Here's the original (bad) pouch design that I came up with:
I folded the edges of the canvas over and sewed it with a wide hem (or whatever you might call it) so that I could pull the cord through it like a drawstring. (What you see here is not the sling as such - it's just the loops that the sling lines tie to.) I broke three #16 sewing machine needles to make this, mostly where the edges are gathered and folded to make the curves. The canvas itself was kindly donated by my wife, who paints.
My intention was to make it easy to centre the pouch on the sling when I was tuning it. Instead of changing the length of both sling lines (which were tied to the two loops), I merely had to change one and then reposition the pouch to the centre.
It seemed like a good idea at the time, and it was easy to centre the pouch. However, it didn't perform very well. It was difficult to control the release timing. It kept spiking the ball a few feet in front, and I had to keep shortening the sling (down to about 75% of the throwing arm length) and straightening the release pin (to something like 10 or 15 degrees). I eventually was able to hurl a tennis ball about 25 meters, but when I tried a denser payload (rock the size of a walnut) it often failed to leave the pouch at all, just hitting the ground *whap* in front of the treb.
I guessed that the pouch had too much concavity to it, especially while under tension. A dense payload like a rock made it worse as the edges pulled tight, like a drawstring. It probably would only work well with over-sized payloads that hung out over the edges, like canned goods, or maybe softballs.
New Improved Pouch
Scan scaled to 100% of actual size. The red lines represent a 1 centimeter grid.
This time I used bias binding tape on the raw edges to keep them from fraying. The 3 cm slits at the top and bottom were merely overlapped a bit, and sewn with a straight stitch down the centres, then a zig-zag stitch on both cut edges. I sewed the bias tape around the sides after the slits were sewn. Since bias tape itself is cut on the bias (ie. the weave is 45 degrees to the long dimension), it's easy to follow the curved edges of the canvas.
The triangular rings were a length of 12 gauge (I think) copper wire bent around a piece of angle iron, soldered along the overlap (concealed by the canvas) and trimmed with side-cutters. I chose a triangle so that it wouldn't tend to cut the canvas where it comes out.
Here's a photo of the same pouch from the side, opened up. It shows how much concavity it ended up with:
It was a bit deeper than I planned on, but it seemed to work fairly well anyway. My ranges improved by about 20% and the grouping of the hits was smaller. The payload always left the pouch.
New pouch in launch trough, loaded with a road hockey ball: