I cut out the three plates - breast, midcollar, and back as before. I made an error on the breastplate's neck hole - since it wasn't fatal I decided to ignore it since this is a prototype unit anyways. However, I did fold it a bit differently to try to hide the error, as shown below. Purple line indicates approximately where I should have made the original cut, which I hid by progressively shrinking the width of the fold.
As previously mentioned, I attempted to copy the folded & flared look shown in Peter Connelly's drawings of lorica segmentatae. This turned out to be easier than I expected, although I still managed to make a mistake or two in my rush to get these pieces done. For the first plate, I folded the edges as described in this post, except that I switched to the planishing hammer much earlier -- when the fold was still about 45 degrees away from being finished. This resulted in a much neater edge, as seen above.
To flare the edge, I positioned the plate on the anvil such that the fold was hanging over the edge a little bit. (Sorry, no action shot. Maybe next time.) Some gentle hammering along the edge of the closed fold was enough to flare the metal over the edge of the anvil, resulting in exactly the desired result. The trick here is to move the plate along the edge of the anvil to follow the curve of the neck so that you end up with a flare with a constant width. Results are illustrated below.
On the second and third plate, I speeded up this process by starting the flaring before the fold was completely closed -- basically I merged the flaring passes with the last couple of fold-closing passes. This turned out just fine, so I will probably continue to do things this way from now on.
I had originally intended to practice rolling edges a bit more, but since that takes much more time & effort I decided to skip it this time. Perhaps I will practice it on the edge of a lesser shoulder guard plate later on.
When I got home and compared my two sets of backplates, I realized I had made a very basic error -- my folded edges were at slightly different heights because I didn't bother to check the second one against the first one. I guess if I had cut out and hammered both left and right sides at the same time, I wouldn't have made this mistake, but it's something to keep in mind in the future.
Comparison of the mostly-assembled first shoulder and the new neck plates.