Once he knows he has the structure captured, he can go back through the settings through of the field of play and shoot the action shots.  I imagine he is just constantly shooting most of the event. 

I’ve produced a few of these mashups of space and time, most noteably this one with The Old 97′s and the Sarah Jaffe Band at the Doug Fir in Portland, OR – http://photosynth.net/view.aspx?cid=356de7b4-8eb4-4504-997e-ae186b92d46b – Sarah Jaffe is watching her own sound check and most of the people in the image appear twice.

Peter’s work has definitely inspired me to do more.

@GigaView on twitter

Now, any modern version of photoshop will be able to align all the images for you, but with that many, it’s going to take time.   depending on your computer speed it might take anywhere up to a few days of processing for the hundreds of shots included.   I’d guess the hardest, most time consuming thing will be choosing which shots to use.   You’ll have bursts of so many different events, choosing which events to unclude in the final shot will be hard, then choosing the frame to use…   It will all take time.

The actual “work” part of it is probably fairly easy, but I’d imagine that every single little tweak you make to the image would use so many computer resources than a simple levels adjustment could take a few hours for the computer to catch up.

1.He really doesn’t need to pan, just as long he could get the whole field(arena) in one shot.
2.He is using a fisheye, which would mean that if he panned and stitched together than there would be double arcs(line distortion would be doubled and would create a W effect on the bottom horizontal lines.
3.panning would make his post work flow alot harder, it would be much much easier to shoot from a fixed position and than layer and mask in photoshop.

The only reason he would have to pan is to catch the important person in the middle of each frame(as to not distort the person because of the fisheye, the middle of a fisheye is not distorted very much but the top, bottom, right, and left sides of the frame are heavily distorted.

But even if he was panning, it still wouldn’t be too hard to put all the images together. All he would have to do is take an establishing shot(a shot showing the whole field and with the least people) and then take first frame with an important person in it, put it under the establishing layer, drop the opacity of the establishing layer to about 20%, so that you can see through the establishing layer onto the first important person layer and than mask the important person into the establishing layer.

because the panning was done from the same fixed distance, the subject(players) really wont be to different in size. and even if they were different sizes you could just layer the wrong size image ontop of the right size image, drop the opacity of the wrong size layer(layer 2) so you can see through to the right size layer and than re-size the wrong size layer until the subject of the wrong layer matches the size of the right layer and then just mask.

The whole idea is really just an elaborate multiplicity, really not very hard, and not nearly as time consuming as they say in the video.

Not to discredit the photographer, these photos are awesome and are very well done.

however I am only 17, photographing for just a couple of years and I could pull this off pretty easily.