Here's something I posted somewhere here a couple of times before:
Think about the sound leaving a vocalists' mouth, and traveling towards the capsule. Ideally, what should be in between? The correct answer is nothing, if you are talking about good sound; just as in mixing (not possible usually) and Mastering (very possible; I've done it in my Mastering room) there should be nothing between the monitor speakers and the engineer's ears except for air (not a console, racks of gear, etc., etc.)
But we have practical applications to think about, not the least of which is the burst of air accompanying a "P" or other letters, or "swooshes" accompanying certain other sounds.
When the condenser microphone was first made, the plan was not for a singer to virtually "eat it" as he sang less than one inch from the capsule. The mics were designed for much more distant pickup of larger numbers of players/singers. So the design included in most cases a high boost in the frequency response, because that would give more balance from more distant sounds, and also included taking into account the physics of proximity effect, where closer instruments might have a somewhat fuller sound.
But what do we all do today, but get right up on the capsule grille cover, and bloody scream into the mic! So of course high sounds such as "S" are accentuated by the built-in freq rise. And of course the burst of air in a 'plosive is much closer, and accentuated by the naturally occurring proximity effect!
So what has been the normal solution? Why, yet one more thing (or 2, or 3...) in between (pop screen), and de-essers after the fact. Does anything seem askew with this photograph? Once again, humans are treating the symptoms, and not the causes.
Of course, the causes are not that easy to treat. The singer would have to move back, and few think they can do that (no pacifier), pops have to be removed (fairly easy now, but still not perfect) or a different type of mic must be used (ribbon? 58/S1-type? Omni?).
Anyway, getting back to that sound traveling towards the capsule, the use of the pop screen in between causes the sound waves to deflect, possibly hitting all sorts of things, and perhaps even bouncing around within the dual layers of the screen, or the grille (already grilles can be bad enough on some mics), or the stand(s), etc. The different parts of the frequency spectrum may well encounter comb filtering, causing the cancellation of certain freq's, or the potential buildup of others, and at best, may arrive at very slightly "wrong" times to the actual capsule front. None of this is very appealing if you want your sound to be "perfect," let alone "good."
The problem is exacerbated by the singer's movements. As we all know, singers will, and probably need to, move around. The cancellations may be quite different given differing positions from whence the sound has emanated.
I recently did extensive (although NOT scientific) testing of this effect, and found that, by looking at many different "sibilant" sounds on a frequency analyser, that no two were exactly the same. Often however, there was not just a rise in the high freq, but TWO rises, with a steep valley in between. The average hi-freq rise (the sibilance) might be about 5-6 dB, but the VALLEY might be as much as 12 dB! This meant that, even though the sound-not-liked was "sibilant," at least a tiny part of that disagreeable sound was caused by MISSING high frequencies, not extended ones! If using a normal de-esser, the "bad sound of the sibilant syllable" would not disappear, only get quieter. I discovered that by applying a corrective EQ, namely approx +6 to +9 of the missing freq range, at a very low bandwidth, centered on the valley's lowest point, made the formerly "sibilant" sound actually better, because, although it was still louder than normal, it was at least MORE NATURAL!
Then, and only then, and if desired, by applying a de-esser the entire sibilant sound could return much closer to normalcy.
But all of this is AFTER THE FACT TREATMENT. Prevention is a much better cure!
So the ONLY pop filters I will use now are the foam ones (as supplied with a new U-87 and some other mics). But these are subject to eventual degradation (Klaus has a Sticky about testing the foam). And this is something you may not want to use one a vocalist that "showers" the area in front of him (another story for another time). Also some complain about theoretical high frequency loss. Perhaps there is some, but I much prefer that to horrible comb filtering and massive treatment later! Also, a slight loss in broadband high frequencies is somewhat preferable to me in some cases, as it counteracts to a degree the high boost built into the mic for more distant usage...for that distant usage, where you need the boost, of course you wouldn't be using any filter anyway!