The video to which you linked describes increasing signal-to-noise ratio (SNR) by pixel averaging. Astrophotographers routinely use this technique, and other techniques like bias and dark frame subtraction, to boost SNR.
For image averaging, the percent reduction in the original random noise goes as:
1/SQRT(n)
where "n" is the number of images used in the averaging operation. So, for example, if you use 4 images, it will reduce the original random noise to 50% of the original.
Using this technique requires that the noise is random, and does not have a pattern. There are obvious limitations to this technique, the same as most computational techniques - camera shake, object motion, ghosting, alignment, etc. all will cause this technique to fail.
FWIW, the DxO ONE camera shoots a raw format the call "SuperRAW" where the camera takes four exposures in rapid succession and creates a single raw file out of the result. This format can be read by DxO software and provides an additional level of noise reduction.
The most useful noise reduction technique depends upon the shooting situation and the exposure parameters that you need to keep fixed to achieve your exposure. As long as your noise is random, there are several solutions out there, each with their strengths and limitations - most of them have trials that you can use to see the results and figure out if the specific tool fits into your workflow.
Ironically, once fairly strong noise reduction is applied you might find that you get a more pleasing image by adding some "grain" back into the image. Adding grain, or noise, sometimes improves the appearance of the image by effectively increasing the image's accutance.
I use Neat Image, when necessary, on RGB images. On raw files, DxO's PRIME noise reduction seems to preserve the most detail in a raw file, but YMMV.
If you want to experiment with image averaging for noise reduction, try some astrophotography software - Franzen Denoise permits the user to also feed it a set of images to average as well as a dark and a bias frame.
In the end, decreasing noise is best accomplished by achieving the best possible exposure. You can also experiment with the in-camera noise reduction function 9should your camera have one) and see if you get better results from the camera's NR than post-processing software.
kirk
and i read a tutorial about take several shots to get the best results
