Sure, here is a project that should demonstrate each control's flaw. Its a null test with FPC. (yes, I know two channels playing at different velocities won't null -- patince! ) Despite round robin on the preset I used, the kits still null perfectly, and the larger number of velocity layers help exaggerate visual peak differences between channels, making it clear that velocity is being altered. If you are ever concerned that round robin is a problem, save a new version of the project and reload it. That will reliably reset FPC's round robin so you can be absolutely sure the imperfect null is because of the level scaler. Keep in mind also, that leaving this project alone also upsets the null after a few minutes. Reload periodically.reflex wrote:I don't see any difference in the way that parameter is handled compared to the others. They're all processed when a voice is created and then on each new tick.TheDichotomist wrote:The bug: "VFX Level Scaler: settings are not applied immediately" is fixed for most parameters except the velocity tension knob. (it is improved, and may be hard to hear differences)
Do you have a good way to notice this difference?
What seems to happen is that changing the Center parameter will activate the selected tab's operation, which can have an effect on the sound. It doesn't seem to alter velocity by itself, at first sight.And the "Center" control still alters note velocity slightly, even though no tension changes have been applied.
How to use the project:
1. Play it "as is" first, to be sure everything cancels by default.
2. Go through each channel and teak it's suggested control in level scaler, and note the differences.
* Note which velocity layer is being activated in both, the reference channel, and the channel you are working on. (use the pad menu in the upper left to select the "Ride" cymbal without triggering it and upsetting the round robin)
* Notice the peak differences in the two active mixer channels. (reference and the channel you are working on)
3. Use the "key-splitter reference" (instructions below) channel to compare a functional implementation of velocity scaling to level scaler's results and the original reference channel.
* Open the keysplitter, and use the leftmost point on the "velocity" tab to change the incoming velocity. (subtle changes first)
* You can adjust the note velocity in the "Reference" channel to match the value sent from keysplitter, and check that these two null currently (with a LOT of fiddling)
* Notice the difference between this and the level scaler; level scaler actually triggers a full velocity note and backs off of it in real time (or so it appears on a peak meter), while the keysplitter does not. (which is the correct behavior, I think)
4. Try the other controls too. Reload the project and see that "offset" and "multiply" also send incorrect values and correct them soon after. (at least according to the peak meters, as described/shown earlier with key-splitter) FPC's velocity readiout confirms this also.
5. Don't forget to reload the project every couple minutes!!!
This brings me back to something I asked about in my last post. The level scaler alters note properties on notes as they are playing, just like FL's slide notes. I was wondering about an option to turn that behavior off, so it just rescales the note properties like keysplitter does (not altering existing voices on every new tick?)reflex wrote:I don't see any difference in the way that parameter is handled compared to the others. They're all processed when a voice is created, and then on each new tick.
In my test project, solo the keysplitter channel and drag the leftmost point on the velocity graph up and down while the pattern plays. Note volumes are not altered in real time like they are in level scaler -- which is nice. It would be handy if level scaler had a switch to enable this sort of behavior (exclude/ignore existing notes), so it would only affect newly received notes, and only change their values once -- permanently for that voice. That would avoid unexpected volume fluttering, pumping/breathing, pitch slides, panning changes, or filter automation on existing notes. I'm not sure if this can be implemented or not (your post above gives a clue about how it might, I think) but it would certainly make the tool a more capable performance/sequence modifier.