What if 36mp DSLRs were around the corner?

[torger]

At the image plane smaller sensored cameras have more DOF

This is true when DoF is by tradition defined with a circle of confusion of 25 microns. This may have been a good definition in the film days when relating to film grain. Today I think it is better to define DoF with a blur spot roughly similar in size to the diffraction blur instead (this is what I use in the field), then you get a DoF definition which is not related to sensor size, but to the actual resolution.

[RFPhotography]

I can see your point.  I can see some practical difficulties with it and here's where I come back to the idea of practical concerns.  What you're suggesting is a different 'standard' for DOF on an individual camera basis.  And again, while I see the point you're making, for the majority of photographers they're not going to be interested in - or perhaps not capable of - making those types of calculations.  With that methodology, though, you then run into an issue where DOF begins to become less once you pass the point of diffraction of the lens, right?  Is the DOF under that scenario then not also based on the lens, which adds a different variable?

[Sheldon N]

This is true when DoF is by tradition defined with a circle of confusion of 25 microns. This may have been a good definition in the film days when relating to film grain. Today I think it is better to define DoF with a blur spot roughly similar in size to the diffraction blur instead (this is what I use in the field), then you get a DoF definition which is not related to sensor size, but to the actual resolution.

There is only less depth of field with a higher resolution camera if you print larger.

The "old" circle of confusions had nothing to do with film grain, they had everything to do with a standard print size and visual acuity/viewing distance.

What you are proposing is essentially a DOF where the new "standard" print size is 100% pixel view on your monitor at home, so every camera is held to a different standard.

[uaiomex]

At first I thought the same and had a little bit of trouble but since I was used to often crop from square, cropping to 3:4 is not much of a deal anymore. Besides, I've found out that that art directors now often use the whole format for a personal choice. As a matter of fact AD's had trouble too but learned how to deal with the "new" aspect ratio. I remember more complaints from delivered square slides. One great thing, 3:2 matches a double spread. I still prefer 3:4 and 4:5 for verticals though. This issue will completely disappear with the mirrorless FF and the multiaspect sensor in the fashion of the Panasonic GH2. This kind of camera will seriously get into DMF territory. Two more EVF generations and we are there. Digital ground glass anyone?
Eduardo

but will it still be 3::2 format?  Ugh!  After 4x5, 6x6 ... I just can't get used to the DSLR format anymore.

[BernardLanguillier]

True.  But those smaller pixels will need to be 'enlarged' more to make that 20" print.  That additional 'enlargement' causes degradation.  Whether it's DOF or accentuating camera shake.

Bob, sorry, but no. A pixel is a pixel and has no memory of how it collected information.

The mathematical truth is that DoF is inversely proportional to the format size, which is easy to confirm with everyday usage of cameras with different formats.

What you probably mean is that:
1. The acceptable circle of confusion also decreases when using a smaller sensor since the pixels become smaller. True also but the impact of this is less than the impact of the increase of the focal lenght.

Per the link below, you can establish that the depth of field is linearly proportional to the circle of confusion (and therefore format) but that it is inversely proportional to the square of the focal lenght.

http://www.bobatkins.com/photography/technical/digitaldof.html

-> the net result is that you do have more DoF in your final image with the smaller format.

2. It takes a higher quality lens to gather equivalent quality light information at the level of each of these pixels. Yes, but on the contrary you need a much smaller image circle, and it is clear that a small image circle of high quality is easier to achieve than a large image circle.

Cheers,
Bernard

[bjanes]

I can see that being the case at the image plane.  But in a print it doesn't make as much sense.  It's similar to the depth of field issue, I think.  At the image plane smaller sensored cameras have more DOF but in a print not so because those smaller pixels have to be 'magnified' more to make a print of the same size as a larger sensored camera.  It seems the same would be so for 'pixel blur' in this instance. 

Look at Roger Clark's article on comparing depth of field on a P&S and a dSLR. If you shoot at the same f/number, the P&S will have more depth of field, and this is common experience and is shown in Roger's figure 1. Your explanation about magnification is incorrect. However, the P&S has to use a larger aperture to maintain image quality. If you keep the lens aperture (the actual diameter, not the f/number) the same on both cameras, the depth of field will be equal, also shown in figure 1.

