The Falacy of FF Cameras having Flatter DOF

[xpatUSA]

Thanks for that link Ted. I've been looking for what look to be the correct equations for some time. COCDL or what ever he calls it - they have to be the same irrespective of format ???? The correct amount of blur can do some interesting things to some photographs due to the way we see.

Indeed. There's some good history about the choice of CoC here:

http://www.dicklyon.com/tech/Photogr...Field-Lyon.pdf

Pardon me if you've already seen this link - I do post it a lot.

It would be rather difficult to make sense of actual pixel resolution of a camera sensor because of debayering which take 4 pixels that contain 3 colours and then use what might be called a kludge to resolve the image for any type of RGB display. Some people have assumed a 2 pixel pitch for resolution. Also I think 3 or more but I would have to get a book out to check. It might be 4. The larger one seems to work on microscopes but probably because by accident it accounts for the flaws these usually have.

As a dedicated Foveon user, I don't have to worry about demosaicing too much.

There is a resolution limit via the lens and that for full frame and crop is circa 45 line pair per mm, if a great lens and used at it's best aperture. On a 50mm Milvus for instance that is at F4 at F1.4 it's about 32. Just as a comparison as it's interesting an Olymus 25mm F1.8's best it around 72 line pairs per mm at F4 and 65 at F1.8. Not so even across the field as the Zeiss as that drops 5 or so, 60 and 50 at the edge on Olympus. It's not the best lens resolution wise. I regularly get images that could be cropped considerably. I have on older full frame - yes it's very old same there but not really on more recent crop cameras. They will take some but no where near as much.

Probably what gave rise to Koren's liking for the Subjective Quality Factor. He didn't invent it but has gone on about it quite a bit. Bob Atkins, too.

I've read that Carl Zeiss came up with the idea of using MTF50 resolution figures. Not sure if this is true or not, The idea is that beyond that point the contrast levels aren't suitable for photographs however really good quality lenses probably do give out useful detail past that point if the contrast is high enough.

John
-

I'm not sure either. Trouble I find with MTF50 is that it is but a single point on the curve and tells you nothing about other frequencies.

[ajohnw]

Not that it matters in practice but I suddenly became curious about what the MTF50 figure should be at F4 if to the usual 1/4 wave error which is completely and utterly out of the question when there are lots of pieces of glass in a lens. Comes out at circa 135 lp/mm. Truly diffraction limited and it's going to be 160 or more.

When I did the sum the numbers seemed way to large - forgot to use metres. Fixed with this page as a check

http://www.imajtrek.com/new_page_32.htm

A good simple one for any one that is interested but the comment about designed to match pixel pitch isn't really correct now. They are but not in that was now on some cameras. It shows some plausible curves too.

Not sure if I have seen the pdf Ted.
No I hadn't. Glad to see he mentions Merklinger and even though he doesn't like it the few times I've used I can say that it works. I posted a link earlier. I went down to 3mm gravel rather close to the lens. He does mention that the distance aspects are odd but his aim is to get sharp images that include the far distance. Not tried that yet but will. He is more interested in the sharp as needed and what is possible at the far distance. It's a little bit different to the hyperfocal view. From memory he mentions a resolution miles away that a lens can't resolve but that is ok as it will have been set up to give sufficient resolution for the foreground which means that the camera could be focused some where else to yield better resolution there too but there is no need.

John
-

[xpatUSA]

Not that it matters in practice but I suddenly became curious about what the MTF50 figure should be at F4 if to the usual 1/4 wave error which is completely and utterly out of the question when there are lots of pieces of glass in a lens. Comes out at circa 135 lp/mm. Truly diffraction limited and it's going to be 160 or more.
John
-

Yep, my MTF spreadsheet says 180 cy/mm for MTF50 at f/4 (perfect lens) and 152 cy/mm for 1/4-wave rms.

[William W]

. . . I'm a convert. Ted may be the only one who agrees with me but IMO for viewing images prints can't compare with a high rez display. The only thing that comes close is a metal print with gallery style lighting.

I think it is that we are in the process of the change.

It is all about the saturation in the market place of the necessary technology for viewing.

About ten years ago, we invested about $18000 for (at the time) the best and most wonderful screen which we wall mounted in our client viewing room - the "wow" factor was second to none. , , but still a 36 x 24 print would cream that display (which was great because we'd sell the print).

The image of the swimmer is about four years old, That Client was viewing it on a TV which probably cost $4000~$5000 and a 30" print still creamed that.

But in (what?) a year's time with an 8K Screen and the smart technology inbuilt to it I agree the print will look tacky . . . then we have to wait for that saturation of enough 8K screens and of course the purchase price drops accordingly with that market saturation.

But that all applies to the critical eye and those who want to view "BIG" . . . for the mass market and mass consumers of photographic imagery that day happened years ago . . . only very few, perhaps a thousandth of one percent the consumers of photography visit forums like CiC . . , for the greater amount of Photographers and Viewers in the world, the compactness and 'legality' of the phone camera; the speed of the internet to transmit data from point to multi point; and the multiple galleries available to host that one image which is ultimately only ever viewed on a screen of whatever quality, is where Photography is now and has been for many years.

WW

[ajohnw]

Not that it matters in practice but I suddenly became curious about what the MTF50 figure should be at F4 if to the usual 1/4 wave error which is completely and utterly out of the question when there are lots of pieces of glass in a lens. Comes out at circa 135 lp/mm. Truly diffraction limited and it's going to be 160 or more.

When I did the sum the numbers seemed way to large - forgot to use metres. Fixed with this page as a check

http://www.imajtrek.com/new_page_32.htm

A good simple one for any one that is interested but the comment about designed to match pixel pitch isn't really correct now. They are but not in that was now on some cameras. It shows some plausible curves too.

Not sure if I have seen the pdf Ted.
No I hadn't. Glad to see he mentions Merklinger and even though he doesn't like it the few times I've used I can say that it works. I posted a link earlier. I went down to 3mm gravel rather close to the lens. He does mention that the distance aspects are odd but his aim is to get sharp images that include the far distance. Not tried that yet but will. He is more interested in the sharp as needed and what is possible at the far distance. It's a little bit different to the hyperfocal view. From memory he mentions a resolution miles away that a lens can't resolve but that is ok as it will have been set up to give sufficient resolution for the foreground which means that the camera could be focused some where else to yield better resolution there too but there is no need.

John
-

I worked from the cut off frequency Ted using this. Probably only of interest to you. From Smith who seems to be pretty reliable.



It's probably for sine rather than square wave. In fact it is as he gives another which is for square and it does come up with larger numbers. I'll bet that the manufacturers do that too for that reason.

John
-

[xpatUSA]

I worked from the cut off frequency Ted using this. Probably only of interest to you. From Smith who seems to be pretty reliable.



It's probably for sine rather than square wave. In fact it is as he gives another which is for square and it does come up with larger numbers. I'll bet that the manufacturers do that too for that reason.

John
-

Thanks for the diagram. On my spreadsheet for f/4, MTF50 and 1/4 wave error, the value of V/Vo was 0.34, FWIW.

In Smith's diagram, I don't know what n, U, or OPD are. I would agree with your comment re: sine waves, though.

[ajohnw]

When I searched for a new monitor last year I focused also on the resolution in dpi. What I have now is 108 dpi and that works for me. Dell p2715q has a resolution of 163dpi. But what is the use? The smaller the pixel size, the less pixel peeping. It looks wonderful, that picture you're viewing. But you're the only one. So if you're shooting for the web, it's worthless to me.

George

I feel into the pixel pitch trap a long time ago when the software wasn't ready for it. I had to set the resolution to obtain decently sized text. Much later displays came with a disc for a while which told the system what it needed to know about the display for scaling things like this. Now when the display is connected with the correct cord the system reads the data from the display.

