Here's my experience with this. C5 will certainly make the most difference. The cap in your picture can probably be improved upon. I find Audience "Auricaps" can sound quite good. I like the sound of the higher voltage versions more than the lower voltage versions (450 or 600 vs. 250V). The green Russian cap that Tubtec showed you is magical in a U47 clone circuit (.47mF/250V). The Bosh MP that was mentioned is in the original U47 and can still be found, but they're quite costly now. Those are my three favorites, but I've been experimenting with SMR caps because they are the fastest discharging caps currently being made. Results, so far, for the SMRs are promising. Teflons sound good, but the voltage restrictions, insane cost, and very long burn-in times have kept me away from trying them as output caps.
C2 can also be heard, if you're really paying close attention. When doing a U47 clone, I ended up with a ceramic - yes, I said ceramic - cap there and felt that it moved the dial 1% closer to my target sound when compared with the HiFi poly cap that I had started with. Also, I tried paralleling C0G ceramic caps for a mic's output because they are said to have "no distortion". That was a DISMAL failure! So bad I couldn't believe it, but I don't know why because the result was so awful that I wasn't willing to put even one more second of my time into investigating them. However, I do like C0Gs and NP0s quite a bit for AC bypass and related applications (such as C2 in your diagram).
Just remember that in comparison to changing the capsule (about 85% of the mic's sound) or the transformer (about 10+ % of the mic's sound) , or the tube (about 5% of the mic's sound), changing the caps is only going to make very subtle differences - except if you parallel C0Gs on a mic's output, but I've already warned you off of that.