you do this yourselves, do NOT use any sharpening as part of PS's resizing... use BILINEAR or it's an unfair comparison.

D810

5D SR (resized to 7360 across)

However.... I'm wondering what the point of resizing the image is, as both use the same size sensor, so they'll both appear the same when resized... which they do. In all honesty... apart from the Canon file being bigger, there was nothing to tell them apart anyway.

 

[HoppyUK]

Can't download it.

View attachment 49493

Besides... Many lenses can out resolve a 24MP sensor, and the 35mm ART lens is one of them, as I suspect you know already. However... they're few and far between. Try it with a Canon L series 35mm.. Also... a better test still would be the 5D S image against a D810 image.

DP Review have raws for comparison. The 5DS R image was taken with a Canon 85mm 1.4L lens, and the D810 image with a Nikkor 85mm 1.4G. I've resized the Canon image to 7360 across. (If you do this yourselves, do NOT use any sharpening as part of PS's resizing... use BILINEAR or it's an unfair comparison.

D810

5D SR (resized to 7360 across)

However.... I'm wondering what the point of resizing the image is, as both use the same size sensor, so they'll both appear the same when resized... which they do. In all honesty... apart from the Canon file being bigger, there was nothing to tell them apart anyway.

Thanks for trying David. That's a shame as I think they're worth showing - the difference is clear and obvious when you get to look at them properly compared.

But on the other hand, since you've just conceded that "many lenses can out resolve a 24MP sensor" that's case closed AFAICS.

That's to miss the main point I was trying to make though, which was simply that stating 'this out-performs that' is meaningless without also specifying all the relevant parameters.

 

[Pookeyhead]

But on the other hand, since you've just conceded that "many lenses can out resolve a 24MP sensor" that's case closed AFAICS.

LOL.. that doesn't mean they can make use of a 50MP sensor!!

 

[The W]

Surely until the day we're free of the tyranny of the bayer sensor, 'resolution gain' will occur until we've got all three colours within the circle of confusion?

 

[sk66]

there was nothing to tell them apart anyway.

One of the potential advantages of using a sensor of greater resolution than the lens/scene is greater color accuracy (offsetting the bayer array). In the posted images I see a few examples of that. In the fine text, particularly the white on black, the 5DR has less false color. The same is true in the line drawing of the man's portrait. Interestingly, they both show significant false color in the line drawing of the women, but differently.

The other (debatable) benefit is that if the sensor resolution is far greater than the lens/scene, then the "diffraction limitation" removes the requirement/benefit of an AA filter (which is how the D8xx gets away without one *most of the time*). Interestingly, the 5DR has an AA filter, but the second layer "recombines" the image (I doubt the manipulation is without *any* effect).

 

[Pookeyhead]

Interestingly, the 5DR has an AA filter, but the second layer "recombines" the image (I doubt the manipulation is without *any* effect).

That's what the D800E does.

 

[The23rdman]

Do you think that the OP has lost the will to live yet?

 

[Phil V]

Do you think that the OP has lost the will to live yet?

Who'd have predicted it.

 

[The23rdman]

Who'd have predicted it.

What that someone would ask a moderately technical question and eventually a couple of geeks would take over the thread to the point that the OP gives up? Not me. No, never.

@David and Richard: I'm only gently teasing.

 

[HoppyUK]

What that someone would ask a moderately technical question and eventually a couple of geeks would take over the thread to the point that the OP gives up? Not me. No, never.

@David and Richard: I'm only gently teasing.

Guilty as charged, m'lud

 

[htid]

I for one am reading the thread and find it interesting

 

[sk66]

That's what the D800E does.

I forgot about that.

 

[sk66]

What that someone would ask a moderately technical question and eventually a couple of geeks would take over the thread to the point that the OP gives up? Not me. No, never.

@David and Richard: I'm only gently teasing.

Honestly, I think it was an oversimplified question to a very technical question. And pretty much unanswerable as asked.

I suppose someone could have said "for 90+% of people the D750 will equal/beat the D7200 in 90+% of situations/uses." But I doubt that would have worked or really benefitted anyone.

 

[Nod]

I for one am reading the thread and find it interesting

Me too. Like many such threads, I don't fully understand all the answers but, like women, being fairly unfathomable doesn't reduce my interest in the subject!!!