This is the Milky Way above Kakadu National Park, Australia, and my first try at astrophotography.
Thirty second exposure, f/4.0 at 17mm with ISO 12800 (!!) which in hindsight probably was way too high and resulted in a lot of noise. I also used Picasa to increase contrast, but I probably went a bit overboard there as well.
This is the original RAW file (.CR2), I am looking for input on how to better postprocess this picture to reduce noise.
Firstly had you lowered the ISO whilst staying at 30s f/4 you wouldn't have ended up with any less noise.
There's probably nothing you could have done to prevent the noise, I presume f/4.0 was the maximum aperture and if you went any longer than 30 seconds you would get star trails. You might even get less noise if you raise the ISO but that's another story.
However there's plenty you could do to rescue the image, the main thing is reducing chroma (colour) noise. Most noise reduction plug-ins as well as RAW converters give you the option of reducing only colour noise.
Here is the image with some brute force chroma noise reduction (split to LAB in GIMP and then Gaussian blur of 250 to the A and B channels):
The noise reduction has also fixed the magenta cast caused by noise in the red channel. A dedicated noise reduction plugin could do much better than this. A little luminance noise reduction would help too, but not too much in case it mistakes the stars for noise.
First, I must say that for a start, this is a great photo. Noise is a problem but your photo still looks great.
Now I suggest you have a look to the software DeepSkyStacker, its website and documentation.
They have a great page that explains the theory behind the software.
Basically, you combine multiple similar shots to reduce the noise. You can also add specific shots like dark or flat shots that will help the software to understand the noise generated by your camera. All of this is very well explained in the documentation.
EDIT: As Matt Grum pointed in the comments, you will not be able to use this technique with a foreground as the sky will rotate during your shooting because of earth rotation. DeepSkyStacker can keep track of stars in your different photos even if they moved but will not be able to handle foreground.
I humbly offer two views on the problem presented.
PictureCode's Noise Ninja proved worth to rely upon every single time I used (needed!) it.
Anyways, I did some researches for this answer; and it seems that nowadays Topaz Labs' DeNoise is the best plugin available.
There's a review of DeNoise 5 at Photographyblog and Google pointed out a comparison between these and two other plugins at Colormancer.
From the EXIF:
I strongly agree with Matt's remarks on ISO setting... just halving the ISO (and doubling the exposition!), would form a picture with much less noise.
Canon EOS 60D's sensor presents severe noise above ISO 1600 (as noted in this review).
Hence, you ought to keep ISO at 800; preferably, below that.
But, you HAVE to be told about your lens.
At 27mm a mere f/4 is, sorry, a very lousy performance... this lens worth nothing.
For hundred bucks you get at Amazon a way more faster lens, f/1.8 at 50mm.
Wonder how impossibly expensive would be f/1.4 at 50mm? Wonder not, since it's pretty affordable shall you desire it enough.
With fast lenses like these you easily can set ISO at 200 (or even 100) whilst keeping exposure times short enough for those static stars, without the need for post (lossy) processing.
Instead of a longer exposure, make multiple exposures. For example, 40 shots of 20 seconds each. Find software that lets you shift the image position, and lines it up for you for the best match of the star points. I saw such software a number of years ago but do not remember its name. This will eliminate the star trails, though the ground objects will be blurred. The noise will be reduced. Just another option to play with.
And some of that noise could be fixed noise at various pixels. Recording a lot of dark frames (shoot same ISO and same time, but with the lens cap on) and subtracting out the sum in the right proportion could reduce some of the noise.
