DOF... where's it going?!

[Pookeyhead]

I never get this - why introduces arguments that are not part if the equation to a debate?

I said above.. it's academic... I was just saying that technically Hoppy is correct. That is NOT your problem here though so stop worrying about it.


You want to solve this? Rule out stuff via a process of elimination

Go and shoot a brick wall like I suggested, and find out if there's a camera/lens fault or not.

 

[Steve Smith]

Hoppy said that cropping changes DOF. IT does not. I haven't changed magnification

Cropping in itself doesn't change magnification. In fact there is no magnification until you view the image either on screen or as a print as you have nothing to reference it to.

But if you keep one image as it is and crop a second similar image then view them both at the same size, either on screen or printed, then the cropped image has greater magnification and will show less depth of field.


Steve.

 

[HoppyUK]

You won't win the DoF debate with Hoppy... it's a war of attrition and he's a veteran. I've tried and failed like many others...

There's always a possibility of equipment failure (sensor misalignment and/or optic mis-alignment) so you might as well do tests in controlled lighting using brick walls and focus charts and all that jazz.

But like others say, if that isn't the issue then you need to look at your technique and how you approach the situation in terms of exposure values, to give you the best shot at nailing it every time.

Haha But it's not an argument with me, it's with the physics.

There are some key factors about DoF that seem to be missing from other explanations.* Yes, as stated DoF is all about magnification (including the lens aperture bit, re the diameter) but depth of field does not exist as a measurable concept until the image is output (printed, displayed on screen) and viewed from a standard distance.

DoF calcs assume a standard size print viewed from a distance equal to the diagonal, eg a 10in print viewed from about 12in. When you take a larger FF image as here and crop it, that changes the magnification, and if you take a larger print and chop the edges off that creates a smaller print that is then naturally viewed from a closer distance (the diagonal) and that changes the magnification again, ie you're now looking at the same image as the uncropped version, but more closely and therefore able to detect more detail than previously and DoF is reduced.

That is what is happening here, and DoF is not as great as the OP estimates. (For more on this, search against my user name and be prepared to wade through a lot of debate.) The other factor is the image is quite seriously back-focused, and there can be no argument about that. Combine the two and you have the answer.

*Edit: crossed post with Steve - he's got it right!

 

[woof woof]

Please explain how? DOF is a balance between focal length, distance from subject and aperture. How could cropping an image change that? Put it this way - if I print a picture out 8x10 and then use a knife to cut off an inch all the way round, how has that changed the DOF?

Cropping does one thing and one thing alone - it reduces the physical size of a file. Any change in perception of The image doesn't change the data acquired

I don't think it's a DoF issue as such as I think you've slightly missed the focus or suffered subject or camera movement despite your insistence that the flash should prevent this... and I think you've made it more visible by making the subject and the fault bigger in the frame.

In the original uncropped image the slight misfocus/movement issue is probably hidden to a degree as the subject will be smaller in the frame but print the image larger, view it closer or crop it and print larger (effectively the same thing as printing the uncropped image larger and viewing it from a close distance) and the faults in the image are made easier to see.

That's all it is. IMVHO as nothing looks sharp to me. You've just made the faults in the image easier to see by effectively printing the image larger and viewing it closer. Anything, well almost anything, will look nice and sharp if it's small in the frame or it's a small image.

 

[Pookeyhead]

It's not movement... he's right, at that range, with the ambient lighting available and exposure levels, the flash will pretty much render anything sharp, even if there is still some underlying movement from the ambient lighting.... you'd still have the flash evidence to establish focus accuracy.


It's focused on the right hand guy.... or more accurately, slightly back focused on the right hand guy, and he's over estimated the DOF available.

 

[HoppyUK]

David yes, but this quote from your previous post is not right...

"print a 10x8, and then physically trim an inch from all the way around, then no, it will change nothing, as you've not magnified the image... you've just turned a 10x8 print into a 8x6 print. The image magnification will remain the same."

...as it ignores viewing distance that changes perceived magnification and is a key aspect of DoF calculation. It often gets overlooked and misunderstood because it's a given assumption that rarely gets spelled out.

