![]() |
More aggressive settings for Temporal Smooth - Printable Version +- tyFlow Forum (https://forum.tyflow.com) +-- Forum: tyFlow Discussion (https://forum.tyflow.com/forum-1.html) +--- Forum: Feature Requests (https://forum.tyflow.com/forum-4.html) +--- Thread: More aggressive settings for Temporal Smooth (/thread-4822.html) |
More aggressive settings for Temporal Smooth - Findus - 06-13-2025 Temporal Smooth seems very weak, even with maxed out settings. I tried to get rid of some jittering in cloth (as per the instructions with a reference flow). I could see only minimal improvement, so I set up a test case. It seems like you need a lot of stacked smooth operators (no pun intended) to get any meaningful result. Which is clumsy, but would be ok if it wasn't for the fact that we have to trigger all of those manually each time we open the file unless we cache the result to disc. The particle motion is smoothed before the spawn / spline paths which only serve for visualization (circle/green is smoothed) Unless I'm missing something, here, would it be possible to offer a more aggressive algorithm or at least unlock the spinners (with a warning regarding calculation times / RamĀ if necessary)? Or a setting for iterating the overall operator x amounts of time automatically? Obviously, there are other ways to smooth that spline, but that's not the point of this exercise. Example file is TyFlow 1.128 and Max 2023 RE: More aggressive settings for Temporal Smooth - tyFlow - 06-13-2025 The Temporal Smooth operator works best on fully cached particle data. Operators that inject forces into the sim (ex: Path Follow) will influence the result of the smooth operation...so your best bet is to create a reference sim (Birth Flow - Flow Update) and smooth that. In the next build I will also add an option to remove velocity from particles, so that even force operators won't influence the smooth result. Also your sim has an extremely small timestep (1/32)...so that vastly increases the number of sampled points, thereby reducing the effect of the smoothing algorithm by 32x. I guess in that case, 100 max iterations is not enough - in the next build I'll unbound the max iteration value, so you don't need to stack operators, you can just crank up the iterations value. But really there's not really a reason to have such a low time step in a setup like this. A workaround in the meantime is to increase your time step, and then after you cache your particle data in the Temporal Smooth operator, disable the Path Follow operator. RE: More aggressive settings for Temporal Smooth - Findus - 06-13-2025 Thanks a lot for your reply! The small timestep was a remnant of the final application being a cloth sim, but I didn't really think about how that might affect the smoothing. " in the next build I'll unbound the max iteration value, so you don't need to stack operators, you can just crank up the iterations value" Thank you so much, it's really appreciated! |