Sensor size and DoF

[howiesmith]

... there's a lot more going on which influences the final perception of DoF as well as the manner in which it is calculated, specifically choice of CoC.

[a href=\"index.php?act=findpost&pid=80345\"][{POST_SNAPBACK}][/a]

My point exactly Ray.  A lot of factors go into DoF.  I never claimed otherwise.  Format simply is not one of those factors.

In your example, just how many variables did you change in addition to format and attribute to format?  I saw at least some pixel counts, and lens sharpness.

If you want to understand DoF, at least learn what affects it and how.

[AJSJones]

My point exactly Ray.  A lot of factors go into DoF.  I never claimed otherwise.  Format simply is not one of those factors.
[a href=\"index.php?act=findpost&pid=80373\"][{POST_SNAPBACK}][/a]

Ray, I do believe I understand this thread!

Howie seems very much into DoF but doesn't seem to care about the actual size of the piece of film he uses, only the dimensions of the image falling on it.  I presume film size matters to the extent that it needs to be sufficiently large for the print size he has in mind when capturing the image. (The above response derives, I think, from the fact that he often crops quite a bit, so the "original" format isn't relevant)

He therefore would not be concerned about the conventions of other people who have developed their own workflows and practices by associating them with particular format sizes and who crop (fairly) minimally or not at all (some insist that cropping of any kind after the exposure is "manipulation").  He uses CoC and enlargement factors directly to guide the design of his prints and therefore has no need of the DoF tables and calculators of others - or even the concept of "format" as used by other photographers in such discussions.  It is therefore not surprising that you and he would not communicate well about DoF and Format.

 It took me a while to understand, but I have put together a sequence of quotes that I think show that you both agree on what affects DoF and that enlargement factor and print size are critical.  (BTW I use a 6x loupe in the field with my 4x5 because I know I will not do much cropping post facto and my max print size of 24x30 represents a 6x enlargement - if it looks sufficiently sharp through the 6x I'm comfortable it will be OK on the print)

This method works for me and my style of shooting. It probably isn't for everyone, maybe no one else. But I do get anoyed when I'm told it doesn't work or nobody does that. But then I don't photograph for anyone else but me.
.....
I use 4x5 and 6x6 film and can measure image sizes directly from the ground glass. No need to know the pixel size or pitch because there aren't any. If there were, I don't need to know thatanyway. (I do need to know not to try to make CoC less than the pixel size though.) Then it is a simple matter of saying that the 1 inch high tree on the ground glass will be 20 inches on the planned print. A 20X enlargement. Not hard to do and not determined by format. Note that if I am using my 4x5 and think I am using my 6x6, I still get the same results - a 20X enlargement.
.....
Then I divide the CoC on the print by 20 to get the value to be used in planning the in camera work. The rest is simple math using the easily determined factors like lens focal length, f/stop, focus distance, etc. It is just a matter of taking time to plan what I'm doing.
.....
I usually crop before enlarging.
......
It is not necessary to base DoF on an average angular resolution of the human visual system. It is merely a convenience. Because DoF is personal, I can base it on anything I want, like my own visual system.

While Howie does photography his way, he does acknowledge a "common" situation where folks know the format and their desired print size and use the ratio to determine DoF considerations, with minimal (or no) cropping.  Implicitly, he acknowledges that if you change the format you would evaluate the DoF issues differently, because the ratio (of format to print)  would be different.

Format size is useful if you plan to make only full frame enlargements. Then just ratio up the format dimension to the print dimension.

 So, "conventional" DoF tables are based on format- and print-size-based CoC's and "conventional" viewing distances, but Howie disdains those conventions and makes DoF a more personal affair.  Nothing wrong with that, but it does generate a bit of semantic confusion.

Howie, I hope I've understood your point and not misrepresented it by this.  There are many who feel that the same image falling on two different sensors can't have different DoFs, because they have not thought beyond that capture to the viewing situation.  You clearly take into consideration the properties of the print, the degree of enlargement and the viewing distance as factors that affect the appearance of the image - just a little unconventionally!

