DOF and Cropping/Enlargement continued...

[flakin stephen]

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

Absolutely agree, infinitely thin planes are meaningless in this context.

Is it time for a new definition of depth of field?

Perhaps there's a mathematician reading this who can be bothered to quantify the relationship between circle of confusion, pixel size (allowing for Bayer or not), image size and resolution of viewing medium, and resolution of the human eye (1 arc second is often assumed, but best eyes will resolve twice that).

All I ask is that the word 'Flakey' is included in the name of the formula!

cheers

Flakey.

 

[mike703]

In any case, here is a wrench in the works.

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

This doesn't change anything. The situation you are describing is one in which the resolution of the sensor is too low to record the necessary detail. So instead of a slightly blurred area being rendered clearly by measurements from several pixels, all that fine detail is lost.

You simply have a poor quality, low-resolution image whose sharpness and detail are limited by the recording medium - like an image from a 2MP camera made into a big print, or an image taken with ISO 1600 film that has grain the size of golf balls. None of that changes the definition of depth of field.

Of course you could make a small enough print - or stand far enough back - that you can't see the individual pixels / grains: at this point the lack of resolution wouldn't be a problem and the in-focus parts of the image would appear sharp to you; the smaller magnification of the image has increased the depth of field - as it always does.

Best wishes

 

[bigpigbig]

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

Absolutely agree, infinitely thin planes are meaningless in this context.

Is it time for a new definition of depth of field?

YES!

Perhaps there's a mathematician reading this who can be bothered to quantify the relationship between circle of confusion, pixel size (allowing for Bayer or not), image size and resolution of viewing medium, and resolution of the human eye (1 arc second is often assumed, but best eyes will resolve twice that).

All I ask is that the word 'Flakey' is included in the name of the formula!

cheers

Flakey.

--
flakey

Thank you for verifying that I am not crazy.

Sometimes I wonder.

 

[bigpigbig]

In any case, here is a wrench in the works.

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

This doesn't change anything. The situation you are describing is one in which the resolution of the sensor is too low to record the necessary detail. So instead of a slightly blurred area being rendered clearly by measurements from several pixels, all that fine detail is lost.

You simply have a poor quality, low-resolution image whose sharpness and detail are limited by the recording medium - like an image from a 2MP camera made into a big print, or an image taken with ISO 1600 film that has grain the size of golf balls. None of that changes the definition of depth of field.

Of course you could make a small enough print - or stand far enough back - that you can't see the individual pixels / grains: at this point the lack of resolution wouldn't be a problem and the in-focus parts of the image would appear sharp to you; the smaller magnification of the image has increased the depth of field - as it always does.

Best wishes

 

[mike703]

Like I said in my other reply enlarging images will multiply aspects of them that are already there, but it will not change them.

But it will. A clear example is that a photo will appear more noisy when viewed at greater enlargement.

lol

You're very good at not answering direct questions, aren't you? You still haven't replied (despite two requests) to the questions I asked you a while ago:

Loading…

www.dpreview.com

To summarise the questions:

(1) Why is it that every depth of field calculator out there shows that if you change the crop factor of the sensor, and leave all other paremeters the same, the DoF changes? I gave you a specific example in the above post.

If you think they are all wrong, please explain why.

(2) Here are some comments from the 'Cambridge in Colour' tutorial on depth of field that explains how DoF depends not just on the physical parameters used to make the image (lens, f-stop and so on), but also on viewing conditions

http://www.cambridgeincolour.com/tutorials/depth-of-field.htm

"When does the circle of confusion become perceptible to our eyes? An acceptably sharp circle of confusion is loosely defined as one which would go unnoticed when enlarged to a standard 8x10 inch print, and observed from a standard viewing distance of about 1 foot.

At this viewing distance and print size, camera manufacturers assume a circle of confusion is negligible if no larger than 0.01 inches (when enlarged). As a result, camera manufacturers use the 0.01 inch standard when providing lens depth of field markers....

A different maximum circle of confusion also applies for each print size and viewing distance combination."