Regards,

Bill

Regards,

Bill

[clkirksey]

I don't see that it changes anything regarding technique.  Cameras with equivalent (or smaller) pixels already exist on the market and have for several years.  A given absolute amount of shake (due to wind, slap, etc.) will have just as much pixel-level blur on these existing cameras. 

The degree of camera shake and/or subject motion that degrades the resolution of a camera is a fuction of the cameras instantaneous field of view (IFOV) which is basically pixel size/FL. This general has remained about the same for say the 5DII and 7D. For the newer 36MP camera the IFOV would be reduced by sqrt(21/36) or about 1/1.3. So the SS would have to be increased by a similar amount.

[EricWHiss]

Just stitch, crop or keep not using a Dslr. :-)

Cheers,
Bernard

"To a hammer everything looks like a nail" ...  and to Bernard everything can be stitched?  But Bernard unlike you, I shoot people and things that live and move - except for some art reproduction work. It's not really an option for most of what I do.  The only way for me is to crop a lot of the image away, and then what's the point? So while I'd like a camera that does multipoint AF and can shoot a usable ISO 3200, I guess I will stick with my MF cameras for now.

[ErikKaffehr]

Hi,

I don't see your point, you just tell your people: "Freeze! Don't move!"

Seriously, I don't mind cropping. In my view crop should fit image and not the other way around. On the other hand, I find it hard to compose for say a quadratic crop using a rectangular viewfinder. Most of my pictures are shown either in HD (2 MPixels) or on A2 format prints, and 3:2 fits both of those formats decently.

Regarding the 36 MP, it really corresponds to what we have on APS-C, and I generally would say that it would be welcome, in my view, but I'm the camera on tripod with low ISO kind of guy.

Best regards
Erik

"To a hammer everything looks like a nail" ...  and to Bernard everything can be stitched?  But Bernard unlike you, I shoot people and things that live and move - except for some art reproduction work. It's not really an option for most of what I do.  The only way for me is to crop a lot of the image away, and then what's the point? So while I'd like a camera that does multipoint AF and can shoot a usable ISO 3200, I guess I will stick with my MF cameras for now.

[torger]

I can see your point.  I can see some practical difficulties with it and here's where I come back to the idea of practical concerns.  What you're suggesting is a different 'standard' for DOF on an individual camera basis.  And again, while I see the point you're making, for the majority of photographers they're not going to be interested in - or perhaps not capable of - making those types of calculations.  With that methodology, though, you then run into an issue where DOF begins to become less once you pass the point of diffraction of the lens, right?  Is the DOF under that scenario then not also based on the lens, which adds a different variable?

The definition is especially interesting for landscape shooters like me that often want maximum DoF ("whole picture sharp") and high resolution large prints. If I use 25 um as the CoC for acceptable DoF that corresponds to about the same resolution I get from f/22 aperture, it would also make a bias for large sensor system for no reason. It makes more sense to relate DoF to what the system can resolve, which means that for f/8 I use a smaller blur spot than for f/22 to know where the DoF ends.

For short DoF photography these calculations are not really important I think, because then you just want focus at a spot and a suitable DoF to get the look you want. One may need a large sensor in that case to get as short DoF as one would like.

DoF calculations are used by those that want large DoF, perhaps even do focus stacking to extend it, and then I think it is relevant to relate it to resolving power. It is not a different definition per camera, it is just related to resolving power (which varies with aperture setting).

With this definition the DoF increases when resolving power is reduced, since it makes no sense to have a smaller circle of confusion than the system can resolve. So at f/22 you get an increased DoF both from the smaller aperture as such, but also due to a larger CoC.

[EgillBjarki]

I hope and I think that this will be the case, small body DSLR's with a high MP sensor. Canon has made allot of money of the 5DII and I think they are keen on doing the same with the update.

I'm very interested to know how the lenses will hold up to a big MP jump. Personally I have invested in prime lenses in hopes that they will hold up better with increased resolution in the future body's.

[torger]

There is only less depth of field with a higher resolution camera if you print larger.

