Digital Reduction of Film Graininess
|A photo of NGC7023 by Tony Hallas (used by permission) was used as a reference for this image processing technique. Here is the before and after results of a process developed for grain reduction using the tools in Picture Window:|
Here's the process:
1) Define a mask which will protect the light area and will allow further processing of the grainy low light areas.
a) Load the image into PW
b) Select the Mask Function
c) In the Mask Window select the Light Curve Button
d) Select the Probe and with the left mouse button pressed point to the image at the point where the grainyness ends and lighter area begins:
e) Now move the left bottom up-arrow to the grid line shown by the red vertical line and press the "Apply" button
f) Select the masked image, magnify to a 3:1 with the button (on main menu bar) and move the image to a bright star area. Hit the feather button , slide the amount to "1" and hit "Apply". The idea here is to increase the mask to the edge of the bright stars. High resolution images will likely require a higher amount of feathering to reach the edge.
g) Hit the "OK" button in the mask window. You should now have the completed mask as partly shown below:
2) Now use the mask with a Gaussian Blur to remove the grain in the low light areas:
a) Select the "Blur" function under the "Transformation" button on the main menu and reset the image back to a 1:1 using the button. Select "gaussian" in the blur window. Point and Click the "Input Image" box and select the original color image. Point and Click the other box and select the untitled mask image. Move the white slider bar to the far left and the black slider bar to the far right.
b) Check the "Auto" box and enlarge the preview image to a 1:1 so that you can see the final result side by side with the original image. For this example the "R" was set to "1" and the Threshold was set to "28". These two setting will vary from image to image. I usually start with R at 1 and then vary the threshold up from 0 until the grain just disappears. Too much of either with make the image look motted.
c) Hit "OK" when you have what you want and the final image is ready to be saved.
Return to AF9Y WebPage on Moonbounce and Weak Signal Detection