Stack Images
This transformation lets you combine up to 16 input images into a single output image. Each image can be assigned a relative weight that defines its contribution to the final image, and this weight can either be a constant value or may vary according either to a mask or a brightness curve.
There are a number of common uses for this transformation:
•To simulate multiple exposures.
•To combine multiple images of the same scene to reduce noise.
•To simulate a single very long exposure
•To combine multiple bracketed exposures of the same scene to produce an image with an extended dynamic range.
•To use multiple exposures to remove moving objects and reveal the invariant background.
Image Selection Bar
Across the top of the dialog box is a row of 16 thumbnail buttons - one for each of the 16 possible input images. If the thumbnail is blank, there is no image defined for that position, so if you only want to stack 2 images then all but two of the buttons will be blank. Clicking one of the buttons makes the corresponding image the current one and displays its information below.
If no input images are selected, there is no way to compute an output image, so a checkerboard image is produced instead. Once you select at least one input image, the checkerboard is replaced with a normal output image.
Tabs
Across the bottom of the dialog box is a series of three tabs that are used to select which set of controls are displayed.
Brightness Curve -- This displays a brightness curve that is applied to the image before combining it with the other images. Each image can have its own brightness curve. The initial brightness curve is a straight line from the lower left to the upper right which has no effect on the image. You can adjust this curve to alter the brightness or contrast of the input image before stacking.
Density Mask -- This displays a curve that is applied to the image only when the operation is set to Blend. The density mask determines how heavily the image is weighted when you combine it with the other images. The initial density mask curve is a straight line across the top, indicating that all brightness levels in the image are weighted equally. Adjusting the density mask curve for a given input image changes the weighting of the lighter or darker parts of the image. Where the curve is near the bottom of the graph, the amount for corresponding brightness levels is small; where the curve is near the top, the amount is large.
Final Curve -- This curve is applied to the end result of combining all the images and is normally used to adjust the contrast and dynamic range of the result image.
Operation
This control lets you specify how the input images are combined. If you select any operation other than Blend, the Density Mask tab is grayed out as it applies only to Blend.
Blend -- Blends the pixel values for each pixel position from among all images. Use this option when combining images of different exposures into a single composite image.
Find Lightest -- Chooses the lightest pixel for each pixel position from among all images.
Find Darkest -- Chooses the darkest pixel for each pixel position from among all input images.
Median -- Chooses the most popular value for each pixel position from among all the input images. You can use this option to remove moving objects from a series of exposures.
Filter -- Multiplies image values. Essentially, filter simulates sandwiching slides -- i.e. the darker the image data, the greater is it’s filtering effect. When Operation is set to Filter, an Exposure Compensation slider is also displayed. Increasing the exposure compensation brightens the output image to compensate for the darkening effect of the filter operation.
Subtractive Filter -- Multiplies the negatives of image values. and then flips the result. Essentially this simulates printing a sandwich of negatives of the two images. When Operation is set to Subtractive Filter, an Exposure Compensation slider is also displayed. Increasing the exposure compensation darkens the output image to compensate for the lightening effect of the subtractive filter operation.
Exposure Compensation
This slider is only displayed when Operation is set to Filter or Subtractive Filter. It lets you compensate for the tendency of Filter to darken images or Subtractive Filter to lighten images.
Image
Depending on which input slot is currently selected in the Image Select Bar, this control lets you choose what image, if any, is loaded into the current slot.
Amount
Depending on which input slot is currently selected in the Image Select Bar, this control lets you set the amount for the current image.
The amount sets the relative contribution of each image. If all the image amounts are set to the same value, the images are all weighted equally and the result image is the average of all the input images. This produces a result similar to a multiple exposure. If some of the amounts are set higher than others, then the corresponding images are more heavily weighted in the average and consequently are more visible in the final image. For example, when combining multiple exposures, you would normally weight each image according to the length of its time exposure.
If you select an amount mask, you can vary the weights across different parts of the image. Using a gradient as a mask, for example, you could weight the left side of an image more than the right side.
Final Shaping Curve
When the final shaping curve button is depressed, a curve control is displayed. This curve is applied to the result of blending all of the stacked images and is normally used to make final adjustments to contrast and brightness.
Settings Menu
Output 16 bits-per-channel -- when checked causes the result image to be 16 bits/channel even if the input images are 8 bits/channel
Build Curves -- automatically builds curves for combining bracketed exposures (see below).
Load a Set of Images -- clears out the current settings (with confirmation) and then lets you select a one or more image files which are opened and selected as inputs to Stack Images. If a file is already open, the open copy is used instead of reading in a new copy. This saves a lot of time compared to opening and selecting multiple files one at a time.
Using Build Curves
The Build Curves command is used to calibrate your camera and create the necessary brightness curves and density masks to combine up to three bracketed exposures in order to create a single image with a wider dynamic range than any of the individual images.
Step 1 - Photograph the target
If you have a Kodak Q-13 or an IT8 test target, you can use that. Otherwise, you can create your own simulated Q-13 test target with the Test Patterns/Simulated Q-13 transformation and print it out on a printer that provides a good range from solid black to pure white. The exact densities of the gray patches are not critical. Then, photograph the target under even and constant illumination at the range of bracketed exposures you want to combine later, such as -2EV, 0EV and +2EV. Check the histogram when taking the normal exposure to see that the test data is more or less centered in the histogram and not biased to one side or the other.
Step 2 - Prepare the images
Process the test images as you normally would with no tonal adjustments and open them in Picture Window Pro. Crop each image so only the gray scale is visible. Make sure the lightest patch is on the left and the darkest is on the right. Retouch out any specks.
Step 3 - Load the images
Start up Stack Images and select the images as follows: Image 1 - underexposed, Image 2 - overexposed, Image 3 - normal exposure.
Step 4 - Build the curves
Select Build Curves from the Settings Menu. This brings up the Build Curves Dialog Box:
Target Type -- enter the type of target you are using.
Black and White Points -- The black and white point settings let you determine what level the black and white patches in the normally exposed test target will be assigned in the final image. Setting the black point at zero and the white point at 1.0 will expand the normal image so that its black patch becomes pure black and its white patch becomes pure white, but doing this does not leave any room to add new tonal information below black or above white which is why we are going to all this trouble in the first place. Setting the black and white points closer together results in a lower contrast final image, but this leaves room to render shadows darker than the black target patch and highlights lighter than the white target patch. The default values of 0.15 and 0.85 are a good starting point, but for very contrasty scenes where you are bracketing by more than two stops you might want to use 0.25 and 0.75 or even 0.30 and 0.70.
Include Normal -- Select Yes or No depending on whether you want to build curves based on just the underexposed and overexposed images or on all three images
Click OK to generate the curves and then use the Settings Menu to save the settings for future use.
Tips
To remove moving objects from a scene, combine a series of exposures setting Operation to Median and leaving the curves set to their defaults. As long as most of the images do not have an object in them, the object will disappear.
When you select an input image by clicking one of the long row of buttons at the top of the dialog box, the corresponding input is selected. Thus, if you click the button to see the result full screen, you will see the selected input image instead. To view the final result in this situation, just click on the Stack Images thumbnail in the image browser.