It is important to note that the dimensions of the image will always be equal to the first image brought into the workspace. If you subsequently bring in other images that have other dimensions, they will be squeezed to fit into the present workspace dimensions (which is almost always an unacceptable situation, so don't do it in the first place).
When opening individual image files you can select multiple files and bring them all in at once. When opening a Quicktime movie you have several options to help you guide the file-opening process. One of the most important tools at this stage is the ability to crop the movie frames as they are brought in on the fly, rather than cropping them after they have already been brought into the workspace. The advantage of this is that the individual frames never have to take up more memory than their initially cropped size, thereby allowing you to bring in a much larger number of frames. You can specify whether or not the cropping rectangle should align to each frame in the movie as the frames are read. This is a good idea since it prevents cropping of the primary object in the image should the object drift outside your originally specified rect.
There are some other features of the Open Movie Dialog as well. You can choose to only see every Nth frame when you clip on Accept all Remaining where N is a number you can set. By choosing not to see all frames, the opening process will go a little more quickly.
This step is only necessary if the images were captured with a RAW modded camera that captures the Bayer matrix without deBayering it. If that is the case, you must deBayer the frames before any other processing can seriously be performed.
There are four possible Bayer patterns, and only one pattern is right for each frame. However, different frames are likely to require different patterns. The best way to handle this situation is to manually deBayer the reference frame only, by moving the mouse over the first clip and tapping the 'B' key. A dialog will be presented in which you can choose a Bayer pattern and a deBayer algorithm. Trial and error on the four patterns should produce the correct deBayered image. You can then deBayer all the other frames in the working space automatically by selecting all, tapping the 'B' key, and then selecting the option in the deBayer dialog that specifies that the deBayer pattern should automatically be matched to the reference frame.
There are also a number of deBayering algorithms available. They are sorted from top to bottom in order of increasing complexity and required processing time, but also increasing quality of the final result. It is highly recommend that you use the linear interpolation method in all cases. It isn't too much slower than the other methods and will produce much higher quality results.
The first thing you must do besides resampling is align the frames. You must become familiar with the Clips Window and how it works if you are going to quickly navigate and manipulate a large number of clips, so play around with it for a while. Notice that as you move the mouse, the clip under the mouse is always hilited with a thick yellow border. Also notice that the hilited frame is always shown in the Frame Inspector Window. Realize, although this is obvious, that you don't have to use the mouse button to hilite a frame. Simply moving the mouse is enough. It is often useful to quickly move the mouse around over the clips to get a sense for the general quality or alignment of the frames. You can also resize and reshape the Clips Window with the Grow Box in the lower right corner and the Zoom Box in the title bar.
To align the frames, move the mouse around quickly in the Clips Window and find a good quality frame. It isn't vital that you find the best frame, just quickly find a decent frame. This frame will be the reference frame that all other frames are aligned to. Therefore it must be a good frame to begin with or the alignments will be noisy and produce an unnecessarily blurry stack.
When you find a frame you like position the mouse over it and hit the 'R' key. 'R' triggers the Set-Reference-Frame operation and toggles the clip under the mouse with the very first clip in the upper left corner of the Clips Window. Each time you hit 'R' these two clips trade places. The first clip is always referred to as the Reference Frame for the purposes of terminology. Many operations use the reference frame, not just alignment.
It is time to learn another Clips Window command. While moving the mouse around in the Clips Window, pause on a given clip for a moment. As stated above, that clip's frame is shown in the Frame Inspector Window. Now hold down the 'space' key. The Frame Inspector Window instantly changes to show the Reference Frame. As long as the 'space' key is down, the Frame Inspector Window will remain locked to the Reference Frame. When you release 'space' the Frame Inspector Window will revert to its default behavior of showing the clip under the mouse. This is a very useful tool. There are numerous occasions when it is desirable to compare two frames in precise detail. This is the most feasible way to accomplish this goal. For example, say you are searching for a good frame to use for alignment. Rather than look at several frames and try to remember how they compare, simply pick one and set it as the Reference Frame. Then go to one of the other potential good frames and flip back and forth quickly with the 'space' key. If the new clip seems better, set it as the Reference Frame. Then continue to explore the clips for another good frame. After doing this a couple times you will have found a very good clip.
Okay, you've picked a good Reference Frame. Now you must select it. Do this by shift-clicking on the Reference Frame. The clip will now appear with a thin yellow border. The reason you must do this is that you need the Reference Frame to be visible in the Frame Inspector Window after you have move the mouse outside of the Clips Window. Ordinarily, the Frame Inspector Window is blank when this happens (because there is no way to guess which frame should be displayed). However, if exactly one frame is selected then that frame will be shown persistantly in the Frame Inspector Window except when the mouse is over the Clips Window.
