Luminous Landscape Forum
Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: Bart_van_der_Wolf on August 25, 2013, 08:49:21 am
-
Hi folks,
As serious photographers, we are occasionally faced with shooting scenarios that we have less experience with compared to our usual subject choices. It can be helpful if we can test several setup scenarios before doing the actual shoot.
How large can we print, which lenses do we need, how much of the subject will be in critical focus? Those are just some of the questions we may want to be able and answer in advance, and go in prepared.
Many, if not all, of those questions can be answered with the use of so-called Depth of Field calculators. Several of those calculators are available on the internet, or as a dedicated application, also for smartphones and tablets. However, one of the biggest shortcomings of those calculators is that they require to input an important calculation parameter, the so-called Circle of Confusion diameter limit, but they do not offer any guidance as to what value is best to use.
In fact, most of them offer preset values based on the selected camera model, thus totally ignoring the intended output quality requirements and viewing conditions. Producing large format output requires different settings than those required for producing images for a webpage. The one-size-fits-all approach doesn't fit at all if quality is your concern.
"It's a DoF calculator, Jim, but not as we know it."
The tool I am offering is a bit different. Of course it uses the same fundamental algorithms to calculate the physical consequences of lens settings such as focus distance and aperture, but it does so with our intended shooting goals and available equipment in mind. It also applies some automatic refinements, e.g. when diffraction is a limiting factor.
Therefore those shooting and viewing goals need to be determined as a first step, and the available equipment as the second, with a few simple selections. Only then will we be able to do meaningful calculations. The results of those calculations may occasionally surprise us because the calculated results are goal specific, instead of generic. That's particularly useful in uncommon shooting situations.
The tool will try and stay close to the input that was given earlier, if possible, therefore the order in which the input changes are made matters. Some of the choices will alter prior input, which is inevitable because everything is part of a single possible solution.
The fastest way to achieve the result we are looking for is to follow the sequential settings from top to bottom, that's why I've numbered them, unless one of the later parameters is already known to be a given.
An example is the choice of aperture / f-number, e.g. if the optimum lens performance is more important than the creative DoF consequences (e.g. in case of reproductions), then by all means select it first, and the rest of the parameters will follow this setting, unless the tool is forced by your
subsequent input to adjust the aperture again.
Another example is with Photomacrography, where it is often more logical to set the magnification factor first, and then move the subject into the focus-range. Then by all means, set the focal length of your macro lens, and the magnification factor first.
To simplify this first setup, I've provided a number of generic shooting setups that you can choose from with a pull down menu. Those settings will get you in-the-ball-park.
Now, without much further ado, here is the link to the on-line web application:
The Depth of Field output quality planner (http://bvdwolf.home.xs4all.nl/main/foto/dofplan/dofplan.html)
As of this posting, it's a beta-testing version release, although I've tried to already eliminate as many potential issues as possible. It was tested on several web-browsers, and I've also built in some prevention for overly 'creative' input, such as negative distances. Under very rare input situations, I hope, one may run into an issue. In that case pick one of the generic scenarios to get a fresh starting situation. Refreshing the web page will ultimately solve any issue, but also requires to redo all input from the beginning.
I welcome user feedback, it will help to make it an even more useful tool.
Cheers,
Bart
-
Hi, Bart.....
Need this in an iPhone app :-)
-
Need this in an iPhone app :-)
Hi John,
I know that would make it easier to use on a phone, but it would also open a whole other can of worms on my side, getting it done. To name one, I don't have an iPhone ... Another is that the number of Android phones (which I do have) and Tablets have already overtaken the iPhone/iPad share in my neck of the woods, so I would have to prioritize on at least two platforms.
Anyway, you can just open the tool as it is on a browser tab, and leave that tab open. The tool doesn't phone home after it has loaded, so you can just carry it along, even without cellphone connection.
Cheers,
Bart
-
It's encouraging to see some original stuff appear here.
Edmund
Hi John,
I know that would make it easier to use on a phone, but it would also open a whole other can of worms on my side, getting it done. To name one, I don't have an iPhone ... Another is that the number of Android phones (which I do have) and Tablets have already overtaken the iPhone/iPad share in my neck of the woods, so I would have to prioritize on at least two platforms.
Anyway, you can just open the tool as it is on a browser tab, and leave that tab open. The tool doesn't phone home after it has loaded, so you can just carry it along, even without cellphone connection.
Cheers,
Bart
-
Great tool Bart, thanks so much for doing it. An android app would be great but I will see how it works from the phone as you mentioned it.
Thanks again for all that you contribute.
Alan
-
It's encouraging to see some original stuff appear here.
Hi Edmund,
I most certainly wanted to avoid the YADOFC (yet another Depth of Field calculator) syndrome, but even more than that solve some missing links in the use of the concept.
People seem to rarely stop and think about what value of COC to use, and why. Many phone apps even come with fixed values based on camera model, but that's pretty useless, IMHO of course.
I've also built in some informational feedback about diffraction effects, and ultimately even apply an adjustment to the COC when diffraction totally limits resolution (MTF=0). It's all automatic.
Cheers,
Bart
-
Great tool Bart, thanks so much for doing it. An android app would be great but I will see how it works from the phone as you mentioned it.
Thanks again for all that you contribute.
Hi Alan,
Thanks for the kind words.
I've set the application up with a number of exchangeable lens camera pre-selections, to reduce the need to punch in too many numbers. That should already make things a bit easier when using it as a browser tool. When people let me know that they'd like to have a specific model added, I'll consider that if time permits.
For exchangeable lens cameras and those with a fixed zoomlens that are in the database, I automatically guestimate a standard focal length close to the diagonal of the sensor dimensions, to get a starting point that's somewhat in-the-ballpark. I currently have no provision for camera models with a fixed focal length lens, so one would need to type in the specific (real) focal length for the specific camera.
Cheers,
Bart
-
Bart, for us metric illiterate (well, not illiterate, but need to convert/translate) Americans, would it be a big deal to have an option to show distances in feet?
Thanks...John
-
Bart, for us metric illiterate (well, not illiterate, but need to convert/translate) Americans, would it be a big deal to have an option to show distances in feet?
Hi John,
I'm working on it, but it requires a lot of additional code testing, and I didn't want to delay the availability due to only that reason. I may add it one input field at a time. It ties in with a question I've asked, and received useful answers about, here on LuLa, concerning which unit conversions one uses in daily life. Yards appeared to be very unpopular, except for some sports related uses.
Since the tool will correctly handle all situations from long lenses to macro, a lot of different distance units can be produced. The user interface should be able to display that in a compact space, and user input should not require too many leading zeroes to be entered.
The information I received will be used to also do some automatic switching, from inches - to feet - to miles when e.g. the Hyperfocal distance would otherwise start producing excessively large numbers in inches or small ones in feet. Likewise the tool will switch to smaller units when too many zeroes would be produced to depict very small distances or quantities. The algorithms and conversion factors are already in the codebase, now I need to adjust the user interface and add event triggers. As I said, still some work to be done.
Cheers,
Bart
P.S. While it's a bit of cheating, for now you can also use the section 3. copy lens settings button and switch front/rear and focus distance to non-metric equivalents there. When you first set the required units, and then click the copy button, things should display what you want.
-
Thank you Bart, the unknown/doubtful COC factor has always made these DOF-apps a bit of a rough guide for DOF. Really very helpful!
-
Thank you Bart, the unknown/doubtful COC factor has always made these DOF-apps a bit of a rough guide for DOF. Really very helpful!
Hi Peter,
You're welcome.
I obviously agree, COC values without being output referred make little sense.
Section 1) of my tool now also make it much easier to answer the question;
- How large can I print a file from camera 'x' ?
Before, the obvious answer was; How large is the paper your printer can handle?
But that didn't say anything about the quality one could expect.
Now we set our required quality level, viewing distance, and camera sensor parameters,
and out comes the maximum print size. Anything larger has lower quality, anything smaller
will have even higher quality. As simple as can be.
Cheers,
Bart
-
Hi John,
I know that would make it easier to use on a phone, but it would also open a whole other can of worms on my side, getting it done. To name one, I don't have an iPhone ... Another is that the number of Android phones (which I do have) and Tablets have already overtaken the iPhone/iPad share in my neck of the woods, so I would have to prioritize on at least two platforms.
Anyway, you can just open the tool as it is on a browser tab, and leave that tab open. The tool doesn't phone home after it has loaded, so you can just carry it along, even without cellphone connection.
Cheers,
Bart
You'd only need some changes to your existing webapp Bart.
I don't think native apps for this would be worth the time as it is all available on your existing webapp.
The problem with using it on an iPhone, Android, Windows or Blackberry phone is the downscaling, which makes interacting with the app difficult on a small screen (though much more useful for use in the field, where people carry phones, but are less likely to have a laptop or iPad, etc. handy).
Adjusting the existing layout with a modified stylesheet would make it more responsive and open up the possibility of having a single webapp that looks and works good on a phone through to a desktop.
No need for native apps (and as for screen emulation, it can all be done in a browser with Dev tools).
Regards,
Peter
-
The problem with using it on an iPhone, Android, Windows or Blackberry phone is the downscaling, which makes interacting with the app difficult on a small screen (though much more useful for use in the field, where people carry phones, but are less likely to have a laptop or iPad, etc. handy).
Adjusting the existing layout with a modified stylesheet would make it more responsive and open up the possibility of having a single webapp that looks and works good on a phone through to a desktop.
Hi Peter,
Good suggestions, thanks.
I thought you might be concerned about the no cell phone connection part, but apparently not so much. I'll have a look at how a different style sheet may help with the downscaling. On my Android phone rotating the phone 90 degrees to landscape orientation makes things readable at full width (I deliberately tried to keep the width down a bit for this reason, but I can still shave of a few pixels), and when I double tap on the screen it zooms in to full size. Maybe that functions differently on an iPhone, I'd have to check that.
Cheers,
Bart
-
Hi Bart -
A glance at this neat app leads me to wonder yet again 1) what sort of rez one should use at printing, and 2) how a current digital camera really stacks up compared to film. I wonder if anyone has published and compared the transfer functions of inkjet printing and enlarging?
Edmund
-
A glance at this neat app leads me to wonder yet again 1) what sort of rez one should use at printing, and 2) how a current digital camera really stacks up compared to film. I wonder if anyone has published and compared the transfer functions of inkjet printing and enlarging?
Hi Edmund,
1) The printer driver dictates the best choice for resampling to native printer resolution. Canon/HP printers require 300 or 600 PPI input for the requested output size, if anything else is received, the printer driver will resample for you, but that robs you of the opportunity to output sharpen at the actual print resolution. Epson printers require 360 or 720 PPI (the finest detail option must also be selected) input, otherwise it will do the resampling itself. One typically resamples up(!) to the nearest PPI
The printer driver resampling is usually not very good (usually bi-linear, or on rare occasions it's plain bi-cubic) and that lowers contrast of micro-detail, for reasons of speed. The lack of control over the output sharpening is my main concern, if one wants optimal quality and images with 'punch' rather than a dull veil.
Therefore my suggestion after the input of section 1.4), to resample the resulting image from the calculated PPI to the native printer PPI, and then apply output sharpening.
2) That is a bit harder to comment on, because film comes in many flavors. But as far as my tool is concerned, you can simply input the actual film image dimensions (e.g. 36x24mm, or 56x56mm, or 120x95mm, if scanned without mask(!)), and the scanned pixel dimensions, and the rest works as expected. It's just a different type of sensor.
The second difficulty with the direct sensor/film quality comparison is that they natively have a markedly different MTF response (with the scanning process adding some more variation). When looking for the limits of resolution (but that's only part of the complete story) a shot on film of my Star test chart with sinusoidal grating will tell the whole story, including the scanner influence (which will also show benefits of scanning at 6000 PPI or a bit more). And it will give you a slanted edge for SFR analysis of the entire imaging chain as well.
As far as the upsampling transfer functions themselves, I've posted some comparisons here (http://www.luminous-landscape.com/forum/index.php?topic=62609.msg505337#msg505337) on LuLa, with some transfer curves and log Fourier space views of the same crops were added a few posts after that.
Hope that helps.
Cheers,
Bart
-
Interesting. I never bothered with other DOF calculators, due to the pre-set CoC - CoC preferences being individual to shooter and intended photo format. In addition to being able to choose CoC, the neat thing here for me is the stack size calculator.
-
Interesting. I never bothered with other DOF calculators, due to the pre-set CoC - CoC preferences being individual to shooter and intended photo format. In addition to being able to choose CoC, the neat thing here for me is the stack size calculator.
Hi Nancy,
Of course the user should be enabled to choose his/her personal value for the COC parameter which lies at the basis of all Depth of Field calculations. However, a problem can also arise in selecting which COC value to use.
Rather than relying on some mythical value, the user can now let the tool assist in finding the optimal value, especially useful for unfamiliar shooting scenarios. It is still possible to manually override the optimal setting, but I rarely see a good reason because the tool incorporates all sorts of tweaks based on image magnification and/or extremely narrow aperture selections, much more accurately than I can.
The focus stacking part to me, is a natural extension for when we need to bend the laws of physics in our favour. It is also rather revealing, as it shows how many slices may be required to achieve the impossible. Even in the macro region, the slices do not have the same range, so users of automated focus rails are advised to use the narrower slice depth as a guideline. It may lead to some additional slices with overlap were a larger step distance could have been used, but it's better than finding gaps later.
All can be tested in a simulated run, without having to shoot and evaluate a single trial frame. That helps to get better results faster.
Cheers,
Bart
-
Personally, I vote for apps for the devices, because some of us shoot in places far, far from any network connection.
But, Bart would not have to do all that coding himself. It would be easy enough to farm this out to a different developer. If the app takes off, he gets some sales profit. If not, he loses nothing.
So, Bart, I may know an iPhone developer who could put this together for you on that platform. Send me a PM if you're interested.
-
Personally, I vote for apps for the devices, because some of us shoot in places far, far from any network connection.
But, Bart would not have to do all that coding himself. It would be easy enough to farm this out to a different developer. If the app takes off, he gets some sales profit. If not, he loses nothing.
So, Bart, I may know an iPhone developer who could put this together for you on that platform. Send me a PM if you're interested.
I can do it.
Edmund.
-
I will be a great tool, but why you don't use fixed dpi settings according to actual printers drivers settings - Epson use 360 or 180-240-360-380-720 steps (and 300 its wrong for Epson driver), HP and Canon use 300/600ppi.
Durst Theta use 254ppi, Lambda 400ppi and so on.
-
I will be a great tool, but why you don't use fixed dpi settings according to actual printers drivers settings - Epson use 360 or 180-240-360-380-720 steps (and 300 its wrong for Epson driver), HP and Canon use 300/600ppi.
Durst Theta use 254ppi, Lambda 400ppi and so on.
Hi,
I'm not sure if you are referring to the PPIs calculated by the tool, or an addition (say section 1.5) for the final output bit. The latter part should be taken care of by the application you print from.
Maybe I need to elaborate a bit more on the PPIs that the tool calculates, yes they are calculations. They are not fixed goals as if to match a certain output device (in fact they are the opposite), but rather the result of the angular resolution requirements to fulfill the required human visual acuity of our viewers' eyes.
When we enter the acuity level we need to satisfy, and the viewing distance, and the size (=magnification at a distance) of our magnified image that's captured on sensor, we can calculate the number of output lines or pixels it would require to achieve that goal. It then becomes necessary to have the output modality adapt to the produced data quality, and that most likely requires an additional resampling step, because the printer's resampling is not good enough.
I've attached an example of the final steps that are required, and used Photoshop CS6 but it is usually similar in most other applications.
Attachment 1 gives the screen captured result, based on the earlier input were an image is produced of 5616x3744 pixels and it should be given an output resolution of 291.1 PPI.
That is done in attachment 2. Note that the resampling checkbox is not ticked, so we only change the metadata of the file as it is. The unchecking of that box is the first thing that needs to be done, before changing any of the other settings! Then, at this stage, the PPI setting is nothing more than a label, the physical file size is not affected, the number of pixels remains the same. We can then either adjust the output size as we entered it in section 1.4 of the tool, or adjust the PPI setting. Both should give the same results, within rounding error precision.
Now on that basis, and in Photoshop we don't need to leave the dialog box, we now do tick the resample checkbox, and enter the PPI requirements for the output modality we want to use. That is shown in attachment 3. Now new pixels will be created, and the file size will grow when we click OK.
In the example I only resampled up to the nearest 360 PPI setting to e.g. print on an Epson printer. Personally I would have gone to 720 PPI, because with deconvolution sharpening I can recreate even more accurate sharpening, and even exaggerate it a bit, because artifacts at that level will be almost impossible to see. I could also add some noise, it will be to small to really see clearly, to give too smooth areas a bit of dithering structure, to avoid posterization in output. I would actually not have used Photoshop's Bicubic Smoother resampling at all, but Photozoom Pro, because that does a better job, but that's for another discussion thread.
Cheers,
Bart
-
I think it would be interesting with such tools integrated with the image capture device itself (just thinking loudly here, not making any demands to you, Bart).
Why can't something like the Nikon D800 run simple Android-like "apps" that allows you to add this kind of functionality and control e.g. aperture directly from the app, possibly integrating liveview preview?
My Sony RX100M2 allows Android/iOS remote viewfinder control across WiFi. The same functionality seems to be offered by the strange DSC-QX100 / DSC-QX10. I wonder how closed that API is. Being able to preview comfortably on something like a retina iPad with great added insight into how camera settings affects the final print could really be something.
-h
-
I think it would be interesting with such tools integrated with the image capture device itself (just thinking loudly here, not making any demands to you, Bart).
Hi,
Well, they have not contacted me, yet ;) . First come, first considered.
I think camera makers would only show interest in adding complexity (for them) if they estimate it would sell more cameras.
Cheers,
Bart
-
I think camera makers would only show interest in adding complexity (for them) if they estimate it would sell more cameras.
Which is why I think that "photo-oriented" people will have to hope for (ugh) facebook integration, angry-birds and appstore. Only when camera makers choose to include such flexibility (to satisfy the masses), can the photo-geeks hope to be able to piggy-back on the APIs/openness to include narrow photo-oriented stuff. I predict a large upswing in photographic creativity as a result.
Samsung is my best bet.
-h
-
As requested, I've just added the initial support for Non-SI units of distance to the Camera-lens settings sections 2.1) - 2.1.2).
A number of associated fields (e.g. 2.1.3) and the Hyperfocal distance) are updated as well, depending on the distance unit choices. The Hyperfocal distance is reported in the same type of units as the Focus distance, because one might want to take that clue as a new focus distance. The type of units for the Front and/or Rear DOF distance will determine the scaled units for the total DOF range at 2.1.3). That Total DOF range will be scaled to relatively meaningful numbers and units, to allow an accurate value in a few digits without too many zeros before or after the decimal point.
There are more distance fields that need to be enabled for Non-SI units, but I wanted to already share this for those who can't intuitively relate to metric units for the most important distances. I also added the relatively unpopular yards, for those who want to try their skills at shooting sports.
I need to do some more testing, but it looks like I didn't break anything vital by adding this.
Later I may add a simple choice to switch all units with a single click, but there may be specific requirements for more detailed unit choices, hence the current high level of control (I also added some larger distance units - km and mi - to the Rear DOF selection, to allow input with fewer digits in some cases).
Cheers,
Bart
-
I think it would be interesting with such tools integrated with the image capture device itself (just thinking loudly here, not making any demands to you, Bart).
Why can't something like the Nikon D800 run simple Android-like "apps" that allows you to add this kind of functionality and control e.g. aperture directly from the app, possibly integrating liveview preview?
My Sony RX100M2 allows Android/iOS remote viewfinder control across WiFi. The same functionality seems to be offered by the strange DSC-QX100 / DSC-QX10. I wonder how closed that API is. Being able to preview comfortably on something like a retina iPad with great added insight into how camera settings affects the final print could really be something.
-h
There are apps that do this now. Helicon Remote has a DOF calculator built in that allows you to change the CoC size to whatever you want and outputs DOF based on that. It doesn't provide a preview; however. That would simply take too much processing.
Typical DOF calculators work the way they do because they're based on a standard. That standard is a CoC of .01 or smaller being considered in focus in an 8x10 print viewed at a certain distance. I forget what the viewing distance standard is off the top of my head. It's a reasonably simple matter to redo the calculation for other print sizes. You want to make a print 4x the size, that enlarges all the CoC by 4x. What does that do do DOF?
Here's the question I have. Standard DOF calculations have served photographers well for decades. Is the difference in the digital age so great that it has a marked difference in a print? I'm asking not to be combative but rather to try and understand. I know some on here think every question I ask is to be combative. ::)
-
I've expanded the support for non-SI units to include the Object dimension settings.
Additionally I've changed the handling of the (optionally chosen) resulting optimized Focal length value. It formerly defaulted to nice rounded (to 5mm) focal lengths, but that also introduced slightly less accurate results for the depending parameter values. Now, one can double-click on the Focal Length field, and it will do the rounding to the nearest 5mm, and additionally recalculate the entire model where it depends on the focal length.
The Object dimension settings allow to automatically set all other related parameters, and can be used if the object dimensions are already known in advance. This can be used with reproduction of e.g. paintings/drawings, or when the main subject has a known size in the plane of optimal focus.
Do note that it only applies to the subject dimensions in the focus plane (that's why that is usually set first, or is optimized for), and that the use of a different focal length will produce a different rendering (size and blur) of the foreground and the background features. The photographer must still make creative choices, but won't have to worry about fitting the main subject in the frame.
This still leaves the 2.7) Total Depth of Field (and it's dependents) to be enabled for non-SI unit input, but that's a tricky parameter to change anyway, because it impacts the original quality goal for the required image size. It is generally not advised to change it there, but e.g. change the Aperture if more or less DOF is needed.
The value of the total DOF is reported at 2.1.3) anyway, and there it uses the distance units that were set for the Front or Rear DOF distance.
Cheers,
Bart
-
Typical DOF calculators work the way they do because they're based on a standard. That standard is a CoC of .01 or smaller being considered in focus in an 8x10 print viewed at a certain distance. I forget what the viewing distance standard is off the top of my head.
Hi Bob,
Some use 12 inches (30.48 cm) viewing distance, others (e.g. Zeiss) assume 25 cm (~10 inches) and also use a lower quality (larger COC, 2 minutes of arc instead of 1 minute). Of course an 8x10 inch (contact) print requires a different (simply scaled) parameter for other distances, and also may not fit everybody's requirement for quality. Sometimes lower quality is permissible, but at other occasions uncompromised quality is required.
A higher quality standard also allows to use a better quality sharpening, because more pixels will be available, and that enhances the almost tactile aspect (cardboard versus leather) of how material structure is rendered in the print. This is very important in product shots, less so in portraits (unless the ultra-realistic style is used).
It's a reasonably simple matter to redo the calculation for other print sizes. You want to make a print 4x the size, that enlarges all the CoC by 4x. What does that do do DOF?
DOF is also dependent on Focal length, which renders foreground and background detail at different sizes due to angle of view. So there is also a 'quality' aspect to DOF, even before considering Bokeh.
Here's the question I have. Standard DOF calculations have served photographers well for decades. Is the difference in the digital age so great that it has a marked difference in a print? I'm asking not to be combative but rather to try and understand. I know some on here think every question I ask is to be combative. ::)
Well, I'm not so convinced that the DOF calculations have served photographers that well. Maybe well enough, given a situation where one already had some other experience with, but I doubt they served them well in very unfamiliar circumstances.
In addition, digital imaging allows the use of images for rather different purposes. Never before has is been so easy to produce poster sized output, sometimes even on one's own printer. Enters the quality question. Being able to do something doesn't necessarily mean it is gong to be of adequate/satisfying/necessary quality ...
For that purpose I think a better, or easier to use, predictive model which includes the specific limitations of the recording medium (such as sensel pitch limitations), is helpful.
Cheers,
Bart
-
Hi Bob,
Some use 12 inches (30.48 cm) viewing distance, others (e.g. Zeiss) assume 25 cm (~10 inches) and also use a lower quality (larger COC). Of course an 8x10 inch (contact) print requires a different (simply scaled) parameter for other distances, and also may not fit everybody's requirement for quality. Sometimes lower quality is permissible, but at other occasions uncompromised quality is required.
Yes, those are the numbers I recall now. I think you've hit on it with that bolded bit, Bart. It's also quite subjective. Reminds me of the issue with temperature ratings for sleeping bags. Those are based on what the 'average person' should be able to achieve. Thing is, there is really no such thing as the average person. We're all individuals.
A higher quality standard also allows to use a better quality sharpening, because more pixels will be available, and that enhances the almost tactile aspect (cardboard versus leather) of how material structure is rendered in the print. This is very important in product shots, less so in portraits (unless the ultra-realistic style is used).
Don't disagree. I'm wondering how big the difference is.
DOF is also dependent on Focal length, which renders foreground and background detail at different sizes due to angle of view. So there is also a 'quality' aspect to DOF, even before considering Bokeh.
True. And the standard calculators take that into consideration, along with distance to subject which is the other component of magnification.
Well, I'm not so convinced that the DOF calculations have served photographers that well. Maybe well enough, given a situation where one already had some other experience with, but I doubt they served them well in very unfamiliar circumstances.
I think to the extent that the people who used them knew what was being conveyed, the tool served them well. Not a lot of people used them, I think is more to the point.
In addition, digital imaging allows the use of images for rather different purposes. Never before has is been so easy to produce poster sized output, sometimes even on one's own printer. Enters the quality question. Being able to do something doesn't necessarily mean it is gong to be of adequate/satisfying/necessary quality ...
Yes, just because we can doesn't mean we should.
For that purpose I think a better, or easier to use, predictive model which includes the specific limitations of the recording medium (such as sensel pitch limitations), is helpful.
Cheers,
Bart
Again, I don't necessarily disagree, it's the quantum of the difference or improvement that I'm interested in. I haven't used your tool yet as I'm on a mobile device, but will take a look later and do some comparisons.
-
Hi Peter,
You're welcome.
I obviously agree, COC values without being output referred make little sense.
Section 1) of my tool now also make it much easier to answer the question;
- How large can I print a file from camera 'x' ?
Before, the obvious answer was; How large is the paper your printer can handle?
But that didn't say anything about the quality one could expect.
Now we set our required quality level, viewing distance, and camera sensor parameters,
and out comes the maximum print size. Anything larger has lower quality, anything smaller
will have even higher quality. As simple as can be.
Cheers,
Bart
Very valuable, thanks!
-
Very valuable, thanks!
You're welcome. I've just added a small bit of additional information to section 1.4), something that I thought might be useful (while I'm also working on the further implementation of non-SI units).
I've noticed that some users have difficulty relating the maximum output size that a camera can produce, at the selected quality level and viewing distance, with how large a 'normal' size print would be. That's because of the viewing distance variable.
Hence I've added a note about those 'standard' size dimensions, assuming that an image with a diagonal dimension equal to the viewing distance, could be classified as 'standard'. It is generally accepted that such (diagonal=viewing distance) dimensions will allow to overview the image content easily because of its moderate angular field of view.
That should make it easier to immediately see if an image would be perceived as impressively large (requiring to step back to take it all in as a total composition), or small (which tends to make people walk up closer to look at details).
Not a major addition, but it could be useful when we produce something with a size or viewing distance that's different from what we're used to.
Cheers,
Bart
P.S. I've also added a comment about viewing perspective and how that relates output size to focal length and viewing distance used. That might also be a useful consideration when the shooting has yet to take place. It's all a lot of information crammed into a tiny space, so maybe I'll think of a different way of presenting it, but it is the logical place in the planning process to consider it.
-
Bart, thank you for this valuable tool.
If you don't mind, I have a few questions related to macro photography,
1) Aperture: Should I use the equivalent aperture (adjusted for magnification) or the lens uncorrected aperture?
2) Focal length: in lenses with internal focusing, does it matter if the focal length varies at close range? Should I consider the adjusted focal length?
3) Can you recommend a procedure to determine pupil factor?
4) Focus stacking: are the steps the same regardless of the method to change focus between images? (Focus lens, change camera+lens position or moving the back only when using a bellows)
Thanks,
Regards
-
Bart, thank you for this valuable tool.
If you don't mind, I have a few questions related to macro photography,
1) Aperture: Should I use the equivalent aperture (adjusted for magnification) or the lens uncorrected aperture?
2) Focal length: in lenses with internal focusing, does it matter if the focal length varies at close range? Should I consider the adjusted focal length?
3) Can you recommend a procedure to determine pupil factor?
4) Focus stacking: are the steps the same regardless of the method to change focus between images? (Focus lens, change camera+lens position or moving the back only when using a bellows)
Hi Francisco,
Great questions! Hope you like the answers ;)
1) You use the regular Aperture that you've dialed in on the camera/lens. The equivalent aperture or effective aperture as it is called by some, is used to allow for an adjustment of the exposure time due to added lens extension.
Because the correction is usually executed by lengthening the exposure time, and not by opening up the aperture (which changes DOF), I've avoided mentioning 'effective aperture'. It tends to confuse, although many use it..
I make a separate mention of the effect on exposure in the comment behind 2.6.1), where both the (required) extension compared to infinity focus is mentioned, as well as the 'bellows factor' or required additional exposure. I've changed the latter from a simple multiplication factor to the number of EVs ('stops') that is required to achieve correct exposure, when it's not measured through the lens. That is mostly used to correct external light measuring and flashes, which cannot know how much extension was applied to achieve the required magnification. On flashes the power can often also be regulated by fixed (1/3rd, or other fractions of) EV amounts, or they are coupled automatically with the lightmetering system which measures through the lens.
2) We have no way of knowing the actual focal length, especially with Macro lenses with internal focusing groups of elements. They may reduce the focal length to allow closer focusing without the need for a longer lens tube/barrel. Therefore, the only result we can measure with certainty is the image magnification. That is also the most common way of dealing with macro issues. I know my EF 100mm f/2.8 II Macro uses internal focusing and extension, and my MP-E 65mm uses primarily, if not exclusively, extension.
So, whether the focal length is reduced, or the extension is effectively increased, we don't know, but the magnification is what will cause the exposure extension and the magnification of the blur/COC and diffraction effects. So in the end, it shouldn't matter much if the extension or the focal length changes, or both, as long as we know the magnification factor. It's usually indicated on a dedicated Macro lens, or simple to measure by focusing on something with known dimensions, such as a ruler or a coin, and compare that to the captured percentage of the sensor/film dimensions.
In Photomacrography, it is not reliable to measure it from the Focus distance used in lens formulae, including my tool, because we don't know the exact optical design and where that positions the optical primary and secondary principle planes.
So my recommendation is, enter the manufacturer's suggested focal length, and use the magnification factor to take care of all other effects.
3) The only procedure that's simple to execute is to look through the lens when not attached to the body, in front of a bright background. Hold it at arm's length and hold a ruler near the side facing you and measure the apparent diameter of the aperture. You measure the entrance pupil with the front of the lens facing you, and the exit pupil with that end of the lens facing you. The Exit (rear) diameter divided by the Entrance (front) diameter is the Pupil factor. It may, or may not, make some difference if the lens is focused at anything else than infinity, due to the internal focusing.
4) There is a difference between using (internal) lens focusing, which follows the magnification rules of my tool as you change the focus distance, and stepping the camera and lens together in a fixed setting on a focus rail. When focusing with the lens, one will need to use progressively larger steps as distance increases, or use the narrowest DOF zone and step through the scene with that (but that will require more slices than strictly needed).
When the focus rail method is used, which I prefer when things get really magnified a lot, like with the MP-E 65mm, then the magnification factor is also unchanged, and the DOF slice will be constant. So in that situation you just calculate the DOF of a given magnification factor, and use that as your fixed step increment.
There is another difference between the two methods, and that has to do with the amount of distance change of the entrance pupil relative to the subject. When the entrance pupil changes position, as is usually the case unless we use equipment which allow to only change the sensor plane position, the perspective changes as well. Dedicated stacking software will compensate by changing the slice magnification, but cannot deal with some of the occlusion effects and perspective shifts between foreground and background features. So, unless the DOF is extremely shallow, there may be stacking errors. Rik Littlefield has written an excellent article (http://zerenesystems.com/cms/stacker/docs/troubleshooting/ringversusrail) about that.
So, whenever lens focusing is used to step through the DOF zones, you can use my tool for guidance on step size and number of required steps, and when a focus rail is used with a fixed magnification factor, you can use my tool to determine that fixed DOF zone depth. It will be difficult to focus by measuring focus distance, because it is not obvious from where on the lens itself to measure, so some common sense needs to be applied, or the shallowest DOF slice distance should be used for all slices. It's better to have no gaps in a focus stack, so I'd err on the side of caution.
Cheers,
Bart
-
Bart,
Thank you very much for your detailed responses. There is a potential issue for some cameras / lenses that use the "adjusted aperture" instead of the uncorrected aperture. For instance, some Nikon camera/lenses combinations use the adjusted aperture. There are also some models of Nikon manual micro lenses with "compensating" diaphragm, so it is important to be aware of this when entering the data in the application.
Regards,
-
Bart,
Thank you very much for your detailed responses. There is a potential issue for some cameras / lenses that use the "adjusted aperture" instead of the uncorrected aperture. For instance, some Nikon camera/lenses combinations use the adjusted aperture. There are also some models of Nikon manual micro lenses with "compensating" diaphragm, so it is important to be aware of this when entering the data in the application.
Hi Francisco,
Thanks for that. I didn't know that.
I also hate that, when camera's make decisions that influence image quality/characteristics without the user being able to allow or disallow that (unless it only reports the wrong/'corrected' aperture value). The physical aperture setting changes foreground/background blur characteristics and changes the diffraction pattern. That's not something a camera should decide on its own or suggest to be using, IMHO.
What they should have implemented, is an option to automatically adjust the exposure time as a function of the magnification factor (or focus distance, same difference), or a suggested increase of the exposure time, in case no through-the-lens exposure metering is possible.
So, in those cases where the camera/lens thinks it knows better, one needs to use the comment in my tool behind the magnification factor about +EV, and enter that as a wider aperture value in the tool when the camera/lens actually changes the aperture, or a narrower aperture when the camera only reports the wrong / 'corrected' value.
The DOF calculations in my tool use the actual physical aperture diameter that is used. The tool even uses more accurate values than the nominal ones, to make sure that an exact 1/3rd stop interval is used instead of rounded values. So, instead of calculating with 5.6 as suggested by the f/5.6 marking on the lens/camera, it uses 5.656854249492381 which is (save for floating point number limitations) exactly one stop narrower than f/4. That may also be the reason for tiny fractional differences in the results between this calculator and others one can find on the internet.
Cheers,
Bart
-
Hi Bart!
Another great tool from your hand - thank you!
Entering the values for my Canon 5D2, and an intended print size of 70x100 cm with a diagonal = viewing distance of 1.2 m, the tool shows me that I have enough resolution even for Higher Quality prints: Max 137.2x91.84 cm @ 104 PPI; 100 x 66.67 cm @ 142.6 PPI.
Does this mean that my efforts with SuperResolution are waisted?
Or can I exspect that the result of the following upsampling to the printer-native resolution will be better if there are twice as many pixels to begin with?
Kind regards - Hening
-
Hi Bart!
Another great tool from your hand - thank you!
Entering the values for my Canon 5D2, and an intended print size of 70x100 cm with a diagonal = viewing distance of 1.2 m, the tool shows me that I have enough resolution even for Higher Quality prints: Max 137.2x91.84 cm @ 104 PPI; 100 x 66.67 cm @ 142.6 PPI.
Does this mean that my efforts with SuperResolution are waisted?
Hi Hening,
At that viewing distance it may not be strictly needed, but at closer viewing distances, or for those with higher visual acuity, it still helps to have more resolution. This also assumes good illumination levels, because when the pupils of our eyes dilate too much (lens edge aberrations), visual acuity suffers, as it does when the pupils need to contract too much (diffraction). Veiling glare (loss of contrast) increases with age.
Or can I expect that the result of the following upsampling to the printer-native resolution will be better if there are twice as many pixels to begin with?
The subsequent up-sampling to native printer resolution will be more accurate, but we won't be able to resolve that additional resolution with our eyes, until we move in closer.
However, one may be able and exploit the additional resolution by getting a higher quality deconvolution output sharpening. That will help the local contrast that we can resolve with our eyes, but it also depends on the subject. So for those who know what they are doing (and have the right tools), the resolution that cannot be resolved anymore by eye, can possibly still be put to some good use.
That's also why I recommend the use of Benvista's Photozoom Pro for up-sampling for large format output. Pixel peeping at display level it may look a bit artificial, but at the proper viewing distance it produces better viewing quality.
Cheers,
Bart
-
Need this in an iPhone app
what is iPhone ?
-
Hi Bart,
Thank you for your reply!
> However, one may be able and exploit the additional resolution by getting a higher quality deconvolution output sharpening.
I don't do the output sharpening myself, but leave that to my print service. However, I plan in the future to *capture* sharpen *after* resampling to print size. I assume, this, too, might benefit from additional resolution?
All in all, however, what you write sounds like the benefits of SR for my preconditions are marginal, and when I get the time, I may try it out.
Kind regards - Hening.
-
Hi Bart,
Thank you for your reply!
> However, one may be able and exploit the additional resolution by getting a higher quality deconvolution output sharpening.
I don't do the output sharpening myself, but leave that to my print service. However, I plan in the future to *capture* sharpen *after* resampling to print size. I assume, this, too, might benefit from additional resolution?
Hi Hening,
Yes, absolutely. It's unfortunate that we need to do this, but in many cases we can still produce better quality output by sending our homebrewed results for plain printing. Not all external parties master the necessary skills, tools, time, to up-sample and sharpen for output as it should be done.
A JPEG which was already converted to the output profile for the off-site printer usually suffices, because they can skip file modifications such as output sharpening and profile conversion. One needs to setup a line of communication with the people who do the printing, which may sometimes be difficult. I usually specify, print without alterations and without additional sharpening, and embed the profile to reduce the risk of double profiling.
All in all, however, what you write sounds like the benefits of SR for my preconditions are marginal, and when I get the time, I may try it out.
Maybe, maybe not. It depends on how large one needs to go, but for many applications adding a bit of viewing distance in the equation can reduce the need for SR. It won't hurt to have the extra files for when it is needed, but you may be able to make that additional operation more exception than rule.
Good Capture sharpening and/or Output sharpening can help a lot, and using the best upsampling method can also make a difference. I hope my tool helps in planning the necessary steps to achieve the desired outcome.
Cheers,
Bart
-
Hi Bart,
thank you for your reply. Your tools are indeed a great help, and so are your software recommendations, which are highly appreciated.
As for the print service: I don't use jpegs other than for web display. I could not afford/justify the purchase of a large format printer, so I searched for the cheapest german print service that does tifs, and I am overly satisfied with digitaloriginal.de. I have the most excellent communication with the owner, Mr. Hamacher. His combined upsampling and output sharpening was better than what I could achieve using PS softproof and Qimage sharpening. (no affiliation with him).
------------------
Proceeding to the Camera-lens Settings and Focus Stacking panes rises some questions:
Example: f=85mm, f/8, I leave the pupil factor at 1, because I am too lazy to determine it; I want the max DOF, so I set Focus Distance to the hyperfocal distance=55.53 m.
In landscape, the whole frame is the "object". The tool shows width=23.48m, height=15.66m. Which significance has this? Does it have any influence on the subsequent calculations?
Then, I copy the settings to Focus Stacking and assume that I want DOF from 5 m to infinity.
What is the Goal Distance Range? (In the example, it is given as 9.999e+ - I have a vague feeling this means "infinity".) - What is the Search: Front-Rear-Aperture? Regardless which of the radio buttons I click, nothing seems to change.
3.5 is called the MAX number of images, 6 in the example. These 6 slices are without any overlap. So shouldn't 3.5 be named MINIMUM number of images? In real life, you want overlap, Helicon recommends it, so you will need more slices. Or ...?
Kind regards - Hening.
-
Proceeding to the Camera-lens Settings and Focus Stacking panes rises some questions:
Example: f=85mm, f/8, I leave the pupil factor at 1, because I am too lazy to determine it; I want the max DOF, so I set Focus Distance to the hyperfocal distance=55.53 m.
In landscape, the whole frame is the "object". The tool shows width=23.48m, height=15.66m. Which significance has this? Does it have any influence on the subsequent calculations?
Hi Hening,
The object/subject dimensions are the scene 'frame' dimensions in the focus plane. This is used for determining if a subject that is focused on, will fit in the frame. It's mostly used to fit e.g. a building facade with known dimensions, or a person of known height, or a car or a truck of known length, a painting of known dimensions, etc., within the frame. You get the idea what it can be used for. A landscape is not typically measured that way. It is directly related to the sensor size and magnification factor, and the latter is directly related to the focal length and focus distance. So in that sense it's related to all other parameters and vice versa.
Then, I copy the settings to Focus Stacking and assume that I want DOF from 5 m to infinity.
What is the Goal Distance Range? (In the example, it is given as 9.999e+ - I have a vague feeling this means "infinity".)
The goal distance range is the depth you want the focus stack to have, the goal you want to achieve by stacking, in case you have a specific range in mind (e.g. an object or subject that needs to be sharp from a given distance to another distance). Since you've focused on the hyperfocal distance, a single slice will already span the distance from half the hyperfocal distance to infinity.
- What is the Search: Front-Rear-Aperture? Regardless which of the radio buttons I click, nothing seems to change.
When you change the total range or depth of the zone you want to have in focus, you need to instruct the program where to look for an expansion of the range. Upon changing the total range, you can have it expand that range by searching for a new front distance (and keeping the rear unchanged), or for a new rear distance (and keeping the front unchanged), or for a new aperture that allows to span the range from front to rear (which may fail when the range is larger than can be achieved by the smallest aperture you can select, and will be heavily diffracted).
3.5 is called the MAX number of images, 6 in the example. These 6 slices are without any overlap. So shouldn't 3.5 be named MINIMUM number of images? In real life, you want overlap, Helicon recommends it, so you will need more slices. Or ...?
The number of slices needed is the automatic result of focal length, aperture, COC, and Total range to cover. You can limit that number of images by entering a lower maximum. Increasing has no use, because the front to rear distance will already be covered by fewer slices. Setting a lower maximum can be useful for limiting the number of slices to a specific number, e.g. the length of a focus rail in macro where each image is shot with the same magnification factor and thus slice depth. Especially when selecting a relatively close front distance, that can result in an impractically huge number of slices which will take a long time to calculate, it can be useful to then set a limit. Other than that, it is not that often used for input, but rather for reporting the total number of slices.
