Pages: [1] 2   Go Down

Author Topic: Format sizes, MPs and printing large  (Read 5306 times)

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Format sizes, MPs and printing large
« on: January 26, 2012, 09:31:56 pm »

Now this subject probably has been covered here and at other forums but I would like to have someone educate me as to the errors in my thinking. So here goes.

First let us look at what might me considered "equivalent images". If you view two images you would hope that at a minimum the two would have the same geometric characteristics. So what are these? A camera projects a three dimensional view onto a two dimensional surface defined as the image plane. To achieve the same projection three geomtric conditions have to be met.

Firstly, The field of view (FOV) has to be the same. FOV is related to sensor size and focal length, basically (sensor area)/FL^2. So the FL of the cameras need to be scaled by the format size. Secondly, the aspect ratio of both formats have to be the same:3:2, 4:3, 16:9, or whatever. Thirdly, the perspective of the two views has to be the same. This is achieved by maintaining the same camera-to-subject distance. Now if these three conditions are met then the two images will have the same geometric characteristics.

What about the optical characteristics? The image that is projected onto the image plane is a transformation of the scene radiance to image plane photon intensity. This transformation causes the ideal geometric transformation of the scene to become blurred. Even if the camera is perfect blurring will still occur. The lower bound of the blurring is defined by the diffraction due to the cameras finite aperture and the particular wavelength of light being considered. If we desire that this lower bound of blurring be the same for both images it is necessary that both cameras have the same aperture. Since the cameras F# is defined as FL/aperture diameter and FL is scaled with format size this will mean that the cameras F#s will also be scaled by format size.

So far we have achieved the same geometric scene projection on the image plane and the same degree of optical blurring of that projection. Next we must look at the conversion of the optical image to an electronic image via the sensor. So what characteristics do we want the sensors to have? Simple: the same number of MPs and the same optical and electrical properties. If we do this then the optical image will be converted at the two image planes into electrical signals that represent the optical images perfectly. Except for noise. Ideally the SNR of the two images will be related by the square root of the sensor area. A simple example. Compare a 35FF and M4/3 and ignore the aspect ratio differences. The 35FF sensor area will four times that of the sensor area of the M4/3. This will result in a factor of two SNR advantage for the 35FF or a two stop advantage.

Now the raw conversion process will also be dependent on SNR. So it will be very likely that the conversions of the two images will be affect. The lower the SNR the more of a negative effect. Image resolution be lowered as the SNR decreases but in a very nonlinear manner.

Now if the above conditions are met can we draw the following conclusions?
Both cameras can print images of the same size.

The depth of field (DOF) will be the same for both images.

The image SNR will scale as the sqrt(sensor area) and the smaller format may appear to be nosier depending on the SNR of the larger format and the ratio of the sensor areas.

The resolution of the two images may not appear significantly different depending again on the SNR of the larger format and the ratio of the sensor areas.

There is no inherent resolution or print size advantage to the larger format.

OK I know I must be missing something. I am here to learn.
Logged
A bird in the viewfinder...

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Format sizes, MPs and printing large
« Reply #1 on: January 26, 2012, 10:38:52 pm »

I should add that there is a limit on F# of .5 (this is a theoritical limit but the realistic limit is probably f/.9 and depnds on FL). Thus a smaller format using a scaled F# and FL may run into a practical limits.
Logged
A bird in the viewfinder...

bill t.

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3011
    • http://www.unit16.net
Re: Format sizes, MPs and printing large
« Reply #2 on: January 27, 2012, 12:03:18 am »

So you are requiring the absolute diameter of the aperture to remain the same, thereby giving the smaller camera an effectively wider f-stop ratio, ie perhap f5.6 for the smaller camera versus maybe f8 for the bigger one.  While that would equalize diffraction due to aperture, diffraction cause by elements other than aperture is also present in optical designs, and that would always disadvantage the smaller optic, even if you increased the quality of the optical fabrication relative to the larger one.

But you have certainly laid out the basic physics of it all very well, as far as I can tell.  All I know is, when things are not so much equal as nominal my 5D2 manifests better image quality than my D7000, which in turn beats the Fuji F31, which is miles ahead of the smartphone, in simple order of scale.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Format sizes, MPs and printing large
« Reply #3 on: January 28, 2012, 04:15:35 pm »

Hi,

This http://www.photodo.com/topic_138.html be good reading on the issue...

Best regards
Erik Kaffehr

So you are requiring the absolute diameter of the aperture to remain the same, thereby giving the smaller camera an effectively wider f-stop ratio, ie perhap f5.6 for the smaller camera versus maybe f8 for the bigger one.  While that would equalize diffraction due to aperture, diffraction cause by elements other than aperture is also present in optical designs, and that would always disadvantage the smaller optic, even if you increased the quality of the optical fabrication relative to the larger one.