Move to the Frame Inspector Window now and click-drag a rectangle around a useful feature. Since this example demonstrates how to work on a planet, the most obvious feature is the bounding rectangle around the planet's perimeter. However, bear in mind that larger Operations Bounds Rectangles (as this rectangle is always called) will require longer periods of time for various operations to complete. In reference to alignment, this rectangle is defining an area which will either be compared by difference with each frame or by centroid location with each frame. In the case of difference alignment, each frame will be aligned to the Reference Frame by finding the translation that causes the minimum difference between that frame and the Reference Frame within the Operations Bounds Rectangle.
It's time to actually align the frames. Select all of the frames. Now press and hold down just one of the '1', '2', '3', '4', or '5' keys. While holding down the number key you have chosen, click on any selected clip. Since they are all selected, you can click anywhere in the Clips Windows (except in the lower right corner which will potentially be blank). This usage is common to most operations that act on Clips. You perform the operation by clicking on any clip that is presently selected. Then the operation will occur on all the selected clips.
Which number should you use? Keys '1', '2', and '3' correspond to Difference alignments. The '1' key should be used for really bad alignments, 15 pixels or more. The '2' key should be used for alignments between 5 and 15 pixels. The '3' key should be used only for very tiny alignments, up to 5 pixels. All three methods will align in all situations. However, each method will perform more quickly than the other two methods within its specified range.
The '4' key corresponds to Centroid alignment. If you use this method, you should always follow it with a minor Difference Alignment of say '3' key, since it is not as precise as the Difference alignments. However, the centroid alignment is definitely faster than the Difference Alignments when the alignment error is large and is therefore recommended as a first-pass approach to alignment.
The '5' key corresponds to Cross-Correlation which is calculated using the Fourier Transform of the frame. Although highly toted in the literature, it has some drawbacks. It performs more slowly than centroid alignment, but probably more precisely than centroid alignment. It is unclear whether it runs faster or slower than difference alignment and whether it performs better than difference alignment. Finally, it can get thoroughly confused sometimes, which can result in completely incorrect results.
It is also worth noting that you can choose whether or not to align the red, green, and blue channels separately. The "difference" method of aligning frames aligns by, as one would expect, matching the difference between the frame being aligned and the Reference Frame. The default behavior is for this difference to be calculated using the "value" of a given pixel in the two images, which is the average of the red, green, and blue components for that pixel. By selecting the Align RGB Channels Separately menu option under the Operations Menu, you can align the red, green, and blue channels of an image separately, so that rather than having one translation (horizitonal and vertical slide) for the image, you have three. This can be particularly useful for objects imaged low on the horizon where the atmosphere often distorts the blue channel out of sync with the red channel.
In order to align the RGB channels separately you simply check the corresponding menu item and then align in a normal fashion by holding down the '1', '2', or '3' key to specify an auto-alignment search size and clicking on a clip.
A hint about alignment: You can manually nudge any clip. While no clips are selected move the mouse over a single clip, or alternatively make a selection of clips by shift-clicking and shift-click-dragging over the clips in the Clips Window. Now use the arrows keys to nudge the frame or the entire selection of frames. By holding down shift, option, and/or control while using the arrow keys you can cause larger nudges. Each of the three keys is equivalent in this usage. The trick is that the more of these three keys you hold down, the larger an nudge you get, so you have four possible nudge sizes, 0 keys, 1 key, 2 keys, or 3 keys. It is often helpful to manually nudge frames some before autoaligning. This is especially useful if, after autoaligning, you realize the autoalignment made mistakes with some of the frames. In such a case, manually nudge the frames into a reasonably close alignment and then throw the autoalignment at them. I do this all the time.
Also be aware that you can manually nudge the red, green, and blue channels separately. To do this, check the Align RGB Channels Separately menu option under the Operations Menu. Then, while holding down the '1', '2', or '3' key, tap or hold down an arrow key (be aware that this won't work in larger nudges with the shift, option, or control keys. You must nudge separate channels in single pixel increments by only holding down a number key). Only one channel will be nudged, corresponding to the numerical key being held down. This can be quite confusing since the same number keys can specify two entirely different things depending on the situation. When clicking with the mouse they specify auto-alignment search sizes. When using the arrow keys they specify a single channel. Sorry about this. I am working on a new interface for a future version that will alleviate this problem.
