Joined: 01 Aug 2006
Location: Richland, Washington State, USA
|Posted: Fri May 15, 2009 5:44 pm Post subject: FAQ: Why aren't my images tack sharp?
|At 100% (actual pixels), my images are soft. Why is that? Can I tell from looking at the image how to make it sharper? Or to put it another way, how can I distinguish between diffraction, lens aberrations, camera movement, subject movement, and mis-focus? |
Well, you've pretty much covered the bases with all those options! (We'll presume that your lens was clean and not foggy from condensation.)
Let's rule out the easy cases first.
Take a look at the smallest point highlights in your image.
If the point highlights are consistently elongated in the same direction, then you're dealing with either camera or subject movement. If the blurs are the same across the whole frame, it was probably camera movement. If not, it was the subject.
If the point highlights are nice and crisp in some area, but not the area you care about, then you're probably dealing with mis-focus.
If the subject was shot with flash, especially with flash at low power, then you can probably rule out motion.
After that, it gets tougher.
If the point highlights are nice round blurs, then there is essentially no information in a single image that lets you distinguish between lens aberrations, diffraction, and some types of vibration.
To separate those, you have to run some controlled experiments and see what happens when you change the aperture.
The basic principle is pretty simple: as you stop down a lens, blurs due to diffraction get bigger while blurs due to lens aberrations get smaller.
So, what you do is to set up a nice solid test platform, configured to the same magnification as your original images. Get yourself a subject with lots of fine detail, like a moth wing or a dead fly, and illuminate it with flash. Then shoot a series of pictures, stepping aperture from wide open to fully closed, adjusting flash power and/or ISO setting to make all the exposures uniform.
When you look at the pictures, what you'll see is that the ones with aperture fully closed are probably pretty blurred. Certainly they will be if you're working at high magnification, where your effective f-number fully stopped down may be f/90 or so. This blurring is due to diffraction. As the aperture opens up, the images will get sharper in the plane of best focus. This is because diffraction is being reduced, and aberrations are not yet a problem. If you have a really good lens, the images will keep getting sharper until the lens is wide open. But more often, they'll get sharper only until the lens is a couple of stops below wide open, and then they'll start getting more blurred again. That transition is where lens aberrations have started to dominate.
With these test results in hand, you can now go back to the images that made you ask the question in the first place.
If the original images look as good as the best of your aperture series, then you know that there's nothing to be done except find a better lens.
If the best of your aperture series looks better than your original images, and the original images were shot at a different aperture, then the problem could well have been either diffraction or lens aberrations. In either case, for subsequent photos you should set the aperture to whatever the test series said was best.
If the best of your aperture series looks better than your original images, and the original images were shot at the same aperture, then the problem could have been a symmetric vibration mimicking optical blur. Bad luck, but it happens sometimes when you're using continuous illumination. In that case, you'll need to either track down and eliminate the vibration, or switch to flash illumination.
If you're interested in doing some calculations to complement the experiments, then take a look at http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm. Just remember that the discussion there is really based on effective f-number, which typically is larger than nominal f-number by a factor of magnification+1. (f/4 at 5X becomes effective f/24.) It's easy to be off by a factor of 2 in estimating diffraction effects, but this will give you at least a ballpark feel for whether diffraction is part of your problem.
I hope this helps -- pixel-level blurs can be tough to figure out.