But you have certainly laid out the basic physics of it all very well, as far as I can tell.  All I know is, when things are not so much equal as nominal my 5D2 manifests better image quality than my D7000, which in turn beats the Fuji F31, which is miles ahead of the smartphone, in simple order of scale.
Logged
Erik Kaffehr
 

bill t.

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3011
    • http://www.unit16.net
Re: Format sizes, MPs and printing large
« Reply #4 on: January 28, 2012, 05:13:40 pm »

Based on a lifetime of using many different kinds of cameras and making more prints than there are useless filter adapter rings in the drawers at a New York camera dealer, I feel the following statement sums it all up...

Large format photographs sometimes look smoother than those from smaller formats.


bill t.'s Conjecture, 2012
Logged

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Format sizes, MPs and printing large
« Reply #5 on: January 30, 2012, 12:45:08 pm »

Hi,

This http://www.photodo.com/topic_138.html be good reading on the issue...

Best regards
Erik Kaffehr

So Erik what conclusions do you draw from this article? Lens resolution as measured in LP/PH seems to be equal. Film resolution should be the same measured in LP/mm. I have no real experience evaluating film negatives so I am unclear what conclusion to draw. Perhaps the test just indicates that the film out resolves the lenses. Not sure.
Logged
A bird in the viewfinder...

Rob C

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 24074
Re: Format sizes, MPs and printing large
« Reply #6 on: January 30, 2012, 02:27:25 pm »

Heysoos!

I sense another Perspective circle coming on... with lots more pissing in the pond. Can hardly wait.

Rob C

EricV

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 270
Re: Format sizes, MPs and printing large
« Reply #7 on: January 30, 2012, 08:04:49 pm »

If we desire that this lower bound of blurring be the same for both images ....
Careful here -- do you want the same blurring on the sensor or on a final print?
Logged

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Format sizes, MPs and printing large
« Reply #8 on: January 30, 2012, 10:34:17 pm »

Careful here -- do you want the same blurring on the sensor or on a final print?

Does it matter? The number of MPs is the same and the optical blur is the same, relative to sensor size. So you will get the same image printed. Why would they be different? Curious.
Logged
A bird in the viewfinder...

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Format sizes, MPs and printing large
« Reply #9 on: January 30, 2012, 11:49:04 pm »

Hi,

What Lars Kjellberg shows that some of your reasoning holds for real. But I would say that all photography is not diffraction limited. Whatever format I have I normally strive for f/8 (or so) where lenses perform best.

Best regards
Erik

So Erik what conclusions do you draw from this article? Lens resolution as measured in LP/PH seems to be equal. Film resolution should be the same measured in LP/mm. I have no real experience evaluating film negatives so I am unclear what conclusion to draw. Perhaps the test just indicates that the film out resolves the lenses. Not sure.
Logged
Erik Kaffehr
 

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Format sizes, MPs and printing large
« Reply #10 on: January 31, 2012, 09:01:43 am »

Hi,

What Lars Kjellberg shows that some of your reasoning holds for real. But I would say that all photography is not diffraction limited. Whatever format I have I normally strive for f/8 (or so) where lenses perform best.

Best regards
Erik

But I think that is the point. In order to produce "equivalent" images from different formats the smaller format must use lenses with larger F#s and shorter FLs which in some cases do not exist or can not be made at all. It seems that a larger format is more likely to produce an "equivalent" image to a smaller format than vice versa. BTW thanks for your input. I have been trying to get my hands around this subject since there so many formats available and so many claims of performance advantages. Just witness the latest roil here at LULA.
Logged
A bird in the viewfinder...

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Format sizes, MPs and printing large
« Reply #11 on: January 31, 2012, 11:35:30 am »

But I would say that all photography is not diffraction limited. Whatever format I have I normally strive for f/8 (or so) where lenses perform best.
You have a strange narrow concept of best lens performance, ignoring the sharpness of parts of the scene that are not at the focal plane; because with f/8 in all formats, a larger format like 4x5 or 8x10 is going to have heavy blurring of parts of the scene that are quite sharp in a smaller format.

What those tests at PhotoDo suggest to me is:
once
1) the lens is stopped down enough to have adequately low aberration effects (which can happen at far lower then f/8 these days; even down to about f/4 with some zooms, let alone primes),
and
2) the film or sensor resolution is as good as say TMAX-400
then a larger format will not give a sharper image overall.
At best, you can increase sharpness new the focal plane at the cost of decreasing it away from the focal plane, by accepting greater OOF blurring. (For example, by using equal f-stop and thus larger effective aperture size in a later format.)

