Hi All! This question is probably for the Jeffs, Wagner or Lait.
I'm noticing drastically different behavior in H18 with regard to the built-in FLIP surface tension. Opening files created in H17.5 usually results in a collapsing/exploding fluid volume.
The other issue is that with previous versions I could tune sim settings with a lower-res sim, and then adjust Particle Separation down for higher-res while still keeping most of the behavior intact, with only minor tweaks needed
I've attached an H18 test scene with multiple FLIP sims that have identical sim settings, other than adjusting Particle Separation for each. In this setup, adjusting psep gives drastically different results even when using the same ST value. Low res sims are affected much less and high-res sims collapse/explode.
Am I missing something with this or is this expected behavior with the changes to FLIP in H18?
Thanks for any insight!
–Dave
https://vimeo.com/davegrafix [vimeo.com]
https://www.behance.net/DaveStewart3D [www.behance.net]
FLIP Surface Tension in Houdini 18 - changed behavior
3415 4 3- Dave Stewart
- Member
- 39 posts
- Joined: May 2014
- Offline
- AslakKS
- Member
- 158 posts
- Joined: Feb. 2016
- Offline
- Dave Stewart
- Member
- 39 posts
- Joined: May 2014
- Offline
Hmm interesting, thanks Aslak.
I remember reading the June 5th note, but I'm seeing the exact opposite in the sample scene posted above, where grid size change (via reduced psep) does change the effect drastically.
The takeaway from all this is that all sims created in earlier Houdini versions will need to be rebalanced, and the ST coefficient will need to be reduced when increasing sim resolution.
The timescale note is consistent with previous behavior, and shouldn't affect this particular case as Timescale is set to default of 1 anyway.
Cheers,
–Dave
I remember reading the June 5th note, but I'm seeing the exact opposite in the sample scene posted above, where grid size change (via reduced psep) does change the effect drastically.
The takeaway from all this is that all sims created in earlier Houdini versions will need to be rebalanced, and the ST coefficient will need to be reduced when increasing sim resolution.
The timescale note is consistent with previous behavior, and shouldn't affect this particular case as Timescale is set to default of 1 anyway.
Cheers,
–Dave
- tamte
- Member
- 8449 posts
- Joined: July 2007
- Offline
it used to be the case that you have to scale the ST by inverse of the particle separation to have consistent look with changing psep
in 17.5.??? (possibly 17.5.279) this was fixed (June 5th note) however how Journal logs seem to work, once H18 was released all the duplicated entries were merged under highest version label, so it looks like that that change has happened only since 18.0.128
all that being said, the behavior should be consistent when changing the psep if using 17.5 or 18.0 build past that date
as mentioned this was not the case before 17.5.279 so obviously scenes before that version will need to be readjusted and that may be the difference you are seeing. As exploding is caused by this change applying internally higher ST needed for ST to be consistent, however you need more substeps, which was also the case when previously compensating it with above mentioned equation
but if you are not getting visually consistent ST for different psep (with enough substeps to converge) then maybe submit a bug
EDIT: just noticed you posted a file, seems to be pretty consistent in 18.0.348, had to up substeps though as mentioned
in 17.5.??? (possibly 17.5.279) this was fixed (June 5th note) however how Journal logs seem to work, once H18 was released all the duplicated entries were merged under highest version label, so it looks like that that change has happened only since 18.0.128
all that being said, the behavior should be consistent when changing the psep if using 17.5 or 18.0 build past that date
as mentioned this was not the case before 17.5.279 so obviously scenes before that version will need to be readjusted and that may be the difference you are seeing. As exploding is caused by this change applying internally higher ST needed for ST to be consistent, however you need more substeps, which was also the case when previously compensating it with above mentioned equation
EDIT: just noticed you posted a file, seems to be pretty consistent in 18.0.348, had to up substeps though as mentioned
Edited by tamte - Feb. 6, 2020 19:59:02
Tomas Slancik
FX Supervisor
Method Studios, NY
FX Supervisor
Method Studios, NY
- Dave Stewart
- Member
- 39 posts
- Joined: May 2014
- Offline
Thanks for the detailed reply Tamte, I generally bump substeps up higher than in the example file but I'm also getting a much stronger ST effect than before, and need to re-balance those values in H18 apparently. That plus substepping will be the trick, just need to re-think the starting values I've been using…
–d
–d
-
- Quick Links