-
Once again - velocity X-Fade and phasing
Hello all, I'm aware this topic has already been touched upon several times, but perhaps someone can shed some light on some of the insecurities I'm still having regarding the issue. I understand that when crossfading between velocity layers of solo instruments, phasing artefacts can occur, and that it is advised to use expresion CC instead to perform dynamic changes. In practice- let's say I'm creating a line with the SE1 solo horn and want to gradually rise to the f area where the patch starts to dip into that forced "growly" timbre. The threshold of the SE1 horn would be at velocity value of 109. Is it a viable technique to turn off velocity crossfade, set the key velocity to 109 or more and draw a gradual expression line in the CC lane for this one particular MIDI note? Also, what does velocity crossfade actually do within a single velocity layer? For example, the SE legato patch only has two velocity layers - 0-88 for mp and 89-127 for f. However, using the velocity crossfade slider in the whole area between 1 and 88 still has an audible effect. What is actually happening here "under the hood"?
-
There is no phasing with velocity crossfade in sustaining vsl instruments. They are different samples, and do not phase at all. I have done extensive use of velocity crossfade with almost all the instruments - solo as well as ensemble - and they are remarkably free of phasing. Nevertheless, the crossfade is most effective in 4 velocity samples, however, you can use it with the two velocity as well- for example, the horns a8 sound better with velocity crossfade than they do with the pure original samples for certain crescendos and diminuendos, and dynamics such as mezzoforte that are not actually sampled and the difference is not distinguishable, especially in an orchestral context.
The use of expression controller is purely an amplitude change that is done for adding to the timbral dynamic changes. It is very important, but can never subsitute for an actual dynamic sample crossfade because that involves timbres as well as levels. It is best used AFTER programming the velociity crossfades to do additional level changes - especially extreme diminuendos - within a certain basic timbre.
-
Thanks for your reply. William.
Indeed, I have tried simulating a mp-f crescendo by setting a high key velocity and then gradually increasing the expression value, but the result really wasn't all that pleasing. The effect of hearing a vivid and sibilant timbre at an unexpectedly low volume doesn't sound convincing at all. So, I don't really see another option with non-dynamic patches other than using velocity crossfade. I mostly work with ensemble patches for strings and brass and I generally use velocity crossfade and expression the way you suggested - general sculpting of the phrase with vel.x-fade and additional touches with expression afterwards. The reason I'm bringing up the whole thing is that I just recently became aware of people pointing out issues with x-fade and solo instruments - with VSL and also libraries of other manufacturers - and started to investigate this a bit. But honestly, I don't seem to hear a prominent or worrying amount of phasing or doubling/chorusing effects, while some other users seem to have serious issues with it. I felt here and there that it can perhaps very briefly become a little bit wonky especially with oboes and clarinets when crossing certain thresholds, but that's all I could ever spot and I thought it was actually pretty negligible. I also feel that reverb helps further masking the effect. But you know how it is - sometimes you're not sure how much you can trust your own hearing at a certain moment in time. After all, I remember messing around and practicing for years before really understanding and being able to hear audio compression. So there's always the question - is it something that isn't there or I don't mind, or something I haven't learned to listen after yet?@William said:
The use of expression controller is purely an amplitude change that is done for adding to the timbral dynamic changes. It is very important, but can never subsitute for an actual dynamic sample crossfade because that involves timbres as well as levels.
-
There is no phasing with velocity crossfade in sustaining vsl instruments. They are different samples, and do not phase at all.
Maybe 'phasing' isn't the right word, but there's an audible artifact when multiple layers play simultaneously. I don't imagine there's any way around that, as long as you're playing two oboe-samples at the same time instead of one sample, it's going to sound like two oboes instead of one.
-
If you're generally working with ensembles, then the phasing effect is less pronounced than with exposed solo lines. You might have decent luck playing an ensemble patch crossfaded and then adding a solo player on top and using a slightly different curve. Generally though when the whole orchestra is playing, the phasing effect is really not noticeable in anything but the most exposed instruments. Some instruments hold up really well to velocity crossfade too, like flutes. I've played around witih a lot of the ensemble patches too and compared the velocity crossfade version to the dynamic recording and you really can't tell them apart.
I've found MIR is brilliant at hiding weak crossfades too!
Anyway your results might be even better after upgrading to Standard instead of SE because certain patches have more velocity layers in Standard. Legatos still only have two, but the legato patches are sampled in half-tones in the Standard and Extended libraries and many other patches get extra velocity layers.