You will find similar explanations all over the place if you search, on a wide range of authoritiative and respected web sites about photography. Lee Jay provided you with links to several in this post: mysteriously, all of them agree with one another. Could it be a conspiracy, or is it possible that they all say the same thing because they are correct?

Loading…

www.dpreview.com

If you think they are all wrong, please provide a link to an equally well-respected and authoritative source that contradicts Cambridge in Colour, Dofmaster and all the rest but confirms your point of view about DoF.

Happy seaerching and I look forward to your answers.

Best wishes

 

[mike703]

Mike,

I am not talking about some low resolution sensor.

I am talking about a 36mp FF sensor. I regularly shoot landscapes with the entire image having no CoC large enough to be recorded by the sensor.

But the CoC in an image depends on viewing conditions: it isn't fixed at the image capture stage. The size of the CoC is the diameter of the smallest dot that you can perceive to be slightly blurred rather than just a dimensionless point - and that depends how far away you look at it from.

From the Camrbdige in Colour web site:

http://www.cambridgeincolour.com/tutorials/depth-of-field.htm

"When does the circle of confusion become perceptible to our eyes? An acceptably sharp circle of confusion is loosely defined as one which would go unnoticed when enlarged to a standard 8x10 inch print, and observed from a standard viewing distance of about 1 foot. At this viewing distance and print size, camera manufacturers assume a circle of confusion is negligible if no larger than 0.01 inches (when enlarged)....

...A different maximum circle of confusion also applies for each print size and viewing distance combination."

So the statement that 'no CoC is large enough to be recorded by the sensor' doesn't make sense because you can't fix what the viewing conditions will be.

In the case you describe, if an image is blown up to the extent that every individual pixel on a 36MP sensor can be perceived as a circular blob by the viewer, then you have, precisely, enlarged the image so much that the limiting factor in the detail that can be seen by the viewer is the sensor resolution.

For someone admiring one of your 36 MP landscapes, the long edge of the image would be about 7400 pixels. The 'standard' definition of a CoC is that it is about 1/100 of an inch when viewed form about one foot away (see quotation above). This is based on 'normal' eyesight acuity.

That means that the hypothetical viewer of your landscape is looking at an image that is 7400 pixels x 0.01 inches per pixel = 74 inches across. So your print is 6 x 4 feet and it is being scrutinised from one foot away! Under those conditions then the amount of detail that the viewer can see, and the 'maximun sharpness' of the in-focus bits of the image, are indeed limited by the sensor resolution. If the image were viewed form a more sensible distance so that you could actually see the whole thing at once, the size of the piixels and the resolution limitations of the sensor would be way less than the viewer could perceive and DoF works as usual.

Best wishes

--
Mike
http://www.pentaxphotogallery.com/artists/mikeward

 

[Christof21]

In any case, here is a wrench in the works.

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

This doesn't change anything. The situation you are describing is one in which the resolution of the sensor is too low to record the necessary detail. So instead of a slightly blurred area being rendered clearly by measurements from several pixels, all that fine detail is lost.

You simply have a poor quality, low-resolution image whose sharpness and detail are limited by the recording medium - like an image from a 2MP camera made into a big print, or an image taken with ISO 1600 film that has grain the size of golf balls. None of that changes the definition of depth of field.

Of course you could make a small enough print - or stand far enough back - that you can't see the individual pixels / grains: at this point the lack of resolution wouldn't be a problem and the in-focus parts of the image would appear sharp to you; the smaller magnification of the image has increased the depth of field - as it always does.

Best wishes

 

[bigpigbig]

In any case, here is a wrench in the works.

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

This doesn't change anything. The situation you are describing is one in which the resolution of the sensor is too low to record the necessary detail. So instead of a slightly blurred area being rendered clearly by measurements from several pixels, all that fine detail is lost.

You simply have a poor quality, low-resolution image whose sharpness and detail are limited by the recording medium - like an image from a 2MP camera made into a big print, or an image taken with ISO 1600 film that has grain the size of golf balls. None of that changes the definition of depth of field.

Of course you could make a small enough print - or stand far enough back - that you can't see the individual pixels / grains: at this point the lack of resolution wouldn't be a problem and the in-focus parts of the image would appear sharp to you; the smaller magnification of the image has increased the depth of field - as it always does.