[howiesmith]

AJSJones,

I think you have it.  DoF can be format dependant if you first assume it is and base all you printing assumptions around a particular format size.  (One can prove so many things true if you start with assuming they are true.)  Then, as I think you pointed out, the "image size" is the same as the "format size."  If that is the way you print and the way you will always wnat to print, then your assumption is fine.  I do not think it is correct to claim DoF is format dependant.

I also agree that if I make all the same assumptions made for on line DoF calculators or the stamps on the lens barrel, they will work fine, if I print and veiw to those assumptions.  If I want to be more "creative," I need to understand DoF.

Thanks for taking the time to understand this thread and your attempt to set things straight.

[AJSJones]

Howie,

I would submit that before digital the DoF that used to be talked about pretty commonly (i.e. for 35mm cameras) was based on the convention of making an 8x12 ish sized print from the whole frame and viewing it from around the image diagonal.  It made life easier with a common reference point for workflow and discussion and didn't require a preamble discussion of CoC and enlargement factors.

The issues that are raised by the advent of digital, and this thread in particular, are a) the one of cropping either by sensor design or post facto and its effect on DoF and  the discreteness of the captured array and its relationship to CoC.  You comment that it doesn't make sense to make it smaller than a pixel; Jonathan Wienke (and others) make a sensible case for using pixel size AS the CoC while if one goes strictly by "convention" it is based solely on geometry (see above).

I wonder if folks had similar discussions on DoF as the film emulsions got better and better and resolution went beyond the "conventional" CoC  

Andy

[howiesmith]

Howie,

I would submit that before digital the DoF that used to be talked about pretty commonly (i.e. for 35mm cameras) was based on the convention of making an 8x12 ish sized print from the whole frame and viewing it from around the image diagonal.  It made life easier with a common reference point for workflow and discussion and didn't require a preamble discussion of CoC and enlargement factors.

The issues that are raised by the advent of digital, and this thread in particular, are a) the one of cropping either by sensor design or post facto and its effect on DoF and  the discreteness of the captured array and its relationship to CoC.  You comment that it doesn't make sense to make it smaller than a pixel; Jonathan Wienke (and others) make a sensible case for using pixel size AS the CoC while if one goes strictly by "convention" it is based solely on geometry (see above).

I wonder if folks had similar discussions on DoF as the film emulsions got better and better and resolution went beyond the "conventional" CoC   

Andy
[a href=\"index.php?act=findpost&pid=80427\"][{POST_SNAPBACK}][/a]

First, I don't own or use zoom lenses.  Therefore, filling the frame isn't always an option for me.  Cropping is frequently required.  Even if it weren't, I'm not sure I would want to make a standard print to be viewed from a standatd distance every time.

Second, I recognize that with the wide spread use of auto-everything cameras.   Many photogrpahers don't know what f/stop or focus distance their camera has selected for them.  Many don't know how an exposure meter works.  And umteen point exposure calculation, who could reproduce that?

Third, using the pixel diameter for CoC means to me that the photographer is going to get maximum DoF for each image.  Using a fixed CoC also means the photogrpaher prints and views every image the same way.  If that is an assumption you can live with, use it.  Just know and understand what you are doing, and don't simply tell every body that is it.

Lastly, did anyone worry about a CoC smaller than film could resolve and opt to call that "the" CoC?  Not that I ever heard of.

But that's just me, and my old fashioned ideas.  Thanks again AJSJones for your time, effort and insight.

[gkramer]

The statistician John Tukey, commenting on a pre-web blog-equivalent on a statistical technique called "factor analyis," in which enthusaiasts of the technique sought to extract an entity called a "simple structure" from the data, remarked in effect that the notion of a "simple structure" was a concept more applicable to the objects doing the studying, than to the object being studied. The same might be said of the concept of a "circle of confusion," relative to this blog.

