YouRei HP/LP filters, Nebula

VST, AU, AAX, CLAP, etc. Plugin Virtual Effects Discussion
RELATED
PRODUCTS

Post

a listening test could be done for this. the only problem is that to compare interpolated vs stepped, you would want to go directly in-between actually sampled positions with the interpolated version. so if i sampled 50hz then another at 60, you would want to put the program filter position at 55hz. then the issue is that the nearest position with the stepped filter would be a 5hz difference (at either 50 or 60hz), so it will sound different just because of that.

so then i would want to make special programs just to test this. it wouldn't take TOO long... i could do them without dynamics, since we are only after the answer as to whether or not interpolation between steps hurts the sound. i would use the same distances between steps as i did with the released set. so i would do one interpolating program with an HP at 50Hz, then another step at 60Hz, and for the test i would put the filter in between, at 55Hz. then to compare, i would make another program with the filter on the unit actually sampled at 55hz. i would test both LP and HP filters, both together and separately.

i could process one or more ~5 second clips of a full mix (or a few mixes), with maybe 10 different combinations of filter positions, using programs made with and without interpolation. i could take the processed wavs, zip/rar them then put an extra zip/rar in that one with the answers to which is which for all clips. that archive would have a password which i would give later. if you guessed consistently which was which, i would admit that freq position interpolation is just plain wrong with Nebula. you don't have to bother, but i'm giving you an easy way to possibly make your case.
Last edited by Cupwise on Tue Aug 14, 2012 2:56 am, edited 1 time in total.

Post

Cupwise wrote:i'll be right here, waiting. because so far you haven't offered one shred of evidence. i don't expect this to change at this point. you are only using your 'feelings' as your argument that there 'might' be something wrong with interpolation. i'm using facts. as a developer of the platform for going on 3 yrs now, excuse me if i happen to think i know a bit about the subject. ;)
Are we having an argument? Am I challenging what you know about the subject? Did I ever say I think you're wrong?
but while we are on the subject of 'what if' scenarios, why don't you imagine if i were an algorithmic vst plug-in developer, and i were to release a HP or LP filter that were stepped at 10Hz steps. what would the general community here at KVR have to say about that after they tested it? they would be like:
"wtf?" -KVR community
Many of the UAD EQ's step just like their hardware equivalents. The audio engineer community has not said, "wtf."
now i regret it because i feel like i just played along with completely unfounded paranoia, instead of taking the chance to dispel that paranoia.
You can dispel the paranoia by allowing people to experience it for themselves. People learn through experience and experiencing this for themselves will hold more weight than anything you might say on the subject.
instead, how about if i just suggest that you are wrong? or point out that you still haven't brought any proof to this discussion. if you want to claim some process in a program is flawed, bring proof.
I never said it's flawed, I merely said I prefer the stepped versions because I have no need for interpolated versions and thus don't need to risk the possibility of something being flawed. That doesn't mean I think it's flawed.
it's more work for me, it adds more clutter to the program selection menu, it can possibly confuse some users who now have to learn what the difference is and probably don't care anyway (and for good reason), it adds more files to deal with, and it's entirely pointless.
That's entirely fair! I get that! Yes, since all those negatives exist, I definitely agree that you shouldn't offer it in the future. I'm just glad you did it with these so we can experience the difference (of lack of) for ourselves. :)

Post

Are we having an argument? Am I challenging what you know about the subject? Did I ever say I think you're wrong?
no, you just claimed interpolation 'might be' bad when it isn't. i'm using facts based on what i know and so far you haven't used any.
Many of the UAD EQ's step just like their hardware equivalents. The audio engineer community has not said, "wtf."
not the same. i said 10Hz steps. if someone sees a filter with 10Hz steps they definitely will wonder about it.
let's say i used the acqua platform to release this filter to the more general (non-nebula) public. acqua still uses nebula's guts, but you don't need the full nebula program. so if i did that down the line, and let's say i only released this filter in stepped version, more people are going to say 'wtf' about that, than if i had only released an interpolated one.

that's also beside the fact that those UAD EQs are still accurate to the hardware they model, just like my interpolated filters are in this case.
You can dispel the paranoia by allowing people to experience it for themselves. People learn through experience and experiencing this for themselves will hold more weight than anything you might say on the subject.
fair enough. but i could also just explain why there isn't an issue, as i've been trying to do. also i could issue a challenge to a listening test as i just did, and get the facts out there once and for all, and save myself the agony of people wanting/requesting stepped versions of all future releases, when stepped versions are pointless and cost more work/time i could be spending on other things that they might want.