Those tests indicate that at aperture ratios of f/5.6 or higher with a good 35mm prime lens, larger formats could not do better under the requirement of equal or greater DOF.

That still leaves dynamic range, tonal graduations, and such to be discussed, along with the film-vs-digital resolution debate.
« Last Edit: January 31, 2012, 07:58:49 pm by BJL »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Format sizes, MPs and printing large
« Reply #12 on: February 01, 2012, 03:11:36 am »

Hi,

I didn't say that I always use f/8, I said I strive to use f/8. Why?

- Autofocus is inprecise, so is manual focus (except possibly LV). So I want to have some margin for improper focus
- Stopping down more will loose some resolution/contrast

I'm not saying that I always use f/8. Sometimes I open up for short DoF and sometimes close down for long DoF. But in general I use long focal lengths if I need short DoF and wide lenses for long DoF.

I have also found that I want my principial subject to be in focus rather than trying to optimize DoF and having everything slightly unsharp.

The main issue, that we cannot compare an FX lens (fullframe) at f/1.4 with an MF lens at f/2.8. Very few, possible none, of the 1.4 lenses are really sharp at full aperture but many f/2.8 lenses are quite OK. On the other hand, stopping down an MF lens to f/22 will loose a lot of contrast and resolution.

Best regards
Erik


You have a strange narrow concept of best lens performance, ignoring the sharpness of parts of the scene that are not at the focal plane; because with f/8 in all formats, a larger format like 4x5 or 8x10 is going to have heavy blurring of parts of the scene that are quite sharp in a smaller format.

What those tests at PhotoDo suggest to me is:
once
1) the lens is stopped down enough to have adequately low aberration effects (which can happen at far lower then f/8 these days; even down to about f/4 with some zooms, let alone primes),
and
2) the film or sensor resolution is as good as say TMAX-400
then a larger format will not give a sharper image overall.
At best, you can increase sharpness new the focal plane at the cost of decreasing it away from the focal plane, by accepting greater OOF blurring. (For example, by using equal f-stop and thus larger effective aperture size in a later format.)

Those tests indicate that at aperture ratios of f/5.6 or higher with a good 35mm prime lens, larger formats could not do better under the requirement of equal or greater DOF.

That still leaves dynamic range, tonal graduations, and such to be discussed, along with the film-vs-digital resolution debate.
Logged
Erik Kaffehr
 

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: Format sizes, MPs and printing large
« Reply #13 on: February 01, 2012, 07:31:13 am »

The main issue, that we cannot compare an FX lens (fullframe) at f/1.4 with an MF lens at f/2.8. Very few, possible none, of the 1.4 lenses are really sharp at full aperture but many f/2.8 lenses are quite OK. On the other hand, stopping down an MF lens to f/22 will loose a lot of contrast and resolution.

Arguably, only the center part of the image matters at those very wide apertures, and some 35mm lenses are extremely sharp in the center at f1.4.

http://www.dxomark.com/index.php/Lenses/Camera-Lens-Database/Nikon/Nikkor-AF-S-NIKKOR-85mm-f14G/(camera)/485

The only metric that increases significantly as the lens is stopped down is vignetting, for the rest the lens is close to optimal at f1.4.

Cheers,
Bernard
« Last Edit: February 01, 2012, 07:35:35 am by BernardLanguillier »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Format sizes, MPs and printing large
« Reply #14 on: February 01, 2012, 08:04:07 am »

Hi Bernard,

It seem that the Nikkor 85/1.4G is an excellent lens at all apertures.

Best regards
Erik

Arguably, only the center part of the image matters at those very wide apertures, and some 35mm lenses are extremely sharp in the center at f1.4.

http://www.dxomark.com/index.php/Lenses/Camera-Lens-Database/Nikon/Nikkor-AF-S-NIKKOR-85mm-f14G/(camera)/485

The only metric that increases significantly as the lens is stopped down is vignetting, for the rest the lens is close to optimal at f1.4.

Cheers,
Bernard

Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Format sizes, MPs and printing large
« Reply #15 on: February 01, 2012, 11:14:21 am »

Hi,

I didn't say that I always use f/8, I said I strive to use f/8. Why?

- Autofocus is inprecise, so is manual focus (except possibly LV). So I want to have some margin for improper focus
- Stopping down more will loose some resolution/contrast

Sometimes I open up for short DoF and sometimes close down for long DoF. But in general I use long focal lengths if I need short DoF and wide lenses for long DoF.

