[阿魚]

如題~~~~~~~~
thx~~~~~~~~~~~

[MaRcO_o]

NR in dc is also created by a software....
The difference depends on the algorithm ....
such as mean filter, median filter
Both of these are algorithm... not the physical filter

[dede]

邊有physical filter 可以做NR?\n
相機的處理有d唔同,佢係要與你所拍的暴光時間同光圈,,再自己內部"拍"一張,末做NR的處理
software就看那一個軟件的做法了,也有如相機的方法一樣,要你自行拍一張一的全黑相來作混合處理
有d袛取那相最黑的部份來處理(要你自己揀那部份)

[jasonksh]

from www.dpreview.com

Noise reduction is a system for removing unwanted noise from a digital image. It falls into two main categories, reduction or removal of noise from high ISO images and reduction or removal of noise from long exposure images (with "stuck pixels").

High ISO noise reduction
So far we have only seen this feature implemented in the Kodak DCS620x, this method of noise reduction relies on a sophisticated algorithm and on the processing power of the acquiring computer, noise reduction is carried out at the time of conversion of the RAW image for importation. See the Kodak DCS620x review for more details.

Long exposure ("stuck pixel") noise reduction
This method of noise reduction relies on a second frame known as the "dark frame" being taken with the shutter closed for the same amount of time as the main image, this dark frame will be taken either before or after the main image and will be used to identify and subtract "stuck pixels". This has been implemented by Kodak in their consumer grade digital cameras for some time, and more recently in their new DC4800 digital camera. Canon also implemented this method of noise reduction in the Canon EOS-D30 digital SLR.

This method of noise reduction can also be implemented in a software package such as Photoshop. Simply take the original image, then a second image with the lens covered with exactly the same exposure (this becomes the "dark frame"). Open the original image in Photoshop, paste the "dark frame" as a new layer, give it a slight Gaussian Blur (around 0.3 pixels) then change the layer options to "Difference". This will make a fairly good job of removing stuck pixels from the noisy original image. Note: this isn't perfect and will still leave some black "pits" in the image.