Can you really fix bad focus?

I managed to capture a couple of parrots in flight, but they where very close and the light in the forest wasn’t great. I was panning ok but the speed of the bird and the slowness of the shutter and not to mention autofocus cranking away. Ok The birds swooped in and I had little time other than your pick up the camera follow the bird and push the shutter. The result was clearly out of focus birds, yet the composition and action was interesting.
Can I recover the focus? My experience has always been I Can Not! Normally such photos get culled as they are first uploaded.

It just happens that Corel are promoting their registration and support with a link to their new 64 bit add-ins and two of the add-ins caught my eye. The first was Wavelet Sharpen 3, which use a larger radius Unshapen Mask (USM) as well as the wavelet sharpening algorithm popular in GIMP. But such sharpening usually enhances the noise, so Wavelet Denoise also looked interesting. It brings the common controls over both luminance and colour noise filtering. With a little back and forth adjustment between these two add-ins and I was able to bring the images to a better “not quiet in focused” state albeit with some grainy artifacts in the background and some haloing. Definitely not in focus but more acceptable. However I can see there is a lot of power in these tools.

Big Hint: Normally I would recommend using manual focus when panning after birds in flight and have any anti-shake of stability adjustment turned off. Of course you need to pre-set the focus to roughly the distance you expect the bird to be and there is a limit on to how slow you can let the shutter exposure be. (for most of us the limit is something like 1/60th. second) so in low light that means a higher ISO (eg 800)

So much noise

That grainy speckle that often infects digital photos taken in low light, is normally called noise, (maybe it is called dark noise, heat noise or salt & pepper noise). It happens when exposures are longer and/or the ISO setting are higher because adjacent receptors on the sensor are charged up longer or more to become more sensitive. That allows heat or current leaks onto the adjacent receptor which interprets it as more light. There are a whole range of methods and algorithms to detect and try to remove this noise. Many such methods are now being built into cameras. They seek to work out what is an artifact of the noise versus real detail in the photo. Some methods look at variation in brightness or luminance (Lumina) other look are variation in hue (chroma), the little specs usually appear variable colours (see below). I find with my camera the chroma noise reduction leaves a better quality filtered image and removes the “digital camera grain” best. However all noise reduction will lead to a slight loss of detail and quality in the final results, typically a smoothing effect.

BEFORE

AFTER

From my perspective noise reduction is the one remaining obvious thing missing in picasa’s simple photo editing tool kit. Unfortunately the best NR (that's Noise Reduction) Software is still relatively expensive and mainly available only as plug in for Photoshop & Lightroom. Don’t despair, a free download trail of noiseware (either as a plug in or standalone) will give you decent results before you need to fork out big bucks. The above example was created with the Noiseware Community Standalone version (free and no trial period). Good Noise filters are available for Gimp.
There is a lot of hype written on the net about NR tools and most are actually much of a muchness, but without testing it is difficult to appreciate subtle differences. Pro Photography Show gives a decent review of leading NR software by way of a number of downloadable high resolution before & after images so you can see for yourself.

It is wise to have at least one such NR filter method available to help you recover an occasional dark photo or even a whole series from your party when the photographer may not have be concentrating.

Fox on the Run

Trying to capture this fox running into deep shadow had the camera deciding to move down to ISO 3200,  nice and sensitive in low light, but this has introduced a lot of “heat” noise. This occurs as coloured speckles, usually most noticeable in flat single tone areas. The noise is primarily caused by the sensor (CCD or CMOS) itself which needs to be charged to record the intensity of light falling on it. When the illumination is low and the exposure long one pixel’s charge can trick the adjacent pixels on the sensor to thinking it has received more light, leading to the little speckles.

So what can be done about it? there are many despeckling filters, often they will be supplied with the software you got with your camera. These generally look for out of place coloured specks and average (blur) them using the tones of adjacent pixels, trouble is this smart blurring often gets a bit over enthusiastic.  Luckily I have found the Noiseware’'s Community Edition program very useful in “filtering” out this noise, but avoiding the over blurred look.

Remember Higher ISO,Low Light and Long Exposures will increase noise in your digital photos.