Popular Post DonRomano Posted November 8, 2019 Popular Post Share Posted November 8, 2019 Hey everyone, so after my very long thread (https://forums.odforce.net/topic/44053-implement-kernel-maths-in-vex/) I finally found a solution. The setup works great at 0.05 res, you might have to dive inside the solver and change some parameters to get it work at other resolutions (especially to get the droplets). Here are some test : Emission threshold at 60 (depending on the situation, there might be some abusive expansion, like at the end on the rubber toy, but this seems impossible to fix when you want a full thin sheet) Stretch threshold at 0.75 / Surