Program for comparing sharpness of similar images

[larsrc]

I've upgraded from a Coolpix 995 to a DSLR, and one function I find I miss it the 'Best Shot Selector' system which allowed you to take 10 pictures in a row and it would pick the least shaken.  It wasn't perfect, of course, but it allowed me to take pictures in much lower light than would otherwise have been possible.  I still have the habit of taking a series of pictures when the light is low and then handpicking the one that looks the least shaken, but it can be difficult to select sometimes (especially with programs that only show the preview of RAW photos).  Thus:

Does anyone know of a program (preferably available for Linux, second to that for Mac) that can do a comparision of "sharpness" of a series of similar images?  I don't really care how it calculates the sharpness, as long as it's allows me to pick the least shaken picture among a group taken in a row with the same settings.

Thanks in advance,
-Lars

[jmb]

Does anyone know of a program (preferably available for Linux, second to that for Mac) that can do a comparision of "sharpness" of a series of similar images?  I don't really care how it calculates the sharpness, as long as it's allows me to pick the least shaken picture among a group taken in a row with the same settings.

You could just select the file that has the largest size... In general, the files with the largest size will have the most information/detail (and will also probably be the sharpest...).

JMB

[Jonathan Wienke]

Adobe Bridge in filmstrip layout; one image fills most of the monitor and the thumbnails are a scrolling strip along the right side of the screen. It has many other advantages as well.

[larsrc]

You could just select the file that has the largest size... In general, the files with the largest size will have the most information/detail (and will also probably be the sharpest...).

JMB
[a href=\"index.php?act=findpost&pid=86974\"][{POST_SNAPBACK}][/a]

I've been told that too, but trying it out didn't actually get me the sharpest.

-Lars

[svein]

Comparing sizes usually work as long as the images are identical, ie a very fast series or a static subject. If you move the camera slightly or your subject moves/changes then size comparisions wont work. So, just by pressing the shutter for each picture chances are the camera will move.

[tonypassera]

I've upgraded from a Coolpix 995 to a DSLR, and one function I find I miss it the 'Best Shot Selector' system which allowed you to take 10 pictures in a row and it would pick the least shaken.  It wasn't perfect, of course, but it allowed me to take pictures in much lower light than would otherwise have been possible.  I still have the habit of taking a series of pictures when the light is low and then handpicking the one that looks the least shaken, but it can be difficult to select sometimes (especially with programs that only show the preview of RAW photos).  Thus:

Does anyone know of a program (preferably available for Linux, second to that for Mac) that can do a comparision of "sharpness" of a series of similar images?  I don't really care how it calculates the sharpness, as long as it's allows me to pick the least shaken picture among a group taken in a row with the same settings.

Thanks in advance,
-Lars
[a href=\"index.php?act=findpost&pid=86962\"][{POST_SNAPBACK}][/a]

Focus magic gives you a measure of how unsharp the image is
before you click the button to sharpen.  This isn't as convenient
as BSS, and it's not as convenient as a tool that runs in batch mode,
if there is such a thing. However, you might find it to be good enough.

I've found BSS works great for low light and even hand held
macro work.   I've tried simulating it with a d2x by taking 10
frames in quick succession and then using bridge and PS to
find the sharpest frame.  This is not a very convenient process, and it
takes lots of space on memory cards.  I wish nikon would add
BSS to its faster dslr's.  It could be done in firmware.
For times when there's subject movement you can get a better
result than with image stabilization (and BSS combined with image
stabilization would be awesome).
Tony

[Jonathan Wienke]

I'd never use "best shot selection" where the camera makes the choice of which frame to keep and which frame to discard. Sometimes a bit of motion blur enhances the image, and there are also times when an event occurs or a subject has a momentary facial expression that "makes" an image in spite of motion blur. Cameras aren't very good at getting a relatively simple thing like exposure right automatically; trusting one to pick your keepers for you is foolhardy at best. Memory is cheap; memories are priceless. Take the time to choose your own keepers; those who let their camera think for them are not photographers, but merely button-pushers.

[Eric Myrvaagnes]

I'd never use "best shot selection" where the camera makes the choice of which frame to keep and which frame to discard. Sometimes a bit of motion blur enhances the image, and there are also times when an event occurs or a subject has a momentary facial expression that "makes" an image in spite of motion blur. Cameras aren't very good at getting a relatively simple thing like exposure right automatically; trusting one to pick your keepers for you is foolhardy at best. Memory is cheap; memories are priceless. Take the time to choose your own keepers; those who let their camera think for them are not photographers, but merely button-pushers.
[a href=\"index.php?act=findpost&pid=87131\"][{POST_SNAPBACK}][/a]

Right. I've been thinking of designing a PS plugin to select 'keepers' automatically. Basically, it will use a random number generator to pick a preset percentage of your images in a shoot (you can select anything from 1% to 99%.) The rest it discards irretrievably.

If I offer it as shareware, maybe the proceeds will let me buy a Canon 24-105 L IS lens.    

Eric

[larsrc]

I'd never use "best shot selection" where the camera makes the choice of which frame to keep and which frame to discard. Sometimes a bit of motion blur enhances the image, and there are also times when an event occurs or a subject has a momentary facial expression that "makes" an image in spite of motion blur. Cameras aren't very good at getting a relatively simple thing like exposure right automatically; trusting one to pick your keepers for you is foolhardy at best. Memory is cheap; memories are priceless. Take the time to choose your own keepers; those who let their camera think for them are not photographers, but merely button-pushers.
[a href=\"index.php?act=findpost&pid=87131\"][{POST_SNAPBACK}][/a]

The name is marketingese.  It doesn't select "keepers", it selects the sharpest of a limited series similar pictures, which is probably easier to get the camera to pick correctly than, say, focus or exposure.  Besides, I was asking for a program to do it, not the camera, so I'd get a chance to see them on the screen anyways. And of course there's times to not use it, just like there are times to not use the widest aperture or autofocus or flash or any other mechanical assistance.  I'm trying to reduce the time I have to spend on automatable tasks, so that I can spend the time picking between more subjective matters like exposure, focus and composition.  If that makes me a button-pusher, then so be it.

[Jonathan Wienke]

The name is marketingese.  It doesn't select "keepers", it selects the sharpest of a limited series similar pictures,

You don't know what you're talking about; the rest are deleted in-camera, before you ever get to see them. The camera does indeed pick your keepers for you when that feature is active:

"Best Shot Selector: when the Best Shot Selector is active, the camera captures up to 10 frames while the shutter release is held down. The frames are then compared and only the one that exhibits the best sharpness is saved."

Which means the rest are NOT saved, also known as "deleted", which means the camera picked which shot of the series was kept, NOT you. That's not something a photographer would ever allow. It takes more than sharpness to make a good photo; as I mentioned before, there are several cases where using it as the sole selection criterion would be disastrous. Flipping through images in Bridge in filmstrip view is fast, and with a decent-sized monitor you can judge sharpness and other criteria fairly well without having to open each RAW separately.

[feppe]

You don't know what you're talking about; the rest are deleted in-camera, before you ever get to see them. The camera does indeed pick your keepers for you when that feature is active:

"Best Shot Selector: when the Best Shot Selector is active, the camera captures up to 10 frames while the shutter release is held down. The frames are then compared and only the one that exhibits the best sharpness is saved."

Which means the rest are NOT saved, also known as "deleted", which means the camera picked which shot of the series was kept, NOT you. That's not something a photographer would ever allow. It takes more than sharpness to make a good photo; as I mentioned before, there are several cases where using it as the sole selection criterion would be disastrous. Flipping through images in Bridge in filmstrip view is fast, and with a decent-sized monitor you can judge sharpness and other criteria fairly well without having to open each RAW separately.
[a href=\"index.php?act=findpost&pid=88320\"][{POST_SNAPBACK}][/a]

I think you are missing the point. I believe BSS is supposed to be used with the same aperture, exposure and ISO setting, and with the same exact framing for each and every shot. This way the only variable remaining is camera shake, and it's trivial to find an algorithm which finds the "best" (ie. sharpest) picture of the batch. Sharpness is not subjective so there's no need to have a human pick the one from otherwise similar frames.

Whether you want the sharpest picture is another issue. But if you don't, you wouldn't be using BSS or asking the poster's question in the first place. Besides, that's an altogether different debate, one the poster wasn't asking for.

I would be interested in finding a similar program, as pixel-peeping handheld low-light shots to find the sharpest one gets tiring.

[marcmccalmont]

I have been happy using breeze browser pro, it has a magnifying glass that I use to scan images for sharpness. It is quick and inexpensive I started using downloader pro and breeze browser pro together and have never changed.
Marc

[Jonathan Wienke]

I think you are missing the point. I believe BSS is supposed to be used with the same aperture, exposure and ISO setting, and with the same exact framing for each and every shot. This way the only variable remaining is camera shake, and it's trivial to find an algorithm which finds the "best" (ie. sharpest) picture of the batch. Sharpness is not subjective so there's no need to have a human pick the one from otherwise similar frames.

Whether you want the sharpest picture is another issue. But if you don't, you wouldn't be using BSS or asking the poster's question in the first place. Besides, that's an altogether different debate, one the poster wasn't asking for.

The original poster was complaining about the lack of BSS on his new DSLR; citing that lack as the reason he was looking for the program. And in case you didn't notice, I did offer a suggestion for a program that offers a fast and convenient means of comparing images based on sharpness or whatever criteria one chooses, Adobe Bridge.

As to how BSS is "supposed" to be used, it sound great in theory, but except for still lifes and other static subjects, it is a really bad idea for the reasons I outlined earlier. There's a much better solution to the problem anyway--the tripod.

[larsrc]

I think you are missing the point. I believe BSS is supposed to be used with the same aperture, exposure and ISO setting, and with the same exact framing for each and every shot. This way the only variable remaining is camera shake, and it's trivial to find an algorithm which finds the "best" (ie. sharpest) picture of the batch. Sharpness is not subjective so there's no need to have a human pick the one from otherwise similar frames.

Whether you want the sharpest picture is another issue. But if you don't, you wouldn't be using BSS or asking the poster's question in the first place. Besides, that's an altogether different debate, one the poster wasn't asking for.

I would be interested in finding a similar program, as pixel-peeping handheld low-light shots to find the sharpest one gets tiring.
[a href=\"index.php?act=findpost&pid=88341\"][{POST_SNAPBACK}][/a]

(My emphasis) Thanks feppe, you hit the nail on the head.  

My guess is that the sharpness decision uses frequency information taken from the JPEG compression.  File size and simple counting of differences doesn't seem to work for me.

[Jonathan Wienke]

My guess is that the sharpness decision uses frequency information taken from the JPEG compression.  File size and simple counting of differences doesn't seem to work for me.

If your images are sufficiently different from one another that the largest compressed file size is not a good indicator of sharpness, no automated tool is likely to work well for you.

[Jonathan Wienke]

Sharpness is not subjective so there's no need to have a human pick the one from otherwise similar frames.

I guess you've never had AF lock on leaves or branches behind the subject, the fine details of which result in a "sharper" image than the one with the intended subject's face in focus. Leaves and branches tend to have more fine detail than faces, unless the subject is extremely old and wrinkled. Chain-link fences, rock walls, and many other detailed backgrounds can cause the same problem.

[feppe]

I guess you've never had AF lock on leaves or branches behind the subject, the fine details of which result in a "sharper" image than the one with the intended subject's face in focus. Leaves and branches tend to have more fine detail than faces, unless the subject is extremely old and wrinkled. Chain-link fences, rock walls, and many other detailed backgrounds can cause the same problem.
[a href=\"index.php?act=findpost&pid=88404\"][{POST_SNAPBACK}][/a]

You are of course right. But it's not relevant as you're still not understanding what BSS is or how it's supposed to be used properly. If you compose your shot, focus on the subject (manually or with AF), and lock all exposure settings and keep the shutter button down until you have 10 or so shots, you won't have (meaningful) difference in any other area than sharpness in situations where one would use BSS. If the photog lets her AF wander during those shots the problem is with the operator, not the camera feature or its implementation. You seem to look for exceptions which are either misuse of the feature or academic in nature.

And yes, tripod is an obvious solution to the problem, but a photographer doesn't always lug one with her or its use is out of the question. So there is a real need for BSS or a software solution, and I'd be very happy to find one.

Back on topic: i'm going to make tests with compression as I've heard it used elsewhere as well as here. I don't know how Canon 30D RAW files are compressed (I'll look it up shortly) so I'll test with both JPEG and RAW.

[Jonathan Wienke]

If the photog lets her AF wander during those shots the problem is with the operator, not the camera feature or its implementation.

Except that AF isn't necessarily disengaged between shots just because BSS is activated; most digicams don't allow AF to be activated only for the first shot in a series, and many lack any AF lock or manual focus feature whatsoever. So my point regarding focus is hardly mere academic nit-picking. And neither is my point about non-static subjects, like people. The "sharpest" shot is not likely to be the same as the shot with the best pose or expression, even if the subject(s) are trying to remain still.

I'm well aware of what BSS does and what it's theoretically supposed to do; I've been shooting since the early 1980s and have shot over 100,000 frames as a working professional photographer, and have excellent reading comprehension skills. My experience tells me that taking a moment or two to choose my own "best shot" from a series of frames is well worth the minimal time involved, when balanced against the probability that the camera would delete my first choice if I gave it any autonomy in such matters. Memory is cheap; memories are priceless. If you want to risk throwing some of yours away by giving your camera the power to delete your images, you're cheating yourself and your clients, if you have any.

Canon RAWs employ a lossless compression method, so sharper images generally equate to larger files.

[feppe]

Did a quick test. I shot a simple scene including backs of books, bags of Dorito's and candy, Dr Pepper cans and some other weekend paraphernalia. Lots of detail but also some more bland areas (table and curtains). I shot with Canon 30D with RAW+JPEG fine at f/11 with ISO 100 and 1600 using a sturdy tripod in constant lighting level. After using AF to get focus I locked it by going to manual. I varied focus to mimic motion blur (I'm sure people will comment whether this is a valid comparison).

Here are the results.

ISO 100 (Similar results hold for ISO 1600):

in-focus (using AF and confirmed by pixel-peeping at 1:1 on computer screen):
7.3MB (RAW) - 3.0MB (JPEG fine)

very very slightly out-of-focus (pixel-peeping required to notice difference):
7.4MB - 3.1MB

slightly out-of-focus:
7.2MB - 2.5MB

out-of-focus (enough to make it look blurred at screen resolution):
6.9MB - 2.1MB

very badly out-of-focus (enough to bring out bokeh):
6.7MB - 2.0MB

The in-focus shot was _not_ the shot with the largest file as would be expected if sharpness can be judged from file size alone. While the heavily oof shots had indeed smaller file sizes, the slightly oof shots were similar or bigger than the correctly focused shot. JPEG seems to be much more responsive to oof blur which is expected due to the algorithm working very well with smooth areas. Even then the results with JPEG were inconclusive with ever so slight oof.

So, what I see is that judging correct focusing can not be reliably looking at JPEG or RAW sizes _with this scene_. This is not a comprehensive test but it's enough for me to forgo using file sizes alone to pick the keepers.

[feppe]

Except that AF isn't necessarily disengaged between shots just because BSS is activated; most digicams don't allow AF to be activated only for the first shot in a series, and many lack any AF lock or manual focus feature whatsoever. So my point regarding focus is hardly mere academic nit-picking. And neither is my point about non-static subjects, like people. The "sharpest" shot is not likely to be the same as the shot with the best pose or expression, even if the subject(s) are trying to remain still.

I'm well aware of what BSS does and what it's theoretically supposed to do; I've been shooting since the early 1980s and have shot over 100,000 frames as a working professional photographer, and have excellent reading comprehension skills. My experience tells me that taking a moment or two to choose my own "best shot" from a series of frames is well worth the minimal time involved, when balanced against the probability that the camera would delete my first choice if I gave it any autonomy in such matters. Memory is cheap; memories are priceless. If you want to risk throwing some of yours away by giving your camera the power to delete your images, you're cheating yourself and your clients, if you have any.

Canon RAWs employ a lossless compression method, so sharper images generally equate to larger files.
[a href=\"index.php?act=findpost&pid=88417\"][{POST_SNAPBACK}][/a]

I don't think anyone here is advocating the use of BSS or software solutions alone with a non-static scene. So my point remains: if one uses BSS or a post-processing equivalent properly, there's no need to pixel-peep (if one trusts the algorithm of course). Refraining from using BSS or software solutions in non-static scenes should be obvious to anyone reading this site. With such scenes AF shouldn't change targets between shots, but even if it does one can always go manual. And the poster was looking for a similar solution for his current gear which has more advanced features than the camera with BSS.