On a more general note, the way we often view images today, at 100% on screen from close distance, makes a complete nonsense of the standard DoF formula which relates to film use when a 10in or A4 print was quite large!

 

[Pookeyhead]

See paragraph 1, post #39

 

[kendo1]

This is fun!

Anyway, we've still not seen the original or exif data.

DoF is a strange subject. It's not as simple as people seem to think it is, but for normal usage it is. And therein lies the misconceptions that people have.

Ken

 

[HoppyUK]

See paragraph 1, post #39

I'm referring to the quote taken above, and for all we know (unless I've missed it, I don't think we know how much of the image has been cropped?) the cropping could be a significant factor. If half the image area has been chopped away, DoF will be reduced from around 30cm to 20cm even before viewing distance etc has been taken into account.

 

[Pookeyhead]

I agree with you. However, just cutting the the edges off a print doesn't necessarily mean you will hold it any closer.

I am however, puzzled... I've been agreeing with you all along.

 

[HoppyUK]

This is fun!

Anyway, we've still not seen the original or exif data.

DoF is a strange subject. It's not as simple as people seem to think it is, but for normal usage it is. And therein lies the misconceptions that people have.

Ken

Yes

And questions like this come up quite a lot now, simply because generally we no longer output to relatively small size prints like we used to, but often pixel peep on screen at 100% and view from close distance.

Basically, DoF calcs today need to be interpreted differently but at least the DoF standard still has the great virtue of being universal reference. For the record, the whole shooting match is based on one single basic assumption - that the average human eye cannot detect anything smaller than 0.2mm in a 10in print when viewed from a distance equal to the diagonal, ie 12in. That is the ultimate circle of confusion size and everything else is calculated back from that.

 

[HoppyUK]

I agree with you. However, just cutting the the edges off a print doesn't necessarily mean you will hold it any closer.

The fundamental of DoF calcs assumes that smaller prints are viewed closer, and larger prints from further away.

Whether people actually do that or not is not part of the plan, but it mostly holds true. Eg, hand a few 6x4in prints to your gran and she'll reach for her reading glasses to take a closer look. Equally, a street poster looks sharp from across the road but just a splodge of dots from a couple of feet. This is where the 'viewing distance equal to the diagonal of the image' works so well as it's automatically self-adjusting for larger/smaller prints.

I am however, puzzled... I've been agreeing with you all along.

Sorry to be picky, but in that particular quote, it doesn't read that way.

 

[Pookeyhead]

The fundamental of DoF calcs assumes that smaller prints are viewed closer, and larger prints from further away.

Whether people actually do that or not is not part of the plan,

...and therefore academic. If you crop a A2 print down to A5, then sure... you'll be looking t it closer, but I don't think the level of crop being discussed here is even close to that. Would I hold a A5 print closer than a 8x10? I doubt it.

The problem here, as I see it, is poor focusing to begin with, coupled with a lack of understanding of how DOF works... and instead just relying on a DOF calculator without any real world experience or context.

 

[HoppyUK]

...and therefore academic. If you crop a A2 print down to A5, then sure... you'll be looking t it closer, but I don't think the level of crop being discussed here is even close to that. Would I hold a A5 print closer than a 8x10? I doubt it.

The problem here, as I see it, is poor focusing to begin with, coupled with a lack of understanding of how DOF works... and instead just relying on a DOF calculator without any real world experience or context.

That's my point, print size and viewing distance are not academic, but fundamental. They're the starting point for everything and when comparing hard numbers, make a significant difference and should be factored in.

The OP's question is where has the DoF gone, with an estimate that it should be 30cm. Well we now know where the DoF has gone, or rather where it never was in the first place, though I agree that it looks increasingly like simple miss-focusing is the main problem.

 

[Pookeyhead]

...but I do look at A5 print from the same distance as I do a 8x10. It's academic.... for me.

 

[HoppyUK]

...but I do look at A5 print from the same distance as I do a 8x10. It's academic.... for me.

That would be unusual, counter-intuitive, and wrong as far as DoF calculation goes.

 

[Pookeyhead]

That would be unusual, counter-intuitive, and wrong as far as DoF calculation goes.

That is illogical captain.... errr.. I mean...


I don't care I'd hold an A5 print as close as a 8x10 print. (shrug).

