I think the implementation in ACR is very good, but I also believe that there are inherent limitations to how well the chromatic aberration can be corrected. [...] For example, if you correct the horizontal and vertical CA on the corner of a building, a nearby diagonal might now show problems.
If I may throw in my 2c's (more than 3 for actually as they're from Europe )... I hope I won't state too much of the obvious, sorry if it's the case!
First - it goes without saying, but I'd better say it - there are 2 more potential culprits than the 'usual suspect' Transverse Chromatic Aberration : Longitudinal Chromatic Aberration, and fringing around specular highlights due to demosaicing as said Jeff.
For the 2 types of CA, evereybody knows [a href=\"http://www.vanwalree.com/optics/chromatic.html]van Walree's site[/url] I assume?
Fringing and LCA are (first order) constant across the frame whereas TCA is radial (ie zero at center, max in the corners), so the remedy for the latter won't cure the formers - which are what the Defringe dropdown is made for, or am i wrong?
Second, and maybe it's where the ACR CA tool could be marginally improved, the TCA is not always linear across the frame but sometimes quite complex - see for an example the graphs of DPR's lens tests
But I'm essentially talking for others here - my only lens showing significant CA is the EFS 10-22, and dialing something like -17/+17 I don't see much of it anymore (works more or less for the entire focal/diaph range, life's good).