The "old" circle of confusions had nothing to do with film grain, they had everything to do with a standard print size and visual acuity/viewing distance.

What you are proposing is essentially a DOF where the new "standard" print size is 100% pixel view on your monitor at home, so every camera is held to a different standard.

You are right concerning standard print size viewing distance etc, but these where set obviously related to what the current camera technology could do rather than limits of human vision. That you used 25 um on the film was assuming that resolution of a system was mainly limited by film size, which it was. But this is no longer true. A modern 36x24mm sensor can be made to have higher resolving power than an old medium format sensor.

I don't agree with the last point however. I'd say that every camera is held to the same standard - namely to its resolving power. Sure if you know the print size you are going to make you can adapt the required resolution to that, and use that as a reference for CoC, but that would not either translate to a fixed "25 um" CoC (or was it 30 um?), but a different sized CoC depending on print size.

The reason why I use the definition above is that I use the DoF calculations when I want to maximize DoF, get "whole picture sharp" and at the same time maximize resolution (so I can make large prints), and then typically choose between f/8 (typically best corner-to-corner resolution compromise) and the more diffraction limited f/11 or f/16, or tilt the focal plane, or make DoF compromises (let some parts be out of focus). Diffraction limits the resolution somewhat in all cases, therefore I feel that relating the DoF CoC to the airy disc is very relevant and makes the decision process concerning the best aperture and focus placement (and focal plane tilt) more exact. Sure the lens can limit too, especially in the corners so it is not an 100% exact method, but does not need to be and I think it is way better than the fixed size 25 um CoC.

With this method focusing at hyperfocal distance actually becomes usable too (which has got its poor reputation partly due to inadequate DoF definition). Using fixed size 25 um CoC and focus at hyperfocal distance from that those nice mountains in the background won't really be that sharp. For example with 25 um CoC hyperfocal distance is 2.88m with 24mm f/8, but with hyperfocal related to diffraction the distance becomes about 6.7m. Due to difficulty measuring etc using some safe overshoot at ~10m is good, close focus is then at 4m instead of maximum 3.4 which is okay compromise for making sure the distant objects are sharp, I usually prefer some unsharpness in foreground rather than background, since details in foreground is typically larger. Since blurs add up a bit one could argue that CoC should be smaller than airy disc, but I find this a quite ok compromise.

[RFPhotography]

Bob, sorry, but no. A pixel is a pixel and has no memory of how it collected information.

The mathematical truth is that DoF is inversely proportional to the format size, which is easy to confirm with everyday usage of cameras with different formats.

What you probably mean is that:
1. The acceptable circle of confusion also decreases when using a smaller sensor since the pixels become smaller. True also but the impact of this is less than the impact of the increase of the focal lenght.

Per the link below, you can establish that the depth of field is linearly proportional to the circle of confusion (and therefore format) but that it is inversely proportional to the square of the focal lenght.

http://www.bobatkins.com/photography/technical/digitaldof.html

-> the net result is that you do have more DoF in your final image with the smaller format.

2. It takes a higher quality lens to gather equivalent quality light information at the level of each of these pixels. Yes, but on the contrary you need a much smaller image circle, and it is clear that a small image circle of high quality is easier to achieve than a large image circle.

Cheers,
Bernard

I have a question for Atkins but there doesn't seem to be a way to contact him so it'll have to be unresolved.

Here's what I get from reading the article.  First, the standard he's suggesting for evaluating DOF is the same one I've always alluded to.  CoC of 1/100" (~250 microns) in an 8x10 (or 8x12) print.  The first chart in the article shows what the CoC needs to be on the original image (i.e., at the image plane) to achieve that 1/100" in a print.  That first chart also refers to the smaller original from the cropped camera being 'magnified' more than the image from the full frame sensor.  He uses the term magnified.  I used enlarged.  Same concept. 

In order to get that 1/100" CoC in the standard print the CoC at the image plane needs to be smaller on the cropped frame camera so that when it gets 'magnified' in a print, it ends up at the required size.  The rest of the article goes on to further prove that the cropped sensor image will, indeed, have greater DOF at the image plane.  The same CoC sizes in the original (i.e., at the image plane) are used further down in the article for the other calculations to show that the cropped frame image has 1.6X greater DOF at the image plane.  But the first chart shows that that difference goes away in the standard print due to the greater magnification of the cropped frame image. 

