The Standard Macros now include two versions of the Do Stack macro. With a little experience it is easy to predict which will give the best results. The original Do Stack macro is tuned to perform best with pictures in which there is sharply focused detail. The new Do Soft stack often produces better results when edges etc. are less well defined, this is often the case, though maybe not obviously so, when you stack pictures of several mega pixels each. The images below are from a stack by Steve Stuart the first was stacked using Do Stack and the second using Do Soft Stack
So try both macros if you are not sure and choose the best result, or even blend the results from both. Unfortunately the Do Soft Stack is more prone to producing halos around brighter objects. If the result is still not satisfactory read on.
Sometimes the Resizing and Aligning stage fails resulting in artifacts like these (from Rik and Don respectively):
Try running the Strict Alignment macro then try restacking:
Notes: If there is rotation between the frames this can cause similar problems, correct this by the manual alignment method or try the Automatic (Shift+Rotate) Stack Menu item. This macro will not work on really deep stacks.
If your problem is caused by parts of the scene overlapping you may
need to try one or other of the following:
1. Follow Do Stack with Weighted Average
Correction.
2. Use the Weightewd Average macro
3. Use the Deep Stack macro set
For most other problems you will need to perform a simple test to pinpoint the trouble. Then edit the stacking macro to fix it.
The most important function to get right in the Do Stack macros is Find Detail, below is a description of how to optimize this. The frames must all be aligned before this function is used, if you have done a trial run using a Do Stack macro this will have been done, if not use one or more of the alignment functions on the Stack menu, or maybe Strict Alignment as above.
I will use a stack by Brian Valentine to illustrate the method, Brian does not claim that this is a good stack it is just intended to illustrate a particular problem. Do Soft Stack produces the following result, the second picture is one of the original frames. Notice the difference in the appearance of the tip of the stigma at the centre of the pictures.
To perform the test click the menu item Stack->Detail->Find Detail, and if you have done a trial run of one of the original Do Stack macros accept the default parameters 25, 3, then 0 for the Soft version and 25, 1, 0 for the other. On completion the screen will look like this:
The picture on the left shows the actual pixels found by Find Detail(25,3,0) these will form the starting point for the reconstruction of the final stacked image. If there are mistakes or not enough pixels here the stacked image will not be verry good. Press the letter 'D' or use the View->Toggle Depthmap menu item to reveal the picture of the depthmap shown above. The colour of the pixel varies according to the frame from which each comes. Light greenish pixels come from the top of the stack, bluer ones from the bottom.
Notice the area around the tip of the stigma in the pictures above there is nothing in the depthmap here so the stacked image is unlikely to be correct in this region. To try to improve this reduce the first Find Detail parameter from 25 to some other value. My stratagy is to perform what is called a binary search, i.e. try half of 25 that is 12:
The gap is closing, but we could do better so halve the 12 again and try 6:
Notice there are many tiny green dots on areas that should be blue, this is noise. CZ can get rid of these tiny errors, provided they are small enough by means of the Stack->Remove Islands menu item, the Do Soft Stack uses a value of 12 for the parameter by default but you can try other values if you wish. The picture on the right shows what happens after Remove Islands(12) I think that there are still too many errors , but notice that the end of the stigma is now more or less complete.
Following my binary search pattern 12 was not quite good enough and 6 has gone too far so try half way between i.e. 9:
Again the second picture shows the result after Remove Islands(12). The end of the stigma is not quite outlined properly but there should be enough pixels here now, and most of the errors due to noise have gone.
The task now is to edit the Do Soft Stack macro. Just alter the line that reads Find Detail(25,3,0) to read Find Detail(9,3,0) and run it again.
Notice the end of the stigma is more like the one in the original frame shown above.
To prove a point here is the depthmap produced when using Find Detail(9,3,0) on Steves stack shown above:
Even without trying remove islands it is prety obvious that noise prevents a low value like 9 being used for the first parameter. You can see the noise in the second picture above in the form of lots of patches of slightly different colours. Here is the result of the same test using a value of 25, along with Remove Islands(12):
Notice that the amount of detail found, and retained after Remove Islands, has decreased but there is just enough to produce the image at the start of this article. Sometimes it is impossible to get enough detail and not get noise so you will have to make a compromise.