I'm normally of the opinion of "use a good film cap- any film cap- and be done with it"... but there's certain situations where there IS a quite noticeable difference between caps.
We were doing a custom crossover for a Tannoy Glenair (15" dual-concentric)- first go-around we had used a standard Jantzen poly cap (IME and IMHO, completely interchangeable with Dayton poly caps, as far as performance and quality are concerned). During the process, we decided to try the Jantzen Z-Silver cap in just ONE position (the main cap feeding the horn/HF driver in the dual concentric driver).
THAT definitely made a difference. Of course, this is a $10K speaker system, with very precise design and very high resolution... and it was being run by a VAC Phi 110i integrated amp ($22K) and a Sonic Frontiers CD player ($8K) at the time. Dunno if the same difference would have been as apparent, with more "normal" electronics"... but boy, did the apparent soundstage get more "realistic" in perceived dimensions (it sounded "more like the original room", to all of us... yeah, that's a really touchy-feely way of putting it, but I think most people can tell what I mean by that)...
Also, there are certain circuits that are MUCH more "vulnerable" to cap quality issues... I've found that circuits that have proper DC bias on the cap (at least 25% of the rated DC voltage across the cap at idle) seem to have MUCH less sensitivity to cap changes, than those with NO DC bias. Makes sense... it's about saturating the dielectric, to counteract the possibility of hysteresis effects at low signal levels... once the dielectric is "full of charge" from the DC, the AC signals don't have to worry as much about the dielectric "absorbing" signal and distorting the waveform...
Regards,
Gordon.