I have also found that I want my principial subject to be in focus rather than trying to optimize DoF and having everything slightly unsharp.
That still makes no sense to me, as usual with arguments about comparing different formats at equal f-stop (and/or equal ISO speed).

1. For focus precision, what matters in general is being some number of stops below the maximum aperture, the aperture that generates the OOF effects used by manual of automatic focus detection. So lenses of lower minimum f-stops will AF safely at proportionately lower f-stops: a fixed f/8 does not make sense, and the overall trend is for larger format lenses to have higher minimum f-stops, suggesting the need for a higher f-stop to get equally reliable focus.

2. In formats like 35mm or smaller, optimum sharpness as the focal planes these days achieved at below f/8: as I mentioned before, even moderately priced zoom lenses like my Olympus 14-54 and 50-200 seem to be optimal at about f/4 to f/5.6, so that f/8 will in your words "loose some resolution/contrast".

3. On stopping down losing some resolution and contrast: you are apparently saying exactly what I suggested: that sharpness and detail primarily matter to you only at one distance. Is it really the case that most of your photographs have all the interesting subject matter at roughly the same distance from the camera? Or that you can usually control DOF by choice of focal length at a preferred f-stop, which requires either changing your camera position to control FOV, or accepting a FOV dictated by DOF choice. Oh, and accepting perspective relationships dictated by camera position. (Sorry, let's not go there again!)

OK, maybe this is indeed your style, but I doubt that this is relevant to most photographers: I suspect that most of us adjust DOF with f-stop choices, and so for most of us, the two comparisons of greatest practical relevance are:
a) With compositions that are roughly equal in factors including DOF and FOV and perspective, so requiring equal camera position, and focal length and f-stop proportional to format size [as in that PhotoDo comparison.]
b) At optimal f-stop for the particular lens, in your sense of sharpness/resolution/contrast at the focal plane: this has a tendency of meaning at an f-stop that increases somewhat with format size, though in a far less rigid and predictable way than equal DOF comparisons.
« Last Edit: February 01, 2012, 03:29:39 pm by BJL »
Logged

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Format sizes, MPs and printing large
« Reply #16 on: February 01, 2012, 02:00:22 pm »

But I think that is the point. In order to produce "equivalent" images from different formats the smaller format must use lenses with larger F#s and shorter FLs which in some cases do not exist or can not be made at all. It seems that a larger format is more likely to produce an "equivalent" image to a smaller format than vice versa. BTW thanks for your input. I have been trying to get my hands around this subject since there so many formats available and so many claims of performance advantages. Just witness the latest roil here at LULA.
The second sentence of my reply should read: "In order to produce "equivalent" images from different formats the smaller format must use lenses with smaller F#s and shorter FLs which in some cases do not exist or can not be made at all. "

Sorry for the mistake.
Logged
A bird in the viewfinder...

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Format sizes, MPs and printing large
« Reply #17 on: February 01, 2012, 03:47:00 pm »

... "In order to produce "equivalent" images from different formats the smaller format must use lenses with smaller F#s and shorter FLs which in some cases do not exist or can not be made at all. "
Yes, that is one of the most clear-cut situations where a larger format can have a distinct advantage: where one desires or can use DOF so shallow that a smaller format struggles to achieve it without problems of optical aberration and such. Though strangely, with film camera formats from 36x24mm ["35mm"] up, the minimum usable f-stops of good prime lenses has tended to rise almost in proportion to format size, minimizing that potential difference between the mainstream film formats. To put it another way, users of medium and large format film cameras seem mostly to aim at levels of DOF that are in the "comfort zone" of good 35mm format prime lenses. Back in the film era, I am not sure that I ever heard or read a medium or large format film camera user boasting about a shallow DOF advantage over 35mm format; this issue has arisen mostly in the digital era, where the comparisons involve smaller formats and thus smaller f-stops, like 35mm vs APC-C vs 4/3" vs Nikon's 1" ...

Aside: once film or sensor resolution outperforms most lenses at most f-stops, as the PhotoDo tests suggest for B&W film, the other main advantage that I know of is dynamic range, particularly relevant to scenes with a wide subject brightness range.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Format sizes, MPs and printing large
« Reply #18 on: February 01, 2012, 04:41:06 pm »

Hi,

Not really an argument but I would like to make a couple of points.

1) I guess that 4/3 lenses in general may be better corrected than most other lenses. They needed to better corrected because the sensor is small.

2) I have looked at some 4/3 lenses and they seem to be very good overall. But many full frame lenses have weak corners at large apertures. See enclosed dump of Photozone's test of the lens I use mostly. Sony Zeiss 24-70/2.8 ZA. I'd argue it best overall (center and corners) around f/8. Center is very good even at large apertures. I also enclose a screen dump of the Zeiss 25/2.0 test on the Canon EOS 5DII. This lens clearly is best at f/4-f/5.6 but the resolution (at 50% MTF) fall of at f/8 is small.