[howiesmith]

The statistician John Tukey, commenting on a pre-web blog-equivalent on a statistical technique called "factor analyis," in which enthusaiasts of the technique sought to extract an entity called a "simple structure" from the data, remarked in effect that the notion of a "simple structure" was a concept more applicable to the objects doing the studying, than to the object being studied. The same might be said of the concept of a "circle of confusion," relative to this blog.
[a href=\"index.php?act=findpost&pid=80437\"][{POST_SNAPBACK}][/a]

I think Tukey also said something to the effect that the price of calculation is decreasing rapidly, but the price of theorem proving is holding steady or increasing, so economics we spwmd spend more time on calculation.  So why make all those limiting assumptions to skip doing calculations?

[Ray]

In your example, just how many variables did you change in addition to format and attribute to format?  I saw at least some pixel counts, and lens sharpness.
[a href=\"index.php?act=findpost&pid=80373\"][{POST_SNAPBACK}][/a]

In my example I changed no variables. I didn't even change the effective format. There are differences built into the 2 cameras which are necessary design features flowing directly from the demands of different sensor sizes, ie. different formats. Smaller sensors generally have greater pixel density and require higher resolving lenses. These 2 factors alone will change the DoF outcome.

[Ray]

Ray, I do believe I understand this thread!

Howie seems very much into DoF but doesn't seem to care about the actual size of the piece of film he uses, only the dimensions of the image falling on it.  I presume film size matters to the extent that it needs to be sufficiently large for the print size he has in mind when capturing the image. (The above response derives, I think, from the fact that he often crops quite a bit, so the "original" format isn't relevant)

[a href=\"index.php?act=findpost&pid=80387\"][{POST_SNAPBACK}][/a]

Andy,
If you read the thread again I think you'll find I also understand this approach of Howie's and I have addressed the fallacy I see in his reasoning when he uses this approach as a basis for the statement that format is unattached to DoF considerations. I don't see how it is possible to measure the size of an image, which is a subset of the camera's format, without knowing the format. Howie has already stated that he needs to know the relationship between the image size on his viewfinder and the image size inside the camera. Knowing that relationship, which is 1:1 in the case of his 4x5 camera, he knows the format of his camera, even if he pretends not to know it.

The format of the camera is inextricably a part of those 2 essential factors, the ratio of viewfinder image to camera image and the measurement of that image in the viewfinder.

I don't see it as useful or insightful to claim that a 6x7cm back attached to a 4x5" format camera effectively turns that camera into a 6x7cm format camera. Of course it does. That's self evident. Whatever crop Howie makes from his 4x5 format camera, that becomes the new format of his camera for DoF purposes, just as it would if he attached a back to his camera holding a piece of film the exact dimensions of his crop.

[howiesmith]

Andy,
Knowing that relationship, which is 1:1 in the case of his 4x5 camera, he knows the format of his camera, even if he pretends not to know it.

I don't see it as useful or insightful to claim that a 6x7cm back attached to a 4x5" format camera effectively turns that camera into a 6x7cm format camera. Of course it does. That's self evident.

[a href=\"index.php?act=findpost&pid=80457\"][{POST_SNAPBACK}][/a]

Same camera.  Toyo Robo.  Image size on the ground glass is 1:1.  I am using 4x5 format film so its a 4x5.  I use 6x7 film so it's a 6x7.  So far, so good, ok Ray.  I know (but don't need to know) the format.

I use a ?x? back so it's an ?x? camera.  Ooops.  I don't know the format now.  I guess know that the image is still 1:1 won't be of any use.  I'm dead in the water until someone tells me ?x?, huh Ray.

Same is true with my 6x6.  The 1:1 doesn't change (but format does) when someone sneaks a 645, or 70mm (or whatever other formats are made for it) back on.  How do I do this now Ray?  I don't know the format.

I took a picture with my 4x5 camera.  I don't need all that image, so I take some scissors and make it into a 4x3 format.  Did I change the DoF?  No, not if I make the same largement of the remaining image.  I changed the format tough.  How'd that work Ray?