There are two ways to normalize. You can normalize the brightest pixel in the Operations Bounds Rect or the average brightness within the Operation Bounds Rectangle. The second method works better but make sure you don't saturate the frames or you will be throwing away useful information.
The simplest and usually the most accurate method is "manual frame selection", whereby you personally inspect each frame and rule it accepted or rejected. To do this, start passing the mouse over the Clips Window. All frames are included in the stack by default. To reject a particular frame, click on it with the mouse. It will be Xed out in red in the Clips Window and a small red X will appear in the Frame Inspector Window whenever that frame is shown in the Frame Inspector Window. With a little practice you can move through a large number of frames rather quickly.
A hint about manual frame selection: You should have already selected a good frame for the reference frame since you did this to align. Therefore, it is easy to compare the rest of the frames to the reference frame by using the space-bar command mentioned above. As you inspect the frames toggle back and forth quickly to the reference frame to visually appreciate the difference between the reference frame and the frame you are inspecting at any given moment. This will give you a better indication as to whether to include the frame.
Sometimes you have hundreds of frames and it will simply take too much effort to manually inspect all the frames. In that case, it is best to sort the frames by some measure of quality and then only include some fraction of the frames at the beginning or end of the list. Each of the sorting methods requires that an Operations Bounds Rectangle be defined. Select only one frame (I usually use the reference frame), move to the Frame Inspector Window, and click/drag a rectangle around the feature you want to use for sorting.
Each sorting method works best on a specific kind of feature. Details about this are provided below. Be aware that some sorting methods require a rectangle that is an exact power of 2. To get such a rectangle, pick the size of the rectangle from the Power^2 Dimensions menu under the Operations menu and then control/click/drag in the Frame Inspector Window. If the dimensions of the image are too small for a particular power^2 rectangle, no rectangle will be visible in the Frame Inspector Window.
After an Operations Bounds Rectangle has been defined, pick the sorting method you want from the Sort menu under the Operations menu. Some of the sorting methods will instantly sort the frames when you do this. Others will bring up a dialog box where you can control how the frames are sorted and which fraction of the frames are automatically included in the stack.
Also note that the frames in the Clips Window are automatically sorted as you adjust the cutoff frequency and you can inspect the clips both by moving the mouse over the Clips Window and by holding down the space bar to compare a given frame to the reference frame.
To save the workspace at any time, simply choose the appropriate menu item under the File menu. First, a dialog box will be presented to you telling you how much disk space will be required to save the workspace. You may cancel at this time if you realize you don't have enough disk space. Otherwise, you can choose whether or not to attempt lossless compression and whether or not to save nonstack-included frames (red X) and then may proceed to save the workspace file.
The lossless compression scheme is a little unpredictable. Sometimes it will actually produce a workspace file that is larger than you would otherwise get if you don't attempt to compress the data at all. Fortunately, the dialog box shows you how large the file will be in both cases so you can choose the smaller one. Alternatively, you may opt for the uncompressed method because it is generally a slight bit faster to save and open even if it takes up more space on the harddrive.
Likewise, you can open a workspace file at any point in time. Doing so will completely discard any present workspace, so make sure this is what you really want to do.
The stack that is produced will always be a pure average of the images. This is what you usually want in the case of planetary processing.
There are two sharpening methods, Unsharp Masking and Laplacian Sharpening. I am not going to explain how they work in a deep way here. Instead I simply provide a quick description.
There are checkboxes next to each slider that can be used to pin the slider to the value 0 or 1 when selected. The checkbox for 1 is useful to toggable back and forth to get a feel for what effect a particular slider is having on the image. The checkbox for 0 is useful for completely removing a particular mask's information from the resulting image, say to remove a noisy mask.
Also note that as you change the shape and size of the preview area the number of available pyramid levels changes. The possible values are 1, 2, 3, and 4, corresponding to increasing size of the preview area. When you hit Okay, the maximum number of levels possible for the entire image will be used, because using more levels results in better denoising. If you want the preview to most accurately reflect the final denoising, make the preview area large enough to use as many levels as possible, up to the max of 4. The only cost of making the preview area larger is that the threshold slider and preview area will take more time to update.
Incidently, wavelet shrinkage denoising tends to reduce the constrast of an image, so you may want to increase the constrast after performing this operation.
The following image demonstrates the powerful effect of wavelet shrinkage denoising.
That's it. The stack is complete. You can save the stacked image now and you're done.