As for what's going on "under the hood", I couldn't tell you, but it doesn't sound to me like the velocities actually start to mix until you get closer to 68. It sounds, to my ears, like you go from somewhere around 1-30 or 40-ish with just volume being affected, then the second layer begins to get noticeable right in the middle, and then you have just the top layer audible after a certain point.
-
JImmy that is true about the reverb. Also, maybe I should add that I never worry about phasing on solo instruments because I only use them as instruments in an orchestral context, so if there is some tiny amount of phasing on instruments which have very straight tones - like clarinet solo, solo non vibrato trumpet or trombone maybe - it is not detectable because another instrument is playing at the same time. Maybe if you are going to do a clarinet concerto recording, with a very dry sounding solo upfront, it might be an issue to worry about a little. Though there you would want to "show off" the dynamic samples anyway.
However I have to add - the recorders are one solo library that actually does have zero phasing no matter how you crossfade them. I determined that for sure on my recording "Song of the Forest" http://www.vsl.co.at/Player2.aspx?Lang=1&DemoID=5995 which Herb suggested more crossfades on, which I added and had no phasing.
Also, on the solo instruments it is good to use dynamic samples a lot, just because they sound so expressive in themselves and are in general more natural sounding. And depending on the number of velocity layers on a particular instrument, you will have a smoother dynamic change with the dynamic samples than for example with a two velocity crossfade. Though with ensembles even that is different, as I have noticed that the 8 horn ensemble legato - which has just two velocity layers - benefits from using crossfade very freely.
-
@William said:
... the recorders are one solo library that actually does have zero phasing no matter how you crossfade them.How is that possible, unless it leaps suddenly from one velocity layer to the next, instead of fading between them? Don't cross-fades by their very nature play two layers at the same time? Isn't playing two layers at the same time what Jimmy is refering to when he writes, "doubling/chorusing"?
-
Yes, I looked up the value 109 in the histogram. I'm not truly sure what a split point is either, though. The VIPro manual states that 109 is one of the most common split points in VSL sample libraries. And the forte range indeed starts at 109 for a lot of instruments. I would assume that a split point is the first value at which a new velocity layer plays on "full" vithout any crossfaded amount of the previous layer. At least that would make sense, I guess. Perhaps the crossfade script isn't always 100% spot-on and possibly overshoots the split point by a point or two, which causes samples slightly above to stil sound chorusy. Just a wild guess here. But ultimately, it's probably better not to over-obsess over all of this. As Casiquire and William have pointed out – when used in an orchestral context, a slight phasing or chorusing of a solo wind instrument probably won't even be noticed anyway. And I guess the other thing one can try to avoid is to rest at certain CC values and instead work/arrange around them if possible. For example, holding a french oboe sustained note at value 46 is really nasty, but fortunately, the effect is rarely as audible as in that particular case. P.S. That's a very sweet and enchanting piece, William!@BachRules said:
Where did you get the number 109? In the Velocity Histogram, I see lines which the VI manual calls "velocity split points" at 24 and 109, so maybe that's where you're getting the number 109? Plus, at 109 and higher it does sound to me like only one layer is playing, whereas I hear multiple layers in the range 99 - 109.Confusing is that I hear multiple layers sounding when VelX is 29, but 29 isn't just below a "split point" -- it's just above one instead. I don't understand what a "split point" is then, since sometimes it's right below the chorusing, and other times it's right above the chorusing.
-
But ultimately, it's probably better not to over-obsess over all of this.
It's not causing significant impairment in my social or occupational functioning, so I think it's not a mental disorder, and it's just me wanting to know which settings give me one oboe sample versus which settings give me two oboe samples. Kind of like a composer or conductor might care whether it's one or two oboes playing a part, and if he didn't care, he might not be as good a composer or conductor. They decided to give us velocity crossfades instead of velocity leaps. It is okay to ask exactly where the crossfades are occuring. This should upset no one, and if it does, then they are the ones over-obsessing.
There are so many people using so many systems where the number of samples playing at a particular velocity setting is no mystery. To dismiss the question as "over-obsession" would be an exhibition of disgraceful fanboyism.
-
I didn't mean that in a dramatic or reproachful way. I'm all for transparency and detailed information, and it would surely be nice to have some clarification from the people at VSL regarding this topic, especially since it isn't fully explained in the manuals. Crossfade phasing is an issue we are facing in this current generation of sampled instruments, and need to be able to work around. As users we should definitely strive for in-depth knowledge and andvanced usage of our tools - even more so in the professional domain - especially in times where any greasy kid is able to get their hands on a "warez"-ed copy of LASS or Orchestral Essentials, stack up tutti FX and start churning out "epic orchestra music" . It's more something I have to remind myself of here and there - not to get hung up on technicalities too much, as I tend to be super-meticulous and pedantic over countless things; sometimes to a point where I forget to actually get some creativity going.