[howiesmith]

In my example I changed no variables. ... There are differences built into the 2 cameras ... .
[a href=\"index.php?act=findpost&pid=80454\"][{POST_SNAPBACK}][/a]

This could be part of the problem Ray.  You don't when you are changing a variable.

When I use my 4x5 and switch from Plus-X to Tri-X film, I am changing a variable.  When I process Tri-X plus 1 instead of normal, I am changing a variable.  A variable is something that can be changed.  "Changing a variable" is when I change something that can be changed.  Changing variables one at a time is when I change only one variable at a time and keep all the rest the same.  I need to be careful to recognize the variables so when I change one, I don't unwittingly change others.

[Ray]

I use a ?x? back so it's an ?x? camera.  Ooops.  I don't know the format now.  I guess know that the image is still 1:1 won't be of any use.  I'm dead in the water until someone tells me ?x?, huh Ray.

Same is true with my 6x6.  The 1:1 doesn't change (but format does) when someone sneaks a 645, or 70mm (or whatever other formats are made for it) back on.  How do I do this now Ray?  I don't know the format.
[a href=\"index.php?act=findpost&pid=80490\"][{POST_SNAPBACK}][/a]

Well, Howie, I can see you might have a problem there if there's no indication in the viewfinder as to what size back is attached to your camera. I've never used a large format camera, but if it's possible for someone to 'sneak' a 645 back on your 4x5, then I guess it's possible for you to forget that a 645 back just happens to be there and you might end up, either way, producing a DoF adjustment for the wrong format, thus demonstrating that DoF is format dependent.

In this situation, if you don't know your format, you not only get the DoF wrong but the composition also. To argue that the DoF is right and it's just the composition that's wrong is nonsense.

[howiesmith]

To argue that the DoF is right and it's just the composition that's wrong is nonsense.

[a href=\"index.php?act=findpost&pid=80500\"][{POST_SNAPBACK}][/a]

No Ray, to try to reason with you is nonsense.  I'm done.

[howiesmith]

BJL, way back on July 9, you posted:

"1) If you take photos from the same distance and compare DOF on equal sized prints viewed from an equal distance, the DOF will be the same for any combination of focal length f and f-stop N that gives the same effective aperture diameter, A=f/N. For example, f=50mm, f/1.4, f=70mm f/2 and f=100mm f/2.8 all have effective aperture diameter about A=35mm, and so all give about equal DOF in a "same print size, same distance" comparison.

"In other words, adjust f-stop in proportion to focal length to get equal DOF."

While true, when you changed format, you also changed the degree of enlargement to get your equal sized prints.  To compensate for that, you then changed focal length.  To compensate for that, you changed f/stop.  You changed much more than format.

When you crop the format, also crop the print so only format changes.  Change one and only one variable at a time to see clearly what the effect of changing that variable is.  If you do that, I think you will find DoF is not affected by format.

[AJSJones]

Ray, I still think it's largely semantic confusion!

Howie sees image capture size as a continuous variable, given that he crops frequently and maybe a significant amount.  IOW he does not see a 4x5 sheet of film as a 4x5 "fixed format", because he may only use part of what's on the ground glass (no zooms!)  If the image to be captured were the right size for a 6x7 roll back, but he only had brought 4x5 film holders to capture it with (and threw away the rest of the image later), he needs to consider DoF issues based on the size of the image he plans to capture and the print he envisions.  In this case, you or I might look up (or use a DoF calculator) the DoF data under the heading of 6x7 ; at least I would, even knowing that it is "really" a 4x5 format camera.  Following my earlier example, if I wanted to make a 24x30 print from only part of the image on my 4x5 ground glass (already had my longest lens and couldn't get closer), I would need to use a more powerful loupe to examine the critical parts of the image because it was going to be enlarged more than my "standard".