I'm not a computer or a machine... you can't dictate what I do based on what a calculation would deem intuitive or not.

 

[woof woof]

It's not movement...

It's focused on the right hand guy.... or more accurately, slightly back focused on the right hand guy, and he's over estimated the DOF available.

The problem here, as I see it, is poor focusing to begin with,

There you go then... missed focus

It's usual to focus on something... often an eye... and if you focus on something then it should still be pretty sharp even at higher zoom and looking at this image I just can't see what the point of focus was, unless it's as you said... back focused on the RH guy

I just think that arguing about DoF is pointless, it's a slight misfocus.

 

[Pookeyhead]

I just think that arguing about DoF is pointless, it's a slight misfocus.

...but caused by misunderstanding of DOF though. DOF is relevant here for the OP, or an uinderstanding of it is anyway, but I really don't get Hoppy's obsession with it in this context. We all agree what the OP's issue is. I even agree that Hoppy is correct in everything he's said so far... he still seems determined to dot all the Is and cross all the ts though

 

[TPFoster]

As has been said before, test your lens/sensor. Got this exact same problem after damaging the lens mount on a lens. The lens didn't lie flat against the camera body and so one side was always slightly out of focus. I would expect that you would know if you had dropped it or something but I suppose a slight knock could have occurred without you noticing.
Check your lens for back focus too.

 

[Davince]

Just got in from a shoot, hence no reply's, but I can't say this any more times, I'm not sure why I'm saying it again frankly, but as it's being bought up time and time again, I'll give it one last go...

I have simply cropped the image. The image IS NOT MAGNIFIED IN ANY SHAPE OR FORM. If I posted the whole image, it would simply have taken up more of the screen space. Hoppy, Cropping an image does not change the size its displayed on the screen, you know that, only changing the sampling of it does that. You do not lean in closer to the image to see it. You're simply looking at EXACTLY THE SAME INFORMATION on the screen, it's just got the remaining pixels filled with white, not colours. If you do lean in to the screen, then that is changing the DOF, not me cropping it. I don't lean in, I doubt you do either.

Only by resampling an image can you change the size it will appear on the screen. That has nothing to do with cropping. In exactly the same way, you do not change the DOF by chopping off the extraneous material off a printed picture either, you may move it closer or further away when viewing it, we already have example above of someone that doesnt, that is personal, and it is an act of moving the image. cropping the additional material off make NO affect on DOF, only what you do AFTER the image has been cropped affects it. If you bring it closer, bringing it close has changed the DOF, the perception of it. Cropping it, did nothing.

 

[HoppyUK]

You're not seeing the whole picture (pun intended) right through to the end.

Your estimate of DoF at 30cm is based on a set of assumptions used by all DoF calculators, that include the final image output size and viewing distance. If you change any aspect of that, including cropping in-camera (with an APS-C camera) or in post processing or by taking a print and chopping the edges off, you have changed the formula and the calculations are no longer valid.

These are facts, not a matter of opinion for debate. Your effective DoF is less than you think, by a factor of roughly one third/10cm if, for example, you have cropped out half of the image area. That is significant, but since we don't know how much you've cropped, it may be more or less relevant.

That, and the fact that it's quite a long way back-focused, explains your original question. Not sure why you seem to have a problem with the fact that I've provided the answer...

 

[Davince]

You're not seeing the whole picture (pun intended) right through to the end.

Your estimate of DoF at 30cm is based on a set of assumptions used by all DoF calculators, that include the final image output size and viewing distance. If you change any aspect of that, including cropping in-camera (with an APS-C camera) or in post processing or by taking a print and chopping the edges off, you have changed the formula and the calculations are no longer invalid.

These are facts, not a matter of opinion for debate. Your effective DoF is less than you think, by a factor of roughly one third/10cm if, for example, you have cropped out half of the image area. That is significant, but since we don't know how much you've cropped, it may be more or less relevant.

That, and the fact that it's quite a long way back-focused, explains your original question. Not sure why you seem to have a problem with the fact that I've provided the answer...