..risk the possibility of something being flawed. That doesn't mean I think it's flawed.
what's the difference? you are worried about a possible flaw. so you don't "think the flaw is there". you just "think it might be". if i could prove that there is no flaw, then you would still be wrong in both cases. and if it were there, it would also be in every dynamic program you use. so why aren't you questioning them, or worried about them?

That's entirely fair! I get that! Yes, since all those negatives exist, I definitely agree that you shouldn't offer it in the future. I'm just glad you did it with these so we can experience the difference (of lack of) for ourselves. :)
ok but the point is that i want to release the best possible releases. if YOU could prove to ME that there IS an issue with interpolation, i would gladly look past those negatives and make sure to provide stepped versions always.

so no, i'm not arguing with you. i'm giving you (or anyone else) a chance to present factual evidence that there's an issue with this. i'm also trying to explain that a lot of people's concerns about this are based on no factual evidence. i'm using this thread as a place to make my point on this issue, so don't take it personal.
it kind of influences what i do, so it is in my interests to address it.

:)

Post

Cupwise wrote:also i could issue a challenge to a listening test as i just did, and get the facts out there once and for all
Sure, that would be great!
what's the difference? you are worried about a possible flaw. so you don't "think the flaw is there". you just "think it might be". if i could prove that there is no flaw, then you would still be wrong in both cases. and if it were there, it would also be in every dynamic program you use. so why aren't you questioning them, or worried about them?
Again, it's a matter of need. If I believed having the in-between frequencies would provide me any benefit at all, I'd be much more interested in trying them out. At this point, the only reason I'll even be trying the interpolated filters is because I want to hear what you've been talking about. :)

Anyway, I'm glad we're finally on good terms, now I won't feel sheepish about buying your products. ;)

Post

Uncle E wrote:
Cupwise wrote:also i could issue a challenge to a listening test as i just did, and get the facts out there once and for all
Sure, that would be great!
ok. it's going to take a while for me to set this up. i'm trying to actually make progress with this notch filter. if you have any mixes of yours that you'd be ok with me using in the test, you could take a 5 sec clip out of it and send me that in a .wav. 96khz 32bit would be swell, but 44khz and even 16bit would be okay too. if so, pm me. it would speed along the process. i have some stuff but it's mostly all sample/synth based, so if you want something more organic to listen to and have something of that nature...

Uncle E wrote: Anyway, I'm glad we're finally on good terms, now I won't feel sheepish about buying your products. ;)
look, i just took the opportunity to list the reasons why i think there's nothing wrong with a properly sampled filter using interpolation. you kept responding so i figured that gave me a good platform for my side of the debate. i care about it because i have plans for some other stuff that will utilize interpolation, where it will be more important to the final thing- just like how interpolation is 100%, completely and utterly, totally and undeniably important to any and all dynamic programs, which simply couldn't exist without it.

so yeah, i'm going to defend Nebula's ability to handle that well when done right.

Post

ok, well, the 'initial buzz' over these filters seems to have died down, and only one person showed interest in the listening test (i promise i wouldn't have been a d*ck about it! ;) ), so i don't think it seems worth my time to do it now. i don't want to look like too much of an ass when nobody participates, not to mention the time i would have wasted that i could spend working on sampling.
so instead i'll take the 'look like slightly less of an ass' option and retract my offer of doing the test.

so i think i might just set something like this up for the next release with the notch/peak filter. and i'll put that out there on release to try to help promote it or whatever. besides, as you (uncle e) pointed out, this release does have stepped versions (at least for the solo filters), so people can compare those to the interpolated themselves if they feel like it.

Post

Cupwise wrote:ok, well, the 'initial buzz' over these filters seems to have died down, and only one person showed interest in the listening test (i promise i wouldn't have been a d*ck about it! ;) ), so i don't think it seems worth my time to do it now. i don't want to look like too much of an ass when nobody participates, not to mention the time i would have wasted that i could spend working on sampling.
so instead i'll take the 'look like slightly less of an ass' option and retract my offer of doing the test.
hehe, no problem! I'll buy your impulses and run the test myself. I use PMC TB2's, B&W 602's, and Tekton 45's for my monitoring.