There is no real need to overlap the slice boundaries because you've already selected a resolution on sensor that is 'perfectly' sharp (according to COC limitations) in the end result, based on the prior input parameters. In fact the sharpness between the DOF range boundaries will still increase towards the plane of optimal focus in between those boundaries (unless limited by sensel pitch), but you won't be able to resolve that in the output by eye. There will be no visible gaps in sharpness between the slices, which should indeed be avoided, hence the recommendation by HeliconSoft.
Cheers,
Bart
-
Hi Bart,
thank you for your extensive answer!
Kind regards - Hening.
-
This is a really interesting tool, and it would be interesting to have an app to do that in the field (although I guess after a while, one will remember the values by heart).
What I would like to do, is focus a tech-cam by the numbers (i.e basically hyperfocal distance, then change aperture if there is a really nice foreground - or something like that).
Now I thought, it's easy, I just take the CoC value given for my desired printing size from the web-tool, and plug it into a "conventional" DoF calculator. I happen to have TrueDoF-Pro on my iPhone. With my settings, (IQ160, Landscape, 32mm lens, f/8, 0.5m viewing distance, I get a hyperfocal of 21.04m), I get a CoC value of 6um (~1x pixel size, by the way, it seems my choices are picky, since usually one selects ~1.4 - 2 x pixel size).
Well on the TrueDoF-Pro, to replicate roughly the Hyperfocal from the web-page, I need to use a CoC of 12um ! That's strange. The other values are also "similar", TrueDoF-Pro yields Front distance of 11.6m (web-tool 10.59).
TrueDoF-Pro says it takes into account diffraction, so that's a possible source of difference. If I set the f number to 11 in TrueDoF-Pro, I get infinity on both front and rear focus, which could indicate that diffraction smoothes everything out. But maybe I'm trying to debug TrueDoF-Pro or am doing something wrong with it :-)
What do you think ? Is it possible that other assumptions would change things by so much ?
-
TrueDoF-Pro says it takes into account diffraction, so that's a possible source of difference.
Hi,
I think it is. They seem to adjust the COC by a certain amount depending on the amount of diffraction.
In my tool, I only report the reduced MTF that is caused by diffraction (but do not increase the COC), until diffraction results in a zero MTF (beyond which my tool does increase COC). I made that (reporting instead of adjusting) trade-off, because system MTF is also dependent on lens MTF (which is unknown until measured for a particular lens), and unknown AA-filter, an unknown sensel aperture (micro-lens, transfer gates, etc.), and obviously subject contrast is variable and diffraction differs by wavelength.
So I figured that, given so many variables, it only makes sense to not assume an effect on COC (but there will be additional signal deterioration), until it kills all signal. That deterioration knows no hard boundary (except for MTF=0), so one can use the results of my tool as guidance for planning, but absolute performance depends on some additional factors. Besides, deconvolution sharpening can restore some of the visual deterioration, as long as the system MTF is above zero.
What do you think ? Is it possible that other assumptions would change things by so much ?
I'd have to study how their weighting of those variables is assumed to affect the actual COC assumption before I can comment on that. I do get a feeling that they try to use the COC to incorporate other effects that affect the impression of sharpness. I do not know if they made sensible assumptions, but it seems difficult to translate reduced contrast to absolute cut-offs.
Cheers,
Bart
-
Interesting...
I guess one should actually measure the MTF of the whole system, Lens -> Camera -> Processing -> Printing. But that is a can of worms probably nobody is willing to open :-)
-
Interesting...
I guess one should actually measure the MTF of the whole system, Lens -> Camera -> Processing -> Printing. But that is a can of worms probably nobody is willing to open :-)
Hi,
Well, in a sense the can is open (http://www.luminous-landscape.com/forum/index.php?topic=68089.msg538932#msg538932) for those who want to, except for the physical printing part which comes with it's own set of challenges.
Cheers,
Bart
-
Thnaks a lot for sharing this excellent tool. I really like this no-nonsense interface.
So, a question/wish; Would it be possible to let the user adjust the 1.3 px parameters in case of cropped images, i.e. instead "the camera's sensor array" to become the "the used part of camera's sensor array" (as I do all the time)?
Btw: Imo you could prioritize the functionality and "engine" before supporting strange non-SI choices (sorry to the US residents living in America, its really time you catch up with the modern world in term of units ;) :D ).
Again, thanks a lot.
-
Thnaks a lot for sharing this excellent tool. I really like this no-nonsense interface.
So, a question/wish; Would it be possible to let the user adjust the 1.3 px parameters in case of cropped images, i.e. instead "the camera's sensor array" to become the "the used part of camera's sensor array" (as I do all the time)?
Hi Alf,
Currently the only way to do that is by scaling both the sensor (Width and Height) and the pixel dimensions by the same amount, causing the sensel pitch to remain the same. That would require to calculate and change 4 values, doable because you only need to do it once for the session, but not convenient with repeated use.
I could add a 'Scale factor'. I'd prefer to avoid the use of the term 'Crop factor' because that can also add confusion about focal lengths. I'll have to think a bit about this, because it will impact several calculations, and I don't want to break things.
I'm also considering to add some panorama stitching functionality, which should use the full sensor size, so I cannot just change the initially entered dimensions, but have to only add the scale factor(s) to selected/relevant calculations.
I'll see what I can do, it'll take a bit of time.
Cheers,
Bart
-
Hi Alf,
Currently the only way to do that is by scaling both the sensor (Width and Height) and the pixel dimensions by the same amount, causing the sensel pitch to remain the same. That would require to calculate and change 4 values, doable because you only need to do it once for the session, but not convenient with repeated use.
I could add a 'Scale factor'. I'd prefer to avoid the use of the term 'Crop factor' because that can also add confusion about focal lengths. I'll have to think a bit about this, because it will impact several calculations, and I don't want to break things.
I'm also considering to add some panorama stitching functionality, which should use the full sensor size, so I cannot just change the initially entered dimensions, but have to only add the scale factor(s) to selected/relevant calculations.
I'll see what I can do, it'll take a bit of time.
Cheers,
Bart
Thanks again,
Yes, I assume I can approximate by scaling the output values in 1.4 by my scale factor (crop sensor factor).
About wording: A a 'Scale factor' would be different up and downwards seen from me as output from Lightroom. If I scale downwards by cropping I keep the original quality (per pixel). If I scale upwards by "Enlarge" in LR export I change the quality by letting LR extrapolate.