This, as is pointed out in the article, is a generalisation and doesn't apply in special cases such as hyperfocal or macro.

None of this seems inconsistent with what I'd outlined earlier.  In the standard measurement for DOF, assuming two images of the same field of view and same aperture, the DOF will be the same.  But at the image plane the digital image will have greater DOF. 

[joneil]

  If I can leave all this talk about pixels, lenses, dof, etc, behind and ask a more preactical  question - exactly what good will a full frame , 36 MP camera be good for?  Let me give you a direct example.  

   I once met a man who did a lot of stock photography, and his "specialty" was shooting for ads for those large billboards you see beside highways and on the side of buildings, etc.  He told me that all things being equal, the larger the sensor AND the higher the MP count, the better.    Now I have never shot for billboards so I don't know if this is true or not.  My idea of needing detail for large prints is to put away my 4x5 and pull out my 8x10".  

     But seriously, the camera here in question is the rumoured Nikon D800.  I love the D7000 I use right now, but if my intention is to start shooting so that I can make 4 foot by 6 foot prints or have my photos show up on billboards, is the new D800 the way to go?

    What other practical needs and considerations in the studio and/or the field does the D800 solve that say a D7000 not address now?  For me it's money, do I really need a full frame, 36 MP?  What about the rest of you, with the economy the way it is, what need does this camera fill?

joe

[bjanes]

But seriously, the camera here in question is the rumoured Nikon D800.  I love the D7000 I use right now, but if my intention is to start shooting so that I can make 4 foot by 6 foot prints or have my photos show up on billboards, is the new D800 the way to go?

What other practical needs and considerations in the studio and/or the field does the D800 solve that say a D7000 not address now?  For me it's money, do I really need a full frame, 36 MP?  What about the rest of you, with the economy the way it is, what need does this camera fill?

Billboards are viewed from a distance and I do not think that the resolution requirements are any greater than needed when viewing your high resolution screen at a distance of 18 inches. It is the angular resolution that counts. If you occasionally need higher resolution than your D7000 can deliver, take three or more images in portrait mode and stitch. If you regularly need higher resolution, the D800 would be more convenient.

Regards,

Bill

[torger]

   What other practical needs and considerations in the studio and/or the field does the D800 solve that say a D7000 not address now?  For me it's money, do I really need a full frame, 36 MP?  What about the rest of you, with the economy the way it is, what need does this camera fill?

It fills the gap between traditional 135 DSLRs and medium format. I think one should not see it as an expensive alternative to an APS-C camera, it is rather a cheap alternative to a medium format system.

Do we really need this kind of resolution? For fine art prints of landscape photography it is kind of nice. I like to be able to produce images that look sharp even when walking up close. At close range, print resolution up to 300-400 ppi can be appreciated. It adds a wow-feeling if the image contains detail beyond what you can see at a distance. For example I recently made a landscape view where you in one part of the image can see a small boat in the distance creating an interesting wave pattern, walking up close you could even see the man sitting it it. In this case I had used stitching (the moving water was fortunately captured in one frame) to catch the resolution. I think that high resolution also gives the prints a high quality feel. When I frame prints I use the best material, glass etc, I like the finished product not only to be a good picture artistically but have the feel of high quality workmanship throughout.

High resolution does not improve artistry though, but that's not the point. I just like to have the best tools I can afford (assuming it gives visible improvements compared to cheaper tools).

However, honestly I think 36 megapixel fullframe is kind of a niche camera like medium format is, the major part of the users will not need it or even employ the photographic technique (high quality lenses, precise focusing, stable tripod, mirror up etc) to make use of all this resolution. Or maybe they do? While landscape photography is a very small niche professionally, it seems to be one of the larger genres among amateurs.

[ErikKaffehr]

Hi,

It's actually very simple. A full frame sensor is 1.5 times larger than a DX sensor. Sensor pitch is about the same on both. So with the full frame camera you can make prints that are 1.5x larger (on each side) than with a DX camera. That mean that if you feel that A2 prints are OK on the D7000 than you can make A1 prints with exactly the same quality with the D800.

Best regards
Erik

 If I can leave all this talk about pixels, lenses, dof, etc, behind and ask a more preactical  question - exactly what good will a full frame , 36 MP camera be good for?  Let me give you a direct example.  

   I once met a man who did a lot of stock photography, and his "specialty" was shooting for ads for those large billboards you see beside highways and on the side of buildings, etc.  He told me that all things being equal, the larger the sensor AND the higher the MP count, the better.    Now I have never shot for billboards so I don't know if this is true or not.  My idea of needing detail for large prints is to put away my 4x5 and pull out my 8x10".  

     But seriously, the camera here in question is the rumoured Nikon D800.  I love the D7000 I use right now, but if my intention is to start shooting so that I can make 4 foot by 6 foot prints or have my photos show up on billboards, is the new D800 the way to go?

    What other practical needs and considerations in the studio and/or the field does the D800 solve that say a D7000 not address now?  For me it's money, do I really need a full frame, 36 MP?  What about the rest of you, with the economy the way it is, what need does this camera fill?

joe

[joneil]

High resolution does not improve artistry though, but that's not the point. I just like to have the best tools I can afford (assuming it gives visible improvements compared to cheaper tools).

-snip-

  Exactly, i agree completely on all counts.  Hey, I still have 6 boxes of 4x5 tech pan in the freezer, and powdered technidol, so anytime I want "the ultimate resolution" - whatever that may be, it's hard to beat good ole tech pan and a g-claron lens.

  but the issue becomes, at least to me, do people notice anymore?  i still shoot 4x5 (was in the darkroom just two nights ago as a matter of fact), but even there, despite the fact that T-Max and even Ilford Delta is incredibly sharp and fined grain, I like the "look" of plain old Tri-X better, which is not nearly as fine grained as those other films.    Some people notice the difference between a print I made off a 4x5 or even my 8x10 vs an old 35mm or a digital SLR, but a lot do not.  So in these days when the economy is in the crapper and every cent counts, there has to be more to this camera than just extra MPs to justify the cost.

     One reason people buy the D3s or the D300s is they are built like tanks, will take serious use and abuse, and for sports shooting they have processors that will allow you to smack in  6 fps without overloading the buffer, which is really my only grumble about the D7000, and a minor one at that.     I even see some of the news guys around here using the D7000 now.  So I was hoping to hear specific examples of how people find the bigger frame and higher MPs directly useful.

 The other issue too, if people want to get into technical details, what kind of computer power are you going to need to handle 36mp raw files?  I just brought my computer back from the shop today for upgrades I find I am having issues with.     The processors run so hot not only did I need extra cooling fans ( I was even offered the option of a water cooled CPU),  I can darned near heat my office in winter with just the computer turned on.  At the rate things are going, someday we will all have computers  with more ram, storage space and CPU power  than NORAD or the NSA. 

  Don't get me wrong, I would love to own one of the new D800s, just based on rumor alone, but practically, with the way the economy is around here (Ford last month just shut down a major plant, throwing hundreds out of work), Nikon  or Canon are going to have to do a better job showing me why I need a better camera.   
joe

[Sheldon N]

I don't agree with the last point however. I'd say that every camera is held to the same standard - namely to its resolving power. Sure if you know the print size you are going to make you can adapt the required resolution to that, and use that as a reference for CoC, but that would not either translate to a fixed "25 um" CoC (or was it 30 um?), but a different sized CoC depending on print size.

The reason why I use the definition above is that I use the DoF calculations when I want to maximize DoF, get "whole picture sharp" and at the same time maximize resolution (so I can make large prints), and then typically choose between f/8 (typically best corner-to-corner resolution compromise) and the more diffraction limited f/11 or f/16, or tilt the focal plane, or make DoF compromises (let some parts be out of focus). Diffraction limits the resolution somewhat in all cases, therefore I feel that relating the DoF CoC to the airy disc is very relevant and makes the decision process concerning the best aperture and focus placement (and focal plane tilt) more exact. Sure the lens can limit too, especially in the corners so it is not an 100% exact method, but does not need to be and I think it is way better than the fixed size 25 um CoC.