Many out there, including you and me, have "quantized" the image capture size continuum into a discrete set of different "formats" and mentally work with them in tabular form, and assume for some of the tables, a fixed size print viewed from a defined distance.  So, for a 35mm FF situation, we can be well guided by the DoF markings on a prime lens.  However, if we plan to make a larger print to be viewed from the fixed distance (the one used to calculate the DoF scale), we would need to make some adjustment if we wanted to exercise strict control over the DoF perceived when the image is viewed.  Similarly if we want to crop the captured image and make a print of the "standard" dimensions viewed from the "standard" distance, we'd similarly have to make an adjustment.   When I'm trying to photograph smallbirds with my 500+1.4x on my 5D and they're not filling the frame, I'll try to stop down to get more DoF than "necessary" because I hope to make a "standard" size print from an image that's been enlarged more than the "standard" amount. (Tough, since I'm trying to shorten shutterspeed too, but you get my point  )

It may be a little idiosyncratic (or old-fashioned as he calls it) but if degree of enlargement and viewing distance are correctly accounted for, along with the other variables, in his workflow, where's the fallacy?

Andy

[howiesmith]

Howie sees image capture size as a continuous variable, given that he crops frequently and maybe a significant amount.  ...

In this case, you or I might look up (or use a DoF calculator) the DoF data under the heading of 6x7 ; at least I would, even knowing that it is "really" a 4x5 format camera.  ), ...

It may be a little idiosyncratic (or old-fashioned as he calls it) but if degree of enlargement and viewing distance are correctly accounted for, along with the other variables, in his workflow, where's the fallacy?

Andy
[a href=\"index.php?act=findpost&pid=80540\"][{POST_SNAPBACK}][/a]

AJSJones,

Not quite.  "Image capture size" (format?) isn't a variable at all.  (I think that is where we started.)  It has no input to my application.  The only time I use "format" is when selecting a camera to use today, like what pants will I wear today?  The only requirement for format is whether the planned image will fit onto the film.  I seldom try to use my 6x6 with a 150mm lens to capture the width of the Grand Canyon.  (Actually, I don't think I have a camera and lens combo that will do that.)  Then, I look at the Grand Canyon, and move on.  I know I can't capture every moment, so I don't even try.

What do you do when the the cropped "negative" is, say 7.2x8.2?  Do you have a table for that or does your calculator ask for that input?  Or do you pick some cook book value that is "close enough?"  That is an odd format to say the least.

Your last statement is what I really see as the crux of this dicussion.  When using the markings on the lens barrel, tables, on-line calculators. CoC=pixel size, etc., the major factors are not usually taken into account correctly.  Many photographers just use the lens barrel DoF calculator, then make a 30x40 print.  Or worse, look at the image at 200% in Photoshop on their computer monitor viewed from 20".  Not even close to an 8x10 from 20".

I have wondered how photographers make a 3x5 post card to be viewed from 20" and a 16x20 coffee table book to be viewed from 20" and a 30x40 print to be hung on the wall and viewed from 5', all from the same negative.  And they expect the DoF to look the same in each one.

[AJSJones]

Howie,

By "image capture size" I was trying to convey  the variable area of the ground glass that you actually end up using in the final print (you use an example of a 7.2 x 8.2 - that is the part of the image you were wanting to capture, and the rest of the image was discarded - so it doesn't matter how big the film was that contained the 7.2 x 8.2) specifically to distinguish it from a fixed format.

As for the DoF for "weird" sizes, I go back to the workflow that many, possibly most, people follow: that is to frame the composition to fill the frame at the moment of capture by moving the camera or using a different focal length.  That is why the tables and scales are popular  

While I strive for everything to be in focus in my large format landscapes, so DoF is only a concern to get everything within it (my loupe does that on the spot), I think you're being too general about photographers with your last comment   Most people will view a postcard from closer than 20", so your scenario is tending towards the "viewed from a distance of the diagonal dimension of the image" where the DoF does indeed remain the same

Andy

[howiesmith]

OK Andy, I see your "captured image size.

OK, folks don't look at post cards for 20".  What then?  Do you still expect the same negative to produce the same DoF for post cards, coffee table books and large wall hangings?  I know the photographer has no control over his aydience, but do you honestly design a photo for post cards, books and large prints from a single image?  What do you do then the editor/publisher tells you you get a quarter page and not the accross the gutter double?  Make a note to the reader to move in?

