So here goes for my Sunday morning blah :
Maybe we must all look beyond where we normally look, because I don't understand myself by now, where this SQ comes from. It is just too crazy, suddenly.
Last night, track after track, I was and remained amazed of what's all squeezed out for bass, mid and highs, mixed with totally unrecognizable tracks which are so familiar, as if they are played through an other system in another room, or live by the artists themselves, for that matter.
I repeat : I don't understand myself how this emerges. UNLESS - but I did not explicitly check that - it is all because I switched off the OSD texts (not the Wallpaper Coverart itself). So among all my by now wildest settings changes, that OSD Text being off, is really the only common denominator.
Help.
The last two days I played with :
Q1 x xQ1 = 30x40.
Q3,4,5, = 1,1,1.
SFS = 20.
ClockRes = 15ms.
Custom Filter = Highest for 705600.
And just mentioning : Appointment Scheme = 3-5.
And thus OSD Text = Off. Wallpaper (front and back) On.
The OSD Text thing should not be underestimated, as it will do something to the OS itself but also to the Sound Engine which does not need to deal with it (in the loop of Playback, which is what it comes down to). Also :
I did NOT solve the issue of the Volume not working when no Wallpaper Covertart is used since 2.08, because exactly there the SQ change may be implied. So instead I offered the old means of (slow) volume change (with a .tst file, see Release Notes of 2.09).
Anyway, and I think you all may agree, the SQ change is so crazy, it is crazier than ever before. And where you guys may see the largest change between 2.08 and 2.09, for me this is way more difficult because somewhere between 2.08 and 2.09 I will have done something which I implemented on day X and which in aftermath is not recognizable (what it has been). So I G-D don't know myself !
It feels (also on behalf of you !) like "we" played with some setting just for trying, and now something popped in all of our systems. And now all is good (or at least changed vastly). IOW, no matter what settings you dial back into, the base of the SQ remains as it is.
The level of detail is inifinite.
The bass is infinitely loud.
The highs do not observe like wrong - they don't observe at all (brain has to time to observe).
And this gag remains for me : when I am sure the mids stay behind, something happens which shows the mids as the most profound ever.
I also keep on feeling that all "frequencies" are infinitely detached now. So maybe you reall me telling about the electric bass player and his own amplification (which always is true in real life) ... this now happens to everything. But meanwhile each shifted something like an octave in their own domain.
All things which can't be.
But, other tracks with acoustic instruments, like e.g. Clark Terry's trumpet, sound too hard and harsh with these settings,
So Yes Richard, these things happen as long as we can't find our "setting". This is very very tough to judge for real merit because who tells whether that Clark Terry is a decent recording ? I mean, man, if things change so wildly then we must re-learn to listen and this includes re-judge. But of course Clark Terry is fine (not that I ever heard of him) and so we wander for that other setting(s). Well, I just gave mine so maybe that helps you and others (but two days in only).
Q1: 14x40, Q3/Q4/Q5: 0/0/0, SFS: 0.7, Clockres: 1ms, Coustom Filter: High (2nd in the list).
As I can see, all that is pretty different to other reports here.
Richard, maybe not. I had the same for two (other) days with only the change of SFS to 0.9 and the highest custom or native Arc Prediction. This is not miles off from yours, so that really should have something, but also for sure not for all (and this is about the SFS of 0.7 or 0.9). So this could be the wildest stereo imaging effect, but if it does not work for all, then good bye (I forgot where it went wrong for me). The crucial general one is the 14x40 here, which is a kind of intuitive change because of how the dials present it, but anyway that for a base is not wrong at all.
And of course when I see that working, I overdo it with a 30x40 ...
That this 30x40 for "base" does not work out for everybody with descriptions as "
bland" ... I tried to observe that and I think this could be about the same I just told about :
highs ? what highs. And it could be the largest pitfall ever.
So yes, completely true, especially in retrospection it could be that the highs disappeared (sort of). However, just
because I have my ever test signal music, I just know how much of highs there is, but now it looks like highs where highs must be. This may be related to my high sensitive mid/high of 118dB (which is also crazy of course) but the mentioned "infinite detail" - which btw is also about "detail where detail should be" can only spring from "perfectly rendered highs". In other words : it is also very unbelievable how no disturbing highs are present any more anywhere (someone told the same but I forgot who - maybe Nick) which somehow seem to give room for the not-fake detail and which very much includes the again "infinite" clarity which is suddenly there everywhere. Mind you, an album like Joe's Garage (Zappa) has always been a strange one regarding the highs and which always lacked fundament. Listen to it now. Ultimate clarity which ... now has fundament ? no, I don't think it is that what happend. But a boat load of hash disappeared (at least that is how it comes across to me).
The effetcs of the above should be encouraged for by the 30x40 of Q1 but with a quite special side effect : removing fake detail hence noise, now bringing forward what should remain (whatever that is). So, this is actually back to my hundreds of times expressed curiosity : The buffers should all be as large as possible and not as small as possible because the latter theoretically imply noisenoisenoise (and super much overhead). And for those with a pocket calculator : 30x40 = 1200 while 14x1 = 14. So something is more than 85 times quieter.
Previously we needed this "harpening" (same as with photos) to perceive sufficient detail, but while sharpening is done by adding noise, by now, everything in the environment improving, the sharpening expresses as noise. Remove that and the native sound remains.
Blahblahblahblah.
But I still don't know what happened, where.
Anyway, going this direction a kind of obviously also highered the CockRes (this one is dangerous because not really in the same realm, but for (less) overhead matters it should help) and what remains is the Driver Buffer size and don't I also recall a Kernel Streaming buffer size somewhere (I never touch that one and maybe it can't even be changed - I forgot). Of course we have a related XTweaks setting (the Nervous Rate) and the SFS itself at 20 should be high enough not to bother.
Most crucial could be the 30x40 which now even works (as I said earlier, I thought that it could not, ever back that is) which I could try to give a maximum of way more. We must of course be able to run into limits, right ?
Did I say
Help ?
Peter