Post

ok! the only issue is that if you select a frequency with the 'stepped' version, and then pick the same exact freq with the interpolated, there won't actually be any interpolation. so you won't really be comparing interpolated to stepped. you would have to select a frequency between sampled steps to get any interpolation, and then you will have more filtering with one than the other, which will cause them to sound different.
so you wouldn't be able to make a 100% direct comparison.
but in general you could listen for anything out of the ordinary, such as artifacts, phase issues, anything unwanted with the interpolated ones.

so yeah it would have been more ideal if i had made up special programs just for the test..
maybe i'll do that and just give those to you, then you can compare those.
if you send me a pm with your email i could email you a link to a few of these small test-only programs, after i make them (if you're interested).

actually if anyone else is interested they can pm me about it too. it could be a few days before i reply with a link.

Post

Stepped means no frequency adjusting right?

I feel non adjustable programs are the best way to use Nebula. Nebula is very akward for tweaking.
Intel Core2 Quad CPU + 4 GIG RAM

Post

electro wrote:Stepped means no frequency adjusting right?

I feel non adjustable programs are the best way to use Nebula. Nebula is very akward for tweaking.
*sigh*

you can 'feel' whatever you like. i've just gone through a whole big thing here and your comment basically reveals that you haven't read any of it but still felt the need to give your opinion without any details or explanations or facts to back it up.

maybe you have a program that doesn't work well for 'adjusting' as you put it. maybe it's a free thing made and released by a member of the community, not one of the 'official 3rd party devs'. maybe its an older one from the early days or one of the earlier things from the official acustica library which were sampled when nebula was first conceived. the whole sampling methodology has improved drastically since then. whatever the case is, which i don't know because you gave no examples to back up your blanket statement, my only stance is that there is nothing inherently 'wrong' with nebula's interpolation, nor is there any actual 'limits' on what it can do with this regard. well, there are, but those limits aren't applicable here, with my filters. the truth is, that any 'bad' filter program you've ever seen or heard could have been a good one of there were more samples taken when it was made, to provide more resolution.

this is just a true fact, and anyone who disagrees is wrong. the sky is also blue. it's just a matter of having more samples. again, i'm not going by 'feelings', i'm going by first-hand knowledge and facts.

Post

so i made some 'blind testing' programs. interpolated vs not.
http://www.cupwise.com/yourei-interpol-blindtest.zip
anyone can grab these, they should even work with free Nebula.

package has 44.1khz, and 96khz. an info txt explains the setup. there are 4 different combos of positions with the HP and LP filters. they don't have dynamics and only have 3k and nothing is adjustable. i never released a demo for the set, and well, this is a 'demo' in a way, but you won't get much mileage out of these beyond testing. they won't be up forever. in about a week i'll give away a password that unlocks a .zip in the archive that contains a txt that has the answers as to which is which.

i don't really mean to clutter up the forums here with people posting their guesses as to which is which, i'm not trying to host a gearslutz style listening test. it's mostly for people who have any interest in the topic to just check it out for themselves. if you can't really tell a difference but just end up making some guesses, that probably means you didn't hear anything wrong with the interpolation. you could still get 'lucky' and guess even all 4 correct, so if you are just guessing i think you should take that to mean the interpolation is fine. but, if you really think that you hear something bad about one or the other then by all means post, because it will make your word more reliable if you say it before i release the answers.

i really wouldn't have bothered and had changed my mind about doing any test at all, but then i realized i could give out 'blind' programs and thought that was neat, so there it is. a different kind of blind listening test. you can use the filters on whatever material you want, and do the test yourself. if someone actually manages to pick out the interpolated one 4/4 times due to some perceived flaw, i might devise one further test for them, and i would then probably make sure to always offer stepped if they passed that test...

Post

There's something wrong with the zip file, I can't get it to open on my Mac.

Post

Uncle E wrote:There's something wrong with the zip file, I can't get it to open on my Mac.
hmm weird, i downloaded it and it opened fine. i don't know anything about macs. i have had someone tell me that .rars wouldn't work for them on mac so i started using only zip. never heard of any problems with them.

maybe because i used winzip 64bit? i always used 32bit but just grabbed 64.
anyway, i re-zipped it with 32bit winrar, and re-upped it. try it again. let me know...

Post

in case anyone actually listened to these and wants to know which was which, here's the password to the answers.
3tgrg4h5hht4s

Post Reply

Return to “Effects”