theheliosequence
I honestly doubt that the calculation would take much more effort than it would to simply change from linear to fast or slow. I'm not sure why you would make that speculation or assume that it would be a deep processing burden. We can draw LFO automation in seconds, without the CPU flinching... we can calculate transients nodes, instantly... And you think we couldn't change the shape of automation without a CPU hit? A Do you have any evidence or knowledge to support this? I'm honestly curious why you've suggested this, as I have no background in DSP processing/calculation.
Given the amazing power of modern processors, I could be wrong about how much of a burden this would impose, but as a math guy I can tell you that calculating a polynomial curve to smoothly fit three points with slopes at both ends that match the preceding and following slopes/curves (which might also be polynomial) and recalculating it continuously as one of the points is moved in two dimensions is a much more significant task than rendering a pre-defined logarithmic/exponential curve between two points. If you were to string a bunch of these custom polynomial curves together and wanted each to bend smoothly into the next, you'd have a real mathematical challenge on your hands. No doubt there are function libraries out there to help a programmer implement something like this, but at a certain point, I think you would have to enforce some limitations on how many of these curves you link in succession.
As I said before, I'm open to hearing examples where the generic curves are audibly insufficient. At this point it's all hearsay for me, but I understand that my ignorance is not your problem, and it might be too much of a burden to school me.
If there really is a requirement that just can't be met without using a lot of nodes to get just the right curve, I would probably propose adding a limited number of additional curve types using other easily calculated functions to meet that need rather than trying to implement the more unlimited kind of customization you're proposing.
If there's a DAW out there that does his already, then I would say SONAR might need to step up and compete, but if not, I think SONAR should only lead if there's a demonstrable and widespread need for more flexibility (pun intended).