What I am not clear about is just what happens to images as the pixel pitch changes so I have stuck to circa 100 dpi for a long time and just bought bigger and bigger monitors. One of my editors allows me to set the actual dpi of the monitor. If I change that the image size doesn't change however it's also set to show dot for dot. Disable that and the image gets considerably smaller. Maybe that is down to what is in the exif and some setting would cause it to display according to that.

As things stand I think we all generally view in what might be called a pixel for pixel or dot for dot view so changing the monitor dpi just alters the image size.

Hard to be sure though without 2 different dpi monitors.

John
-

[rpcrowe]

Actually using the same focal length lens at the same distance, the full frame camera will have a wider DOF than the crop camera.

When using a 50mm lens at f/5.6 and focused at 10 feet; the full frame DOF will be: 4.25 feet and the Canon crop camera DOF will be: 2.62 feet...

The reason that the full frame camera is said to have a more narrow DOF than the crop camera is that if you frame the scenes equally, you will need to use a longer focal length lens on the full frame camera or get closer to your subject.

As an example, you would need to use an 80mm lens on the full framer to give you the same framing as you would get with the 50mm lens on the cropper...

[ajohnw]

I noticed the "oddity" on the how much blur calculator some time ago. There is a very simple reason for it actually. One axis is % of sensor size, probably the diagonal. The other is distance. I'll post the link again to save people looking back in the posts

http://howmuchblur.com/#compare-1x-8...m-wide-subject

It's rather noticeable that the crop lens has less blur than the full frame when expressed this way. The curious fact is that this means that it does have more dof but on the other hand if the images that are eventually produced are the same size it needs to have a deeper dof other wise it would show more blur due to the increased magnification it needs for the final image size.

So if some one wants no blur or some degree of blur as far as the final image goes given that they are the same size there is no difference between the 2 cameras. The final image is the thing that people want so this is the correct way of looking at the effects. Not just saying crop has more dof for the same framing.

John
-

[Stagecoach]

Why can't we just say?

"If we use the same fixed FL lens at the same aperture on a crop camera and then a FF camera, frame our subject (lets say a head) the same, the FF image when viewed with the eyes has a lesser depth of sharpness than the image taken with the crop camera"

Because that's what I found when I did a test, or am I wrong

[ajohnw]

The crop camera is then working more "telephoto" than the full frame lens if the framing is the same so that will show up if the images are viewed at the same size. This shows though that over significant distances the crop will show more blur as would be expected as the focal length goes up.

http://howmuchblur.com/#compare-1x-8...m-wide-subject

I'd say there should be little difference over short distances from the focal plain if viewed at the same size. Things get difficult to compare when the same full frame lens is used on an aps crop camera as the lens aberrations will be lower towards the edge and corners of the shot as less of the image from the lens is being used and aberrations always increase out from the centre. Maybe be that the lens is better at the distance used on the crop as well. Or even for some reason the focus was better. The fact that the lens might function a little better on crop might even influence that. Also the fact that the AF has a stronger signal to work to due to the telephoto depth of field effect. One AF might be better than another.

It's all about magnification really and people seem to have forgotten the one that is used on the final image so being rude some yoyo on youtube etc just repeats what everybody else states. Usually with a smile on their face.

Only problem is that all of this is based on approximations but the same ones in all cases and they are known to work reasonably well even if not exact. Exact is more or less out of the question in practical terms.

John
-

[ajohnw]

Why can't we just say?

"If we use the same fixed FL lens at the same aperture on a crop camera and then a FF camera, frame our subject (lets say a head) the same, the FF image when viewed with the eyes has a lesser depth of sharpness than the image taken with the crop camera"

Because that's what I found when I did a test, or am I wrong

You would have to take the viewfinder magnification into account.

Who cares anyway most of us are more interested in the final image.

John
-

[Stagecoach]

You would have to take the viewfinder magnification into account.

Now I'm totally lost, why would anyone take the viewfinder magnification into account when comparing two images (taken and evaluated for test purposes) on my monitor screen or printed?

A 'test' would ensure equal framing, whatever the method used to achieve it

Who cares anyway most of us are more interested in the final image.

Exactly, and it is the final images that clearly shows what I quoted

[dem]

It is the viewfinder coverage that might be relevant here. It will be difficult to achieve the same framing if one camera has a VF with 92% coverage and the other with 98%.

Note the smiley faces.

[GrahamS]

I think things are getting over-complicated here.

Told you so.......

[ajohnw]

Why can't we just say?

"If we use the same fixed FL lens at the same aperture on a crop camera and then a FF camera, frame our subject (lets say a head) the same, the FF image when viewed with the eyes has a lesser depth of sharpness than the image taken with the crop camera"

Because that's what I found when I did a test, or am I wrong

Maybe I should have answered that a bit more clearly. I assume you have kept the framing the same so the crop is viewing with the same lens from a greater distance. The magnification the lens is giving has changed due to the difference in size of the sensors. The distance an image is behind a simple lens relates to it's focal length and the distance of the object. As the distance to the object is increased the change in distance behind it due to out of focus objects decreases which effectively means more depth of field. The simple formulae is

1/focal Lengh + 1 / object distance = 1 / image distance

As the object distance gets larger and larger it has less and less influence so dof increases. That formulae can easily be found on the web.

I just love some of the helpful votes that are cropping up. Myths always persist. What started me wondering was dof shooting macro using M 4/3 but as mentioned I had played with the "how much blur" before but didn't think about the implications. Complicated due to the lens I was using but a nagging feeling that I wasn't getting what the usual ideas suggest. Then I realised that I was finishing up with images sizes the same as everybody else via a smaller sensor. As it turns out on macro I may have to use a click or so slower aperture than some one like Geoff uses but that I suspect is largely down to using an electronic veiwfinder as the in focus view is a bit more iffy, What I couldn't do is use faster settings.

John
-

[Stagecoach]

It is the viewfinder coverage that might be relevant here. It will be difficult to achieve the same framing if one camera has a VF with 92% coverage and the other with 98%.

Note the smiley faces.

Exactly Dem, the viewfinder coverage not the magnification. Fortunately both my cameras are quoted as 100% coverage, and the results are near enough for me

[Stagecoach]

All this talk and not one image as an example, theories and fallacies be darned; let's see some comparisons.

OK, here you are John

I post these not to make any other point than to show the real world results of this simple practical test. Ignore the moving palm leaves, there was a bit of a breeze unfortunately.

Both images were taken using a D800. The camera on tripod, remained in the same place and the focal point was on the nut in the centre. Lens used was a Tamron 28-75mm 2.8.

No 1 image was taken with the D800 in FF mode at a FL of 75mm, f/3.2

No 2 image was taken with the D800 in 1.5x Crop mode at a FL of 48mm (indicated, this to achieve same framing), f/3.2

No 1


No 2


Nothing done to the two images other than downsize to 1400px wide using the same downsize sharpening.

[ajohnw]

How did you focus Grahame? I would have been inclined to use highly magnified live view on the car number plate manually just to make sure both were focused in the same place. Or same on a bold end or nut.

The lens could give corner and edge problems on full frame with less on crop especially at it's longest focal length and they often improve a lot at shorter focal length settings but I would have thought that wouldn't have any impact when a shot is reduced this much.

I've been trying to figure out how I could do a direct compare but format ratios would cause problems so can't use same framing. I believe the 80D also uses it's own size of sensor - not the same a the 7D. My 5D batteries may still charge up.

John
-

[ajohnw]

I don't think that the magnification relationship is correct. Here's why. Watch the car and other parts.



Argggg. it's a tough thing to compare.

John
-