With this method focusing at hyperfocal distance actually becomes usable too (which has got its poor reputation partly due to inadequate DoF definition). Using fixed size 25 um CoC and focus at hyperfocal distance from that those nice mountains in the background won't really be that sharp. For example with 25 um CoC hyperfocal distance is 2.88m with 24mm f/8, but with hyperfocal related to diffraction the distance becomes about 6.7m. Due to difficulty measuring etc using some safe overshoot at ~10m is good, close focus is then at 4m instead of maximum 3.4 which is okay compromise for making sure the distant objects are sharp, I usually prefer some unsharpness in foreground rather than background, since details in foreground is typically larger. Since blurs add up a bit one could argue that CoC should be smaller than airy disc, but I find this a quite ok compromise.

There's nothing wrong with calculating what the diffraction limited aperture will be based on a "per pixel" basis for each camera. It will let you know at what aperture you start to begin to trade resolution for deeper DOF, and help you know how much DOF you will have for a given image if you stick with the diffraction limited aperture at the largest print size.  All very good information.

The underlying assumption that go with this method are that you are going to print at the maximum print size possible for your given camera, then view that image at a viewing distance that allows you to see all the way to the "per pixel" detail level.  In other words, print as big as you can then viewing the image as close as possible. You could look at it in terms of "print" size, or you could look at is as being the same way that we view images at 100% on the monitor.  In jest, you could almost call this the "Pixel Peeper's DOF".

The only issue that I have is that it's not a useful as a comparative method between cameras or formats. You can't say that one camera has more DOF or less DOF, since you are subjecting each camera to a different set of viewing conditions based on the camera's capabilities.

[torger]

There's nothing wrong with calculating what the diffraction limited aperture will be based on a "per pixel" basis for each camera. It will let you know at what aperture you start to begin to trade resolution for deeper DOF, and help you know how much DOF you will have for a given image if you stick with the diffraction limited aperture at the largest print size.  All very good information.

The underlying assumption that go with this method are that you are going to print at the maximum print size possible for your given camera, then view that image at a viewing distance that allows you to see all the way to the "per pixel" detail level.  In other words, print as big as you can then viewing the image as close as possible. You could look at it in terms of "print" size, or you could look at is as being the same way that we view images at 100% on the monitor.  In jest, you could almost call this the "Pixel Peeper's DOF".

The only issue that I have is that it's not a useful as a comparative method between cameras or formats. You can't say that one camera has more DOF or less DOF, since you are subjecting each camera to a different set of viewing conditions based on the camera's capabilities.

You can say that a camera has possibility of less DoF, since the minimum DoF is limited by the lens maximum aperture / focal length combination which tend to be larger for larger formats.

On the maximum DoF side what definition you choose depends on what you want to use it for. A perfectly valid approach would be to have a quite loose definition, say that 25 um thing, and just put focus on what's important to be in focus in the scene and let some things be a bit blurry looking close on a larger print, simply not worry too much, and say that the ultimate system resolution is appriciated in the focal plane anyway, so the high resolution is not thrown away. I think that is an ok way to work.

I am myself a quite technical photographer though, and feel more confident when I more closely know the parameters. For example, if I change from f/8 to f/11 in a scene, what exactly will happen, can I go from tack sharp background with slighly unsharp foreground to a very slightly less sharp background with an equally sharp foreground, ok then I might choose f/11 instead, or I decide that foreground is low contrast and do not gain from that sharpness so I use f/8 instead. Or when there's focus stacking, to avoid uneven sharpness it is good to use the "pixel peeper's DoF" :-).

When I do tilt though, then it is much more "just doing it", it seems so far too complicated and unprecise to use calculations, but perhaps I may try making some useful tables in the future. I'm fairly new to having the tilt option at hand. Anyway, today when I do a near-far composition I first check the hyperfocal alternative using DoF calculations, and if that does not work I do tilt and hope for the best ;-), and if really hopeless I may do focus stacking (or just choose another more reasonable composition). With tilt there is nearly always something a bit out of focus, so then my thinking is to get all the important parts in focus, and have not too inconspicous out of focus areas.