-
... I'm all for transparency and detailed information, and it would surely be nice to have some clarification from the people at VSL regarding this topic, especially since it isn't fully explained in the manuals. Crossfade phasing is an issue we are facing in this current generation of sampled instruments, and need to be able to work around....
^ agreed
-
Long time since I even had Velo XFade enabled for a solo instrument, but I did just now go to verify, using the Bb trumpet. 2-layer or 4-layer, there is a pronounced issue of phasing at every XFade so far, I can't use it. Dynamic samples all the way for me with a solo instrument. If it's a 'concertmaster' supposedly in front of a loud orch, maybe we can hide but I know it's there.
8Dio in that Adagio series has rhetoric as though they've gotten around this but I was confused and I think that workflow isn't for me, so...
I'm glad of the dynamic samples in the deeper libraries here.
-
8Dio in that Adagio series has rhetoric as though they've gotten around this
I'm skeptical, and I've found no demonstration on their site. Plus when you modulate the volume of wet samples, you get modulating reverb tails, which I've never wanted.
-
@BachRules said:
With VelX on, doubling at occurs at settings not corresponding with the Velocity Histrogram, and at other settings I hear no doubling.Well, I think it should be obvious, that, as the wole thing is called crossfade, there is not a leap from one CC-value to another, but a certain range in which two layers are actually x-faded. This usually should happen around the split points, but I guess the VSL team made little adjustments to best suit the individual patches.
-
[quote=MassMover]... I guess the VSL team made little adjustments to best suit the individual patches.@BachRules said:
Don't cross-fades by their very nature play two layers at the same time? Isn't playing two layers at the same time what Jimmy is refering to when he writes, "doubling/chorusing"?So when the Velocity Histrogram has a line labeled with the number "24", and the Brass documentation says the split is between 28 and 29, sometimes the split is really between 29 and 30; and that is why I wrote "arbitrary".[quote=MassMover]... a certain range... usually...That degree of precision would satisfy composers who want results around what they intend, instead of results exactly as they intend. This, still:[quote=BachRules]It is okay to ask
-
Long time since I even had Velo XFade enabled for a solo instrument....
I hadn't realized it jumps between layers, instead of fading, when VelX is disabled. Now I hear that it jumps instead of fading when VelX is off.
Many sampled instruments have a sudden difference between a given velocity on note-on and the next, I can't let that bug me. I would rather deal with CC expression to smooth that out than have that phasing, which just sounds awful to me. Now, I'll expose the instruments and as an M.O. I write linearly/contrapuntally with monophonic instruments often so I don't really go for masking this sort of thing in the first place.
-
Hi,
@BachRules said:
Is there any way to ask where exactly the crossfades are occurring without a bunch of people getting defensive?This information is available in your Library Manuals in your User Area, each patch is documented with the split points in velocity.
Best,
Paul
Paul Kopf Product Manager VSL -
@BachRules said:
Is there any way to ask where exactly the crossfades are occurring without a bunch of people getting defensive?This information is available in your Library Manuals in your User Area, each patch is documented with the split points in velocity.
Best,
Paul
Thanks, I see that, but I am wondering where the crossfade ranges occur when VelX is turned on? In that case, there is a range of values inside which range two velocity layers play simultaneously. Looking in the manual I don't see where this range is.
After some experimenting, it seems the crossover range is roughly 9 values long, leading up to the split-point shown in the manual; and it seems this range doesn't have anything to do with the "velocity split points" shown in the Histogram. If that's correct, my question is answered. I am just hoping to know in general, so I don't have to run experiments on each patch individually to find the ranges.
-
Many sampled instruments have a sudden difference between a given velocity on note-on and the next, I can't let that bug me. I would rather deal with CC expression to smooth that out than have that phasing....
I like that I have a choice between jump-points and crossfade-ranges; but I don't like having to run time-consuming experiments to locate the crossfade-ranges, as their location is undocumented. And the people who act like I slapped a baby if I ask where the crossfade-ranges are, because I am not supposed to admit there is a chorus-effect under some circumstances.
Forum Statistics
191,218 users have contributed to 42,789 threads and 257,330 posts.
In the past 24 hours, we have 2 new thread(s), 8 new post(s) and 42 new user(s).