This is indeed a good idea to document everything at this stage, as you say a lot of this stuff just seems obvious now (yeah now I've made all the mistakes you can think of several times, plus some you couldn't think of plus some you wouldn't think you could make!)
A sub stack at it's most basic is used to avoid problems with bits of the subject showing through bits which should be in front of it - antennae being the obvious culprits. If you save the two images in the thread you linked and compare them side by side, look at the area where particularly the left antenna crosses the top of the ant's head. In the first stack which was just a straight stack of all the images I shot, this area shows the ant's head
through the antenna. In the second shot I used a sub stack to fix this. This is a common problem in stacking software - in the case where it finds two bits of detail in the same place at different depths in the stack - it does not know which bit of detail should be in front.
I should point out at the start, to do the merging the sub-stacks must line up exactly with the full stack, so don't trim off any alignment borders from the full stack until we've done the merging of any sub-stacks! In fact don't do any post stack processing on the main stack yet!!
Okay so I worked out that the antenna needed clearing up, so I carefully looked through the input images and found just those images which contained the antenna parts in focus. This was a relatively small set of images in this case, 13 images for the left antenna. So load up Tufuse Pro, browse to the directory with the input images, choose just the range to sub-stack. Setup everything else exactly as it was for the main stack. Fuse. Wait a little while, obviously a lot less time than the full stack!
Now we have two images, which should be exactly aligned. One has the whole subject in focus but some issues with 'show-through'. The other has a fix for one of those areas of show through.
Load up both images into your image processing software, I tend to use Paint shop pro for most of my PP work so I load them into that. In that program, and many others work this way you can use 'control'+'tab' to switch between open files, (much like alt-tab to switch between open applications in windows...). So I select the clone tool, at an appropriate size with a nice soft edge, go to the picture with just the antenna. Position the mouse cursor in the middle of the antenna and then lift the mouse up so it cannot move the pointer. Right click to select clone source. Control-tab to the full stack. Left click to clone in a spot from the other image. Now you can zoom in and just paint over the antenna and where it should be and the good antenna from the sub-stack will be cloned in over the top of the semi-transparent one.
Sub stacks seem to work their best if they are small stacks with large well defined details. Sometimes with small details like crossing over hairs it can be a real pain to do the cloning and not go outside the tiny area you are trying to clone. I really must work out some other method to do the cloning in that case, perhaps using the sub-stack image as a semi-transparent layer, cloning, then removing the layer, or something like that!
To check the images are perfectly aligned before you start cloning you can use this technique in PSP. Load both images into your PSP. Place the cursor somewhere in the middle of the picture. Lift the mouse, roll the mousewheel up a couple of notches to zoom into the image. Control-tab to the other image, without putting the mouse back down. Roll the mousewheel up the same number of notches you did for the other image. Now you can put the mouse back down again. Control tab back and forth between the two images, once zoomed in it should be obvious if they line up exactly or not. If they do not then you need to use a slightly different technique to do the cloning, basically pick a point on the antenna which is obvious on both images - zoom in to the same point on both to be able to find something. Clone from that point on the antenna only shot to as close to the same spot as you can on the other shot. If you are a pixel or two out it probably won't show up!!
I use PSP mostly as I've been using it for years, I'm used to the way it works. Although I have PS and I know it's more capable and powerful I find it more difficult to do the basic stuff than I do in PSP... I suspect these techniques will work in PS but you might need to vary them slightly, for instance I'm not sure if mouse wheel zooms by default in PS..