A few people have requested an explanation of how the action montages I posted here were created so here is a walkthrough of my process. This is the first time I've tried to delineate my post processing so let me know if there's anything I can do to make it more understandable.
When I shot both sequences I wasn't thinking about creating a montage like this so I was hand-holding the camera and moving between shots. This means the framing and orientation of the camera will vary with each shot, so I know that I will have to use some cloning and transformations to combine the images in a believable way.
With forethought I would have used a tripod and framed the scene to capture all the action without having to move the camera at all. That would make post processing a lot easier but I think it would have been hard to use a tripod in this environment where you can't predict the subject's movements or timing.
Click an image to view it larger.
Open all your images in Photoshop. I've got these four vertical images that I shot handheld so they're not going to line up easily.
The first thing we need to do is bring all the images together onto one canvas. To do this we will pick the first shot in the sequence and increase the canvas size. I think 5500 pixels will cover it, we will need a little room to move layers around.
Now we want to drag all the images onto the same canvas. You can just select the arrow tool and drag the images from their canvases into the first. I have ordered the layers so that they run from left to right top to bottom.
Only one image is completely visible at this point but I can see enough to make some basic alignments by moving the layers around. Here I'm looking at the railing in the bottom background and the brick wall on the right to give me some clues about preliminary positioning.
Now the fun part begins. To start revealing the second layer we need to add a layer mask to the top layer. To paint on the layer mask I use a combination of the brush and gradient tool.
You can see in the above image that I've masked out enough of the top layer to see the second figure on the layer underneath.
All good, but immediately we can see that the layer below is slightly offset from the way the white wall near the figure's hands and the railing beneath his knees don't match up.
To fix this I'm going to apply a transformation to the second layer, in this case a slight rotation and skew fixed the problem.
To reveal the third figure we add a layer mask to the second layer. I'm using the brush here to 'paint in' the guy because I know his hands are going to be 'behind' the second figure's hood so I want to be accurate.
As before we add a layer mask to the layer above to reveal the layer below.
This time I've used a circular gradient on the layer mask and I can immediately see the last image is badly aligned with those above.
Once again a transformation on the layer fixes the alignment and we can continue to reveal the last figure. Again I use the brush to carefully paint in the areas where the figure overlaps the one above. In this case his lower legs and feet.
Now the image looks finished. All four figures are complete and it looks pretty convincing.
When you get to this stage it's time to zoom in and look for the less obvious errors.
Here we can see a misalignment between two images showing as a line.
To fix this I use a combination of smudge, clone and blur tools. It doesn't have to be too convincing in such a small area, though you might want to do a better job than I did here!
Finally you can crop the image to eliminate the overlapping edges.