Quite simply because you're mistaking part of the picture not being visible for a change in aspect of the picture. Unfortunately no matter how many times you repeat the same mistake it wont make it correct so i'll simply cease trying to discuss it with you as i have neither time nor energy for your kind. The area shown over would appear absolutely, completely identically had I posted it with or without the area surrounding it. It is not a long way back focused, either, the sharpest part of the image is the RHS persons eyes.

For those who understand the importance of applying basic commonality throughout language for any hope of universal understanding... I've done a couple of tests and there is indeed a problem with the camera, DOF is well below calculators and expected and appears to be uneven across the plane. I suspect ken was correct but it's going in for repairs either way.

 

[HoppyUK]

I'm sorry you feel that way. Perhaps you should be equally rude and ungrateful to the other posters that tried to help with the same explanation.

Your camera must be seriously faulty to make it back focus like that. Or possibly, you got that bit wrong too.

 

[abdoujaparov]

It strikes me that while the crop hasn't changed the size of the portion of the image that's left, it has changed the brain's perception of what there is to see, and will presumably therefore instruct the eyes to focus on the image differently. If you give it less to work with, maybe it can pay more attention to what's left and see the loss of focus better?

 

[Ivan S]

Can you show us these test's? as I am curious as I am sure others will be.

All this talk of DOF and cropping and no one has posted any pics to show what they are discussing?

 

[Davince]

I'm sorry you feel that way. Perhaps you should be equally rude and ungrateful to the other posters that tried to help with the same explanation.

Why would I be rude to others? They aren't intentionally altering the parameters of the debate to make it fit with their flawed understanding.

I'm sorry if you find that rude, I find consistently adding misleading variables into a discussion and ignoring basic premises incredibly rude, C'est la vie

Can you show us these test's? as I am curious as I am sure others will be.

Sure. Took a photo of a grid on the wall. Camera on tripod, spirit level, triangulated for parallel:

Below are 100% crops from top left, mid left, top right, mid left, middle (focus point) etc.

Obviously the distance between edges to sensor is going to be greater than focal point to sensor as it's flat on a wall, but a quick bit of maths puts the difference in distances between centre focus point ~80mm (unless my trig is worse than I thought which it may well be lol), within dof at those settings, but more interesting is the fact that the bottom right corner, though slightly less sharp than the centre, is clearly significantly sharper than top left corner

Then split the grid in two, and hung one part 60mm (1/4 estimated dof at the distance) in front of the other.

100% crop of same bit.

What do you think? That doesn't look within acceptable to me?

What I found odd, as did the camera shop a moment ago when we were discussing it, was in the exif data - the camera was mounted on a tripod and didn't move between the shots at all, I just changed the focal length to 50mm to more closely match the shot I'd used as an example previously, yet exif puts the distance to subject on one as 1.5m, on the other, it puts distance to subject as 2m?!! not sure what that is about or if its symptomatic or what.

Was planning to do some more with different lenses but as I've got to send that lens back anyway, just going to get it all checked out now.

 

[Phil V]

Quite simply because you're mistaking part of the picture not being visible for a change in aspect of the picture. Unfortunately no matter how many times you repeat the same mistake it wont make it correct so i'll simply cease trying to discuss it with you as i have neither time nor energy for your kind. The area shown over would appear absolutely, completely identically had I posted it with or without the area surrounding it. It is not a long way back focused, either, the sharpest part of the image is the RHS persons eyes.

...

Oh how ignorance deceives.

Let's examine the facts objectively:

• You have posted a question in the 'talk basics' section of a forum.
• The question has arisen from your lack of understanding of a problem you've encountered
• Someone who has spent years getting paid to write technical articles on photography has given you an answer to your question (the correct answer)
• You choose to treat him (acknowledged expert in the field) like he's an idiot

Got to hand it to you - straight from 'how to make friends and influence people

There are a few recognised experts on this forum, too many to mention (I started) have a read of some other posts rather than just asking questions and you'll realise the talent pool available here.

 

[specialman]

Davince - is that grid straight or is there some barrel distortion going on? What lens is is and how close to wide open is it? There are obvious tolerances in lens performance and on many lenses, you'll struggle to get the same level of sharpness in the corners as you will in the centre. This can then exaggerated by the f-stop used, especially if you're nearing the maximum aperture (i.e. wide open).