I don't care what many or most people think.  The truth is the truth regardless of who or how many believe it.  DoF is not a popularity contest.  I have no problem with streamlining DoF by making reasonable assumptions.  The user should know what he is doing and why, before just jumping in with the lens barrel scale and then maybe wondering what went wrong.

I sometimes wonder how many folks actually plan an image, or just point and shoot.  Those who claim to take 10,000 images in a week or two, or even 100,00 in a year, don't spend much time planning.  It is my opinion that much of that is point, shoot, bracket like crazy, run on to the next spot, and edit later.  A style yes, but not mine.  I object to "that's how everyone does it."

[Ray]

Here's a shot I took using the HFI method of DoF calculation ('Howie Format Independent' method). I borrowed a 4x5 format camera with a 150mm lens. Based my DoF calculations on the size of the dancer on the ground glass plate and I think I got the background with just the right amount of blur, enough not to distract from the main subject but not so blurred as to be unrecognisable.

What a pity I was unaware the camera had been fitted with a 6x7cm back. The DoF is just fine, though   .

[attachment=1045:attachment]

[Craig Arnold]

Ray is right and Howie is wrong.

Howie simply isn't using the commonly accepted definition of DOF.

I think the notion of HFI DOF is at the crux of the matter. You can call a horse a donkey and then debate about how your donkey runs faster, and looks better and is taller than all the other donkeys. But the problem is that you have misunderstood the definition of donkey.

COC is dependent on film format.

The DOF calculations use COC as a variable.

When you change format you have changed your COC, and so you will change DOF.

Wiki entry - Circle of Confusion

Wiki entry - Depth of Field

Wiki extracts:

In film photography, the circle of confusion criterion is sometimes defined as the largest circle of blur on a film negative that will still be perceived by the human eye as a clean point when printed at 30 cm diagonal size and viewed from a normal viewing distance of 50 cm (and variations thereon).


Accepted values for circle of confusion based on d/1500
Film format Frame size[1] CoC

Small Format
APS-C[2] 22.5 mm x 15.0 mm 0.018 mm
35 mm 36 mm x 24 mm 0.029 mm

Medium Format
645 (6x4.5) 56 mm x 42 mm 0.047 mm
6x6 56 mm x 56 mm 0.053 mm
6x7 56 mm x 69 mm 0.059 mm
6x9 56 mm x 84 mm 0.067 mm
6x12 56 mm x 112 mm 0.083 mm
6x17 56 mm x 168 mm 0.12 mm

Large Format
4x5 102 mm x 127 mm 0.11 mm
5x7 127 mm x 178 mm 0.15 mm
8x10 203 mm x 254 mm 0.22 mm


Of course these values are for d/1500. If your lenses and film/sensor have higher resolution and you vary print sizes and viewing conditions etc you might want to change the value, but for whatever values you use you can draw up a similar chart for COC of different formats.



A list of my favourite Howie quotes from this thread:

(Taken somewhat out of context for humorous effect   )

One can prove so many things true if you start with assuming they are true.

A variable is something that can be changed. "Changing a variable" is when I change something that can be changed.

"Image capture size" (format?) isn't a variable at all.

I don't care what many or most people think. The truth is the truth regardless of who or how many believe it. DoF is not a popularity contest.

Because I don't own a pixel, I need not know anything about its size.

I have not yet run into a case where my CoC was smaller than the lens/film resolution. In fact, I don't even know what that is for my lenses and film selections.

All that I have been trying to say is that if you use two cameras of different formats but with the same focal length lens, same f/stop, same CoC and same focus distance, to make equal prints (same sizw focused image), DoF is the same for bothrints. The only variable changed is camera format. DoF does not change. The logical conclusion is DoF is independant of format.

It isn't that hard. I'm thinking you are either being obstenant or putting me on.

If I make an 8x10 print, then cut it in half, the characteristics of each half are not changed. Only the print format has changed. The DoF of each half is unchanged. It is so simple as to approach absurdly so.