3) I made a small demo of diffraction versus defocus here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1 the left column shows effect of diffraction while the rows show increasing defocus. Nothing revolutionary but it shows the effects in real life.

4) It seem that the really best MF lenses, like HR Digitars, may also reach peak performance at around f/5.6. Would I have a high end back and high end lenses I would not give up resolution for DoF, if I could avoid it.

I would also add another small observation. A couple of months ago I was shooting an autumn subject with yellow leaves against a green forrest background. I was using my 70-300/4.5-5.6 G lens at 90 mm. It was windy so I was considered about subject motion and didn't want to increase ISO because the camera I used (Sony Alpha 900) is a bit noisy above base ISO.

So I decided to switch to my Sony Alpha 55 SLT. On that camera I could use the 24-70/2.8 ZA at 60 mm instead. The Alpha 55 is an APS-C camera using the "sweet spot" of the lens so I could shoot at f/4 without consideration for soft edges and jack up ISO a bit. Both images were similar, but I feel that the Alpha 55 images have some advantage.

I may add that I'm not necessarily a Zeiss fanboy, just took these examples because the 24-70/2.8 ZA is a lens I own, and the 25/2.0 Zeiss is a new lens having very good reputation

Best regards
Erik




That still makes no sense to me, as usual with arguments about comparing different formats at equal f-stop (and/or equal ISO speed).

1. For focus precision, what matters in general is being some number of stops below the maximum aperture, the aperture that generates the OOF effects used by manual of automatic focus detection. So lenses of lower minimum f-stops will AF safely at proportionately lower f-stops: a fixed f/8 does not make sense, and the overall trend is for larger format lenses to have higher minimum f-stops, suggesting the need for a higher f-stop to get equally reliable focus.

2. In formats like 35mm or smaller, optimum sharpness as the focal planes these days achieved at below f/8: as I mentioned before, even moderately priced zoom lenses like my Olympus 14-54 and 50-200 seem to be optimal at about f/4 to f/5.6, so that f/8 will in your words "loose some resolution/contrast".

3. On stopping down losing some resolution and contrast: you are apparently saying exactly what I suggested: that sharpness and detail primarily matter to you only at one distance. Is it really the case that most of your photographs have all the interesting subject matter at roughly the same distance from the camera? Or that you can usually control DOF by choice of focal length at a preferred f-stop, which requires either changing your camera position to control FOV, or accepting a FOV dictated by DOF choice. Oh, and accepting perspective relationships dictated by camera position. (Sorry, let's not go there again!)

OK, maybe this is indeed your style, but I doubt that this is relevant to most photographers: I suspect that most of us adjust DOF with f-stop choices, and so for most of us, the two comparisons of greatest practical relevance are:
a) With compositions that are roughly equal in factors including DOF and FOV and perspective, so requiring equal camera position, and focal length and f-stop proportional to format size [as in that PhotoDo comparison.]
b) At optimal f-stop for the particular lens, in your sense of sharpness/resolution/contrast at the focal plane: this has a tendency of meaning at an f-stop that increases somewhat with format size, though in a far less rigid and predictable way than equal DOF comparisons.
Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Format sizes, MPs and printing large
« Reply #19 on: February 01, 2012, 05:03:58 pm »

... many full frame lenses have weak corners at large apertures. See enclosed dump of Photozone's test of the lens I use mostly. Sony Zeiss 24-70/2.8 ZA. I'd argue it best overall (center and corners) around f/8. Center is very good even at large apertures. I also enclose a screen dump of the Zeiss 25/2.0 test on the Canon EOS 5DII. This lens clearly is best at f/4-f/5.6 but the resolution (at 50% MTF) fall of at f/8 is small.
It does depend on a lot of other factors, but for one thing I was thinking in the spirit of the PhotoDo comparison, using primes rather than zooms, and wide angle lenses of course have more problems towards the corners than normal to telephoto ones. It would be interesting got dig up data on good normal primes for various systems and formats. [So why di dI cite zooms? Because that is all I have in my current 4/3 kit!]

And if one has a primary interest in sharpness of the subject focused on, maybe a bit of corner softness is less important that optimum performance near the center: the OOF affects are likely to overwhelm a bit of corner softness most of the time (how often is something in the extreme corner in completely sharp focus?) At center it seems to be a close context between f/4 and f/5.6 for both lenses, with f/8 in third place!
Logged
Pages: [1] 2   Go Up