Personally, I'd feel better knowing that I'd evaluated ALL aspects of focus performance, AF fine tuning (etc) before relying on the opinion of a camera shop which ultimately, wants your money.

 

[Phil V]

Obviously the distance between edges to sensor is going to be greater than focal point to sensor as it's flat on a wall, but a quick bit of maths puts the difference in distances between centre focus point ~80mm (unless my trig is worse than I thought which it may well be lol), within dof at those settings, but more interesting is the fact that the bottom right corner, though slightly less sharp than the centre, is clearly significantly sharper than top left corner
.

Obvious? No: You appear not to know that the focus distance is a flat plane? (I'll bet most of the people who were trying to help you earlier do)

So there's no difference in distance between the top left, centre, or bottom right. They all should be in focus, I hope that helps with your diagnosis.

 

[Davince]

Davince - is that grid straight or is there some barrel distortion going on? What lens is is and how close to wide open is it?

I was looking at that myself, unfortunately I chucked the grid so can't check it but as it only appeared to be the LHS that was distorted I assumed I just wasn't very careful with the pen lol.

It's a nikkor 24-70 f2.8 shot at 50mm f5.6

 

[HoppyUK]

Thank you Phil

To the OP, don't just take my word for how DoF calcs work, research it for yourself.

If you want to test the lens, shooting a close subject like that will skew the results. DoF runs in a plane parallel to the sensor, so the different distance from centre to edge shouldn't matter. However, most lenses commonly suffer from field curvature, especially fast ones at close distance, so may not show the edges of a flat 2D subject sharp all over, even though that wouldn't be noticeable in a normal 3D subject shot. (And BTW, the Exif data showing distance is only approximate.)

You need to test it at a greater distance, preferably replicating the original shooting conditions, with a 3D subject so you can see exactly where the focus plane falls in front/behind the focused point. Use lowest f/number for shallowest DoF.

Edit: suggest set up a line of three or four cerial boxes on the worktop, line parallel to the camera, boxes set at about 45 degree angle to camera. Focus on the middle of the central box, then note where the plane of sharpest focus falls on the outer boxes.

 

[Davince]

Oh how ignorance deceives.

Let's examine the facts objectively:

• You have posted a question in the 'talk basics' section of a forum.
• The question has arisen from your lack of understanding of a problem you've encountered
• Someone who has spent years getting paid to write technical articles on photography has given you an answer to your question (the correct answer)
• You choose to treat him (acknowledged expert in the field) like he's an idiot

Got to hand it to you - straight from 'how to make friends and influence people

There are a few recognised experts on this forum, too many to mention (I started) have a read of some other posts rather than just asking questions and you'll realise the talent pool available here.

I do realise the talent pool here or I wouldn't have joined. I posted in basic, because I thought there may have been something very basic I was missing. My response to Hoppy has to be based on his articulation skills in the thread, I don't know him.

I'm very happy to be proven wrong if it sorts out my bloomin issue!!! Maybe you can explain it to me then using this scenario:

I take a photograph. I print the photograph out at 10 inches wide. It is glued to a wall. You are sat on a chair, with your head in a strap assembly. You cannot move your head, or the distance it is from the photograph.

A third party comes in and sticks a piece of 2 inch wide paper over the edge of the photograph. The photograph now has 2 inches of it's width covered up. You have not moved, the photograph has not moved.

Has the depth of field changed, and if so, how?

 

[Davince]

Obvious? No: You appear not to know that the focus distance is a flat plane? (I'll bet most of the people who were trying to help you earlier do)

So there's no difference in distance between the top left, centre, or bottom right. They all should be in focus, I hope that helps with your diagnosis.

LOL that's ironic. I always thought it was, but an expert convinced me otherwise while we were shooting a large structure about a year ago.

 

[HoppyUK]

I do realise the talent pool here or I wouldn't have joined. I posted in basic, because I thought there may have been something very basic I was missing. My response to Hoppy has to be based on his articulation skills in the thread, I don't know him.

I'm very happy to be proven wrong if it sorts out my bloomin issue!!! Maybe you can explain it to me then using this scenario:

I take a photograph. I print the photograph out at 10 inches wide. It is glued to a wall. You are sat on a chair, with your head in a strap assembly. You cannot move your head, or the distance it is from the photograph.

