QUOTE (pco98 @ May 21 2008, 06:32 AM)
I've started experimenting with longer night time exposures and star trails. Currently, I have my 5D set to long exposure noise reduction. So far the maximum time I have done is 45 mins + 45 mins in camera processing although I think I could push this up to the hour mark.
Is there a way to replicate Canon's in-camera noise reduction processing in PS so I could double my current exposure times and use the battery capacity to the fullest?
Thanks,
Ross
The best way to do this is not in Photoshop but rather to do the noise subtraction while the image is in the raw state. Raw conversion mixes up the individual pixel data via the process of interpolating the colors from the Bayer array of color-filtered photosites; in addition, the gamma correction is a nonlinear curves adjustment that distorts the original raw data and makes the subtraction even more imprecise. The noise signature of the camera is different from pixel to pixel, and so what you want ideally is to subtract the noise from individual pixels before the raw conversion is done. This is what your camera does when it performs long exposure NR; it holds the image you took in a buffer while it records a dark frame, then subtracts the latter from the former and records the result in the raw file.
The freeware raw converter dcraw has the capability to subtract a dark frame taken separately from a raw image file; astrophotography programs such as IRIS can do the same and more. However they require a bit of manual labor since they are not as sophisticated as ACR (dcraw is a command line application; IRIS doesn't have color management and the user interface is a bit basic).
What would be ideal is if standard raw converters had the capability to do these sorts of noise subtractions that subtract a noise template from a raw image, then one would have better noise reduction and one's usual workflow together, rather than having to make a choice between sophisticated noise processing and less capable raw conversion, or less sophisticated noise processing and more capable raw conversion -- which is the current state of affairs.