I tried messing with different shapes and in the end just do a light chamfer on the inlet and outlet, it seems getting the size right for your particular spring /piston/volume is the critical bit.
Size is everything ! apparantly
For a relative newbie like me I found this 3 part article on pyramidair about transfer ports quite educational.
http://www.pyramydair.com/blog/?s=transfer+port&btnGo=
Here is some data I generated with one of my .25 cal UK Webley Patriots (I have 2). Being a retired scientist I just "have to know" how stuff works. The first Patriot was used as my test rig and was altered to take screw in transfer ports. I might add that the internals of the gun have been altered as well but leaving that aside here is some data generated with a soon to be replaced mainspring the only difference being different size TP's. Lube was Moly which results in a 6 f/s increase in velocity over Moly impregnated Krytox which I soon gave up using because the heat of compression "tears it apart" generating pungent low mole weight toxic fluorocarbons in this gun. The data below was all measured under identical atmospheric conditions and is the average of 10 shots and all pellets were sized to identical head and skirt dimensions.
All velocities are with the H&N 19.5 gr. pellet.
TP: 0.138" Vel. 845.9 f/s
TP: 0.147" Vel. 863.3 f/s
TP: 0.150" Vel. 869.4 f/s
TP: 0.156" Vel. 877.2 f/s
What I will say is that anything more than a minute chamfer on the barrel side of the port will cost you velocity. When everything is optimised (not just TP diameter) velocities in excess of 900 f/s are achievable although better accuracy using JSB 25.4 gr. pellets at circa 800 f/s results accuracy of less than an inch at 50 yards. I might add that I head and skirt size all pellets to suit my guns and don't waste any pellets. Why? Because .25 cal. pellets here in Australia must have gold in them to justify their price!
Cheers