A third party comes in and sticks a piece of 2 inch wide paper over the edge of the photograph. The photograph now has 2 inches of it's width covered up. You have not moved, the photograph has not moved.

Has the depth of field changed, and if so, how?

This is where you're going wrong. DoF calcs assume that when the size of a print is changed, the viewing distance is adjusted accordingly - to a distance equal to the diagonal. We tend to do this naturally, but whether you actually follow that practise or not is irrelevant here - it's what goes into the DoF calculator that matters. Bottom line is, when you crop an image you end up with less DoF than the full uncropped version.

NB See my edit on post #72 re how to test the lens.

 

[ernesto]

LOL that's ironic. I always thought it was, but an expert convinced me otherwise while we were shooting a large structure about a year ago.

That is probably the expert you should have been questioning as they were clearly wrong.

 

[Davince]

This is where you're going wrong. DoF calcs assume that when the size of a print is changed, the viewing distance is adjusted accordingly - to a distance equal to the diagonal. We tend to do this naturally, but whether you actually follow that practise or not is irrelevant here - it's what goes into the DoF calculator that matters. Bottom line is, when you crop an image you end up with less DoF than the full uncropped version.

And this is where our fundamental disagreement comes from. You say that cropping an image changes dof, I say that cropping an image and adjusting your POV relative to it changes dof, and I've not read anything to point to the contrary. This to me is far more relevant with digital

Answer me this: Is there any difference in DOF in these two images on your screen:

 

[Phil V]

I do realise the talent pool here or I wouldn't have joined. I posted in basic, because I thought there may have been something very basic I was missing. My response to Hoppy has to be based on his articulation skills in the thread, I don't know him.

I'm very happy to be proven wrong if it sorts out my bloomin issue!!! Maybe you can explain it to me then using this scenario:

I take a photograph. I print the photograph out at 10 inches wide. It is glued to a wall. You are sat on a chair, with your head in a strap assembly. You cannot move your head, or the distance it is from the photograph.

A third party comes in and sticks a piece of 2 inch wide paper over the edge of the photograph. The photograph now has 2 inches of it's width covered up. You have not moved, the photograph has not moved.

Has the depth of field changed, and if so, how?

It hasn't - but that's not what you'd do, or indeed by cropping and displaying what you have done.

If you're not strapped into the chair, you'll automatically move the image closer, therefore magnifying it. Magnification will show the lower contrast, so the DoF changes.

Look at any large print close up and you'll see that it's nowhere near as sharp as it appears from a great distance, it's exactly that principle at play.

And by showing a 6" wide crop of an image on screen you aren't showing the same thing as a 6" wide original. You've magnified the image and it's flaws.

 

[Davince]

It hasn't - but that's not what you'd do, or indeed by cropping and displaying what you have done.

If you're not strapped into the chair, you'll automatically move the image closer, therefore magnifying it. Magnification will show the lower contrast, so the DoF changes.

Look at any large print close up and you'll see that it's nowhere near as sharp as it appears from a great distance, it's exactly that principle at play.

And by showing a 6" wide crop of an image on screen you aren't showing the same thing as a 6" wide original. You've magnified the image and it's flaws.

Yep, exactly, completely agree with you Phil, just as I agreed with Hoppy - changing your pov of an image WILL change the dof.

But back to what I asked, or even the above example... Has the image changed DOF if you don't change your perspective to it? You know as well as I do, the answer is 'no'. It hasn't, it can't have. Cropping an image, WILL NOT change it's dof, only changing your POV relative to it, will. With the digital era, we do not tend to move closer to the screen, we sit at reading distance and view everything from that distance.

I'm sorry but I don't believe a single person can honestly say that in the example I posted above, they looked at the upper image, then moved closer to look at the lower image.

 

[Phil V]

...

Answer me this: Is there any difference in DOF in these two images on your screen:

No - the subject is the same size.

Are you suggesting that the subject in your earlier point was the same size? because that wasn't ever made clear. And is of course the fundamental point.

Wheel your chair back a couple of feet and watch the writing on the beer towel get sharper.