[pschlute]

Originally posted by pschlute Quote
just to be clear, when you say "is the camera cropping the full frame image", what it is NOT doing is capturing an image on the Full Frame sensor and resizing it.

That is incorrect. The camera does just that.

What i was trying to indicate to the OP was that the crop mode did not use the whole FF image and resample the image to 15.3 mp.

Originally posted by pschlute Quote
This is because you are capturing your image on an aps-c sized portion of the sensor.

You are contradicting yourself. APS-C is a crop, like it or not.

I think my answer here was quite clear. His image on a K-1 in crop mode is going to look just like an image taken on a 15.3mp aps-c sensor camera.

The other two parameters influencing the DOF (indeed, the parameters dominating the effect) are focal length and aperture

Different focal lengths in themselves do not have different DOF. Where two images are recorded using differing focal lengths, but the image size remains constant the DOF will be identical. It is aperture and camera to subject distance which are the two main parameters affecting DOF

[Ian Stuart Forsyth]

To me, Ian's methods are useful for pushing the envelope for what is photographically possible because sometimes reframing or repositioning

Sometimes it is for pushing the boundaries but it also helps to know what happens as the cameras resolution is increased and what effect that has on captured resolution, DOF, diffraction and why it is important to apply the correct DOF model for the resolution that camera can capture. See below for what happens with a model that is for a 24mp camera and a 50mp camera. If I use a model for that 24mp camera on that for 50mp camera I would be loosing the increased DOF that the 50mp give me over the 24mp camera for the same final image resolution.

[Ian Stuart Forsyth]

This obsession with "the same" always leads to me scratching my head. Why wouldn't you want to use the max. resolution you could achieve

There is no obsession with the same I would how ever want to use a model based on the resolution of the camera and not use a model that covers a single resolution. I would like to use the max resolution in all my shots but that max resolution is greatly affect by the DOF and diffraction I choose, this is why I want a model that I can use to get the best I can.

I am totally unclear however as to why that's useful information?

For landscape

My way of doing it is to make the best use of every system, which means i would never be trying to match the resolution of one sensor with another. Especially in a case like this where the DoF will be so much narrower on the 24 MP sensor. The reason you buy the larger sensor is more resolution for the same DoF.

How about more DOF for the same resolution? That would be my reason as to why I want more resolution from a camera. how about you? wouldn't you like that more DOF? I know I sure would

Why would you buy a more expensive larger system, then "match" the resolution of a lower MP system?

I answered this, with more mega pickles I gain more DOF than the system with less mega pickles and this is at any display size too boot.

I will use the 24mp camera verses the 50mp camera as that is what I started with so I will continue with them.
I visit a site where I take the photograph using the 24mp camera and the best all around setting to get everything with equally sharp across the frame is f/16 but it is not as sharp as I want it to be so I go back to f/8 but give up some of the DOF so that what is in the DOF is very sharp but at the cost of less DOF

Going back to the same site next year but with a 50mp camera would I use the same settings to shoot the same scene? What would you do?

My answer is found above. More mega pickles more DOF

[normhead]

ANd how is this different than just looking a lens chart, noting where the lens is sharpest and the rate at which it falls off, and shooting accordingly? Lens charts tell you that for most of your lenses they are going be sharpest at ƒ5.6 and fall off from there. So you shooting starting at ƒ5.6 and use whatever Aperture gives you the DoF you want? If you shoot for shallow DoF then you select on how shallow you want it to be.

I'm not seeing how the AP helps or even works. In your example you ended up using an exposure I could have used. I was hoping you had one that was something I wouldn't have anticipated. Going back with a higher resolution camera next year I'd shoot the same settings, and using a 3 step bracket. Then I'd pay careful attention to the images to see what worked best.

I answered this, with more mega pickles I gain more DOF than the system with less mega pickles and this is at any display size too boot.

Maybe you could point us to a reference for this. As far as I know DoF is defined pretty much by circles of confusion, which are going to be the same size regardless or the number of Mega Pixels. Diffraction and CA are usually reported by the pixel size, so as pixels get smaller CA and Diffraction become worse pixel peeping. Once you get over the size where CA crosses multiple pixels it does a lot more damage to image resolution. Same with diffraction.

By Photozone standards CA is measured by the amount of CA compared to pixel size of the camera tested. So .7 on a 24 MP camera is OK, but that's going to be over 1 pixel on a 50 MP camera. So CA and diffraction relative to pixel size is going to be worse as pixels get smaller. This is perhaps the biggest weakness of older lenses. It' also why I get cleaner images from my K-1 than I do with my K-3. The K-1 has bigger pixel sites and is therefore less prone to CA and diffraction. a 50 MP FF is going to have the same problems my K-3 sometimes does. And user results from the Canon 50 MP FF confirm that.

So this is not making any sense to me from a physics perspective.

I'm going to have to see some science, hopefully from a peer reviewed paper, to buy into this whole line of thinking.

As far as I know, DOF is function of the lens, not the sensor. The CoC do not change in size because of the number of pixels recording them. But I'm open to a scientific explanation if one is offered. Please not some web expert making up stuff to justify his opinio, based on his/her inability to correctly interpret the science, like the whole "total light" thing som many bought into.

---

Last edited by normhead; 12-14-2017 at 07:09 AM.

[photoptimist]

What's interesting about Ian's method is that it's an update to the DoF markings found on old helicoid-focus primes. Those etched lines and numbers worked in the days of film and single-format bodies but the markings need to be different for crop vs. FF and different for different resolution sensors (or intended output resolutions). In essence, the app is calculating the correct near-far DoF limits for a specific focal length, camera format, and camera/print resolution (or reverse calculating the required aperture for a near-far pair with the resulting worst-case blur).


True hyperfocal can't be done by aperture bracketing because the focus point must shift for each aperture to relocate the furtherest point of focus to infinity. And just setting the lens to it's sharpest point does not guarantee that both the background and foreground are in focus. Getting the highest possible resolution over the greatest possible near-far distance really requires either a lot of experience or an app that can calculate all the blur effects of DoF, diffraction, and device resolution.

[audiobomber]

Cambridge In Colour has a diffraction calculator that considers number of pixels along with sensor size.

For a 24mp Pentax APS-C sensor:
Diffraction May Become Visible f/5.4
Diffraction Limits Extinction Resolution f/6.8
Diffraction Limits Standard Grayscale Resolution f/8.1
OVERALL RANGE OF ONSET f/5.4 - f/8.1

For a K-1:
Diffraction May Become Visible f/7.4
Diffraction Limits Extinction Resolution f/9.2
Diffraction Limits Standard Grayscale Resolution f/11
OVERALL RANGE OF ONSET f/7.4 - f/11

Most will find that the f-stop given in the "diffraction limits extinction resolution" field tends to correlate well with the f-stop values where one first starts to see fine detail being softened. All other pages of this website therefore use this as the criterion for determining the diffraction-limited aperture.

Digital Camera Diffraction – Resolution, Color & Micro-Contrast

Almost all lenses start to drop in resolution somewhere between f/5.6 and f/11. Some cameras use no blur filter, others a strong blur filter. An app would need to consider the camera diffraction limit and MTF values for the lens by aperture, in order to provide the optimum hyperfocal distance. Do the apps contain these specific details? Accuracy of results would then depend on how close you get to the calculated focus distance.

I'm not at all confident that I can judge a distance of 12.2 feet without measuring. Then what about the final use for the photo? Circle of Confusion is dependent on approximations regarding resolution and print size. I can see that these scientific gymnastics would be worthwhile for some, but as an amateur hobbyist, I will continue to be more approximate and bracket my choices.

[photoptimist]

As far as I know, DOF is function of the lens, not the sensor. The CoC do not change in size because of the number of pixels recording them. But I'm open to a scientific explanation if one is offered. Please not some web expert making up stuff to justify his opinio, based on his/her inability to correctly interpret the science, like the whole "total light" thing som many bought into.

As long as the CoC is more than a few pixels, you are right that DoF blur at the pixel-peeping level isn't a function of sensor format or megapixels. At that level, the DoF near-far span is physical optical property of the focal length and magnification. And if we add diffraction effects or other lens resolution effects, the CoC and DoF are still only functions of the lens. It's only when we try to get a really small CoC (assuming the lens is good enough) that sensor resolution might add to the blur if the sensor is too coarse.

But if we want DoF at the printed image level (say a 30" x 20" print of beaver pond with sharp foreground cattails and sharp background pine trees), then the acceptable CoC (in microns) will be different for different formats. Confounding this is that the different formats will require different focal lengths (and different optical magnification settings) to create that specific print which also makes the DoF distance ranges different for a given aperture.

[derekkite]

Does sensor size change the hyperfocal distances? Yes. The hyperfocal distance for a given lens, sensor and distance from subject is a calculation based on what distance further away and closer in to the focus point will be perceived as sharp.

You must understand that it isn't describing a line where it is in focus or not. It is describing a point on the gradient from focus to out of focus where detail is perceived to be lost. The calculation is based on producing an image of a given size.

A real world example. I had a setup to take photos of bats in flight. I had a 50mm 70mm and 90mm lens on three bodies. From about 6' to 9' each would have an effective depth of field so that a bat taken in the 12" space would be sharp if cropped down substantially. The DOF calculations gave me just under 2' DOF for each lens, but in reality 12" was about the maximum because I cropped substantially.

So when you crop from full frame to apsc, in some circumstances where it really matters the DOF would be different. Knowing when it matters and when it doesn't is part of mastering your gear. The calculations are useful rules of thumb that you apply along with experience to any circumstance. The closer you are to the edge of the capability of your gear, the more this stuff will matter.

[normhead]

What's interesting about Ian's method is that it's an update to the DoF markings found on old helicoid-focus primes. Those etched lines and numbers worked in the days of film and single-format bodies but the markings need to be different for crop vs. FF and different for different resolution sensors (or intended output resolutions). In essence, the app is calculating the correct near-far DoF limits for a specific focal length, camera format, and camera/print resolution (or reverse calculating the required aperture for a near-far pair with the resulting worst-case blur).


True hyperfocal can't be done by aperture bracketing because the focus point must shift for each aperture to relocate the furtherest point of focus to infinity. And just setting the lens to it's sharpest point does not guarantee that both the background and foreground are in focus. Getting the highest possible resolution over the greatest possible near-far distance really requires either a lot of experience or an app that can calculate all the blur effects of DoF, diffraction, and device resolution.

Well ya, except you know where you want the edge of the frame to be. You set your focus point so it's approximate (in terns of distance) one third into the frame. At that point you can bracket your exposure say from F16 to ƒ5.6 (the sharpest ƒ stop for most consumer lenses, (although really good lenses like the FA 31 and FA 77 will be ƒ4. IN fact, the better the lens is the lower the sharpest ƒ-stop will be. Some Zies' are at their sharpest at F2.8) The thing is, as long as you shoot an ƒ16 to 5.6 or with FF 22 to 5.6 one of them will be right. You can select after. Unless, you are saying I also need to move my focal point as well.

Using a 31 ltd on APS-c you might as well just use ƒ11 as your starting point. You don't lose much in terms of resolution any way.


I'm still looking for visual confirmation that this even works. You just don't know if it makes any difference visually until you see it. There's lots of theoretical nonsense out there. Like people saying they can tell the difference in the image of a lens ar 2600 lw/ph compared to 2700 image and buying a slightly better lens for a higher price. Blind tests are wonderfull things.

How about that DA 18-55 compared to the 31.....(at 28mm)



For some uses @F8 the 18-55 might be the better lens. Mind you, I've always been convinced there was something wrong on this test the 31 on 16 MP.. At 10 MP the 31 tested off the charts, at 16 it was very average. I'm guessing sample variation, good vs. bad.

[angerdan]

The hyperfocal distance for a given lens, sensor and distance from subject is a calculation based on what distance further away and closer in to the focus point will be perceived as sharp.

Hyperfocal distance is calculated by focal length, f-number and circle of confusion limit.
So sensor size directly affects CoC and indirectly the hyperfocal distance.

[normhead]

Well, after checking this out and reading through Circle of confusion - Wikipedia my conclusion it appears to me this for most people is nonsense because...

Since the final-image size is not usually known at the time of taking a photograph, it is common to assume a standard size such as 25 cm width, along with a conventional final-image CoC of 0.2 mm, which is 1/1250 of the image width. Conventions in terms of the diagonal measure are also commonly used. The DoF computed using these conventions will need to be adjusted if the original image is cropped before enlarging to the final image size, or if the size and viewing assumptions are altered.

My assumption would be there are way to many unknowns to rely on this type of math in the field, the two most glaring being M magnification is usually unknown at the time of shooting as is cropping. Just those two unknowns, which must be known for the formula to be useful make this type of calculation at the time exposure untrustworthy. Whether or not it's more or less trustworthy than the on lens DoF guide is open to debate.

I wonder if the "Max resolution" setting on the camera menus does this type of math for you.

[BrianR]

Magnification is easily determined from focal length and focusing distance (both are on the side of your lens), and this is the info most apps will let you input (note the reduction of the DoF equation to depend only on magnification, aperture, and CoC becomes inaccurate at far distances). Cropping... I pretty much always know as I'm photographing to be 'none' or on the rare occasion to a specific aspect ratio other than the camera's native one. ymmv.

The DoF scales on the sides of lenses are a fine thing to work from, try them out and if they give you acceptable results run with it, or adjust as required. The key takeaway from this discussion is that people will have different use cases and tolerances so anyone hoping to rely on the pre-printed tables or scales should know what they're targeted at and be prepared to adjust, and also know that more precise calculations are available if they're interested.

A side takeaway is that there are many ways to approach photography, do what works for you and let other people do what works for them.

[normhead]

Magnification is easily determined from focal length and focusing distance (both are on the side of your lens), and this is the info most apps will let you input (note the reduction of the DoF equation to depend only on magnification, aperture, and CoC becomes inaccurate at far distances). Cropping... I pretty much always know as I'm photographing to be 'none' or on the rare occasion to a specific aspect ratio other than the camera's native one. ymmv.

The DoF scales on the sides of lenses are a fine thing to work from, try them out and if they give you acceptable results run with it, or adjust as required. The key takeaway from this discussion is that people will have different use cases and tolerances so anyone hoping to rely on the pre-printed tables or scales should know what they're targeted at and be prepared to adjust, and also know that more precise calculations are available if they're interested.

A side takeaway is that there are many ways to approach photography, do what works for you and let other people do what works for them.

With this app, i find myself wondering if it works for anyone. It's impossible looking at EXIF to tell if it's been used. Also, for myself, the idea of making the best of an image my camera can't ade uqtely cover is "it may be the best you could have done but, it's still not printable." One has to ask why one is even taking such images. But then that is determined by the individual as well. AN interesting digression form the topic of hyperlocal distances, but clearly off topic.

Really the only time you have to worry about diffraction is one you have gone over the lens' diffraction limit.ƒ 2.8 to ƒ11 you really don't have to worry that much about diffraction....hyperfocal away to your hearts content. Maybe at ƒ16 - ƒ22 you need to take diffraction into account. And at those-stops just back up and get your whole field of view in acceptable focus. I'm really not interested in "the best settings for an image with out of focus or diffraction limited areas. It's the photographer's job to know how to avoid those types if images.

Most of the time it's ƒ8 and be there with the K-1 or ƒ5.6 and bee there with APS-c. If you are thinking of printing really, sharp corner to corner is a thing.

[Na Horuk]

This thread really went far, despite the premise being such a simple question. A hypothesis that could be tested in one day by someone who has an FF and APSC camera and a measuring tape.

[normhead]

This thread really went far, despite the premise being such a simple question. A hypothesis that could be tested in one day by someone who has an FF and APSC camera and a measuring tape.

I've done that a number of times, It doesn't mean anyone listens.

I tested the DA 35 2.4 on APS-c against my FA 50 1.7 on FF. My biggest take away, is the DoF is about the same with the FF 1 stop more DoF , say ƒ8 instead of ƒ5.6 APS_c, but at those focal lengths there is enough difference in the geometry of the scene you probably want to know the lenses used and select the one that looks the way you desire.

Long story short I prefer the 50 on FF to the 35 on APS-c, and that has nothing to do with DoF. It's always more complicated than these discussions lead you to believe.

One stop more DoF on APS-c, but on the FF I can happily go to ƒ22 if needed. On APs-c my upper limit is usually ƒ16 although on rare occasion ƒ22 has worked. Sp I go to ƒ22 on APS-c but not as happily. But I've never found any method that reduces the need for ƒ-stop bracketing in critical situations. I'd often guess what I think should be the best setting, but often it's not, for reasons stairght numerical estimates can't factor in. Like an image that just happens to look better artistically at ƒ4 with some OOF areas, than at ƒ11 with corner to corner hyperlocal sharpness. There is just no formula for that.

Hyperfocal is as much about where to set you focal point more than trying to get a whole scene in focus.