Best wishes

--
Mike
http://www.pentaxphotogallery.com/artists/mikeward

Mike,

I am not talking about some low resolution sensor.

I am talking about a 36mp FF sensor. I regularly shoot landscapes with the entire image having no CoC large enough to be recorded by the sensor.

The Coc used by a dof calculator is 0.03mm for FF which is 6 pixels large... Are you saying that you use Coc 6 times smaller ? Good luck with diffraction !!

For example, a 14mm lens shot at f5.6 and focused at about 4.5 meters will record everything from about 2.5 meters to infinity with no blur spot larger than 10 microns. This will not be diffraction limited.

 

[mike703]

Is it time for a new definition of depth of field?

YES!

NO!

The definition of depth of field was worked out in the first half of last century and has been used happily by photographers and cinemtaographers all of the world since then. It doesn't need 'redefining' just because someone doesn't understand it; and you couldn't 'redefine' it anyway, the laws of physics being what they are.

The links in this post might help:

Loading…

www.dpreview.com

 

[bigpigbig]

But the CoC in an image depends on viewing conditions: it isn't fixed at the image capture stage. The size of the CoC is the diameter of the smallest dot that you can perceive to be slightly blurred rather than just a dimensionless point - and that depends how far away you look at it from.

Mike,

I think you are limiting you definition of CoC to the printed size of a blur circle.

But CoC happens at two times. First, the size of the blur circle in the image projected onto the sensor. This is usually measured in micrometers (thousandths of a meter). An image from a FF sensor needs to be enlarged about 60 times when printed 8x10. What that means is that a blur circle at the capture stage (on the sensor) of 10 microns will print at 600 microns, quite a bit larger than the 254 microns you set as the minimum perceptible size of a blur circle at the printed stage of .01in

This is why a blur circle of 250 microns would be totally acceptable (and diffraction is a non-issue) when shooting large format 8x10 film if the output is to be an 8x10 print.

 

[bigpigbig]

Is it time for a new definition of depth of field?

YES!

NO!

The definition of depth of field was worked out in the first half of last century and has been used happily by photographers and cinemtaographers all of the world since then. It doesn't need 'redefining' just because someone doesn't understand it; and you couldn't 'redefine' it anyway, the laws of physics being what they are.

The links in this post might help:

http://www.dpreview.com/forums/post/54748307

--
Mike
http://www.pentaxphotogallery.com/artists/mikeward

Mike,

I think you are being closed minded.

Imaging has changed from film to digital sensors.

DoF has changed (slightly) as well.

Emphasis on "slightly".

 

[Bob A L]

Got a question I haven't seen in all these days of this discussion. With the theory of the c of c, and how different sensor sizes change things, does this not also have to do with the size of the photosites on the sensor. If so, and we need to factor in sensor size, wouldn't we also have to factor in size of the photosites that are recording the image projected by the lens? So when we add more pixels to the same size sensor would this then under these theories change dof also. I have a apsc camera with 2.1 mp resolution and one with 16 mp resolution so looks like if we have to tie sensor size into the dof factor, this should have affect also. It certainly takes more out of focus to spread the affect over 4 pixels in the 2.1 sensor than the 16. Yet in all these charts and explanations no one mentions this. I'm sure the proponents of this dof change will have lots of highly scientific explanations of why this doesn't matter, but might be interesting to hear them. I'm having a ball reading all this - better than the cartoons in the paper.

 

[Craig76]

Like I said in my other reply enlarging images will multiply aspects of them that are already there, but it will not change them.

But it will. A clear example is that a photo will appear more noisy when viewed at greater enlargement.

lol

I missed the answer to my question. Here's the question again:

In any event, are you really trying to say that if we took a photo of the same scene from the same position with the same focal point, framing, and aperture using a 12 MP D700 and a 36 MP D800, displayed the photos at the same size, and viewed from the same distance, that the photos would have a different DOF?

At your leisure, of course.

I thought that was a joke. Really, thats why I replied as I did and went to bed.

Here is a quote from a previous post in this same thread by me made earlier which I think answers it just fine. Also I don't think you read the original post. The one that started this thread. I have even put it in bold this time.

• My point, which should be vainly obvious even to the most simple, is that a camera with 50mm lens, focused on a stump 10 feet in front of it, with an aperture of f/4 will project an image with the same DOF characteristics. All other things being equal (pixel density) the sensors size has nothing to do with this. The size of the sensor will merely determine how much of the image is recorded.

"http://www.dpreview.com/forums/post/54750631"

Here is my original post quote

"I believe part of the problem here is a confusion between sensor size, and pixel density."

It really is funny because I was then told by Lee Jay that Pixel Density has nothing to do with it… now your arguing that it does and that I just don't get it….. lol

 

[Craig76]

Its time for you to start reading my answers/original post and thinking about what I said. (see reply I just made above about 1 minute ago to GreatBustards question)

 

[Craig76]

Im not answering qustiions anymore. Try READING my posts. Thank you.

Good day all.

 

[bigpigbig]

Got a question I haven't seen in all these days of this discussion. With the theory of the c of c, and how different sensor sizes change things, does this not also have to do with the size of the photosites on the sensor. If so, and we need to factor in sensor size, wouldn't we also have to factor in size of the photosites that are recording the image projected by the lens? So when we add more pixels to the same size sensor would this then under these theories change dof also. I have a apsc camera with 2.1 mp resolution and one with 16 mp resolution so looks like if we have to tie sensor size into the dof factor, this should have affect also. It certainly takes more out of focus to spread the affect over 4 pixels in the 2.1 sensor than the 16. Yet in all these charts and explanations no one mentions this. I'm sure the proponents of this dof change will have lots of highly scientific explanations of why this doesn't matter, but might be interesting to hear them. I'm having a ball reading all this - better than the cartoons in the paper.

Don't confuse two things here.

One: Size of the sensor relative to the enlargement. This is the most relevant.

Two: Resolution (pixel density). This only impacts the smallest blur circle capable of being recorded.

 

[Craig76]

There is another term for this, and this is why I kept saying it was a matter of semantics.

Pixel density has an affect, some people here are calling this DOF alteration or whatever. There is another term for this. Its called diffraction. Its when light bleeds from one microlens to another and creates a softer image.

 

[bigpigbig]

• My point, which should be vainly obvious even to the most simple, is that a camera with 50mm lens, focused on a stump 10 feet in front of it, with an aperture of f/4 will project an image with the same DOF characteristics. All other things being equal (pixel density) the sensors size has nothing to do with this. The size of the sensor will merely determine how much of the image is recorded.

This is wrong.

If you'd ever shot large format film, it would be obvious.

 

[bigpigbig]

There is another term for this, and this is why I kept saying it was a matter of semantics.

Pixel density has an affect, some people here are calling this DOF alteration or whatever. There is another term for this. Its called diffraction. Its when light bleeds from one microlens to another and creates a softer image.

Diffraction and DoF are completely unrelated.

 

[Christof21]

In any case, here is a wrench in the works.

If a blur circle on the sensor is as small as a photosensitive site (or maybe it is 4, I am still not clear on how a single pixel with tone and color is formed) then no amount of enlarging with change DoF. The entire image might blur from insufficient resolution, but a section of the image (or the entire image) will appear sharp regardless of the viewing conditions. In other words, the blur circles exist, but there is insufficient resolution to record them.

This doesn't change anything. The situation you are describing is one in which the resolution of the sensor is too low to record the necessary detail. So instead of a slightly blurred area being rendered clearly by measurements from several pixels, all that fine detail is lost.

You simply have a poor quality, low-resolution image whose sharpness and detail are limited by the recording medium - like an image from a 2MP camera made into a big print, or an image taken with ISO 1600 film that has grain the size of golf balls. None of that changes the definition of depth of field.

Of course you could make a small enough print - or stand far enough back - that you can't see the individual pixels / grains: at this point the lack of resolution wouldn't be a problem and the in-focus parts of the image would appear sharp to you; the smaller magnification of the image has increased the depth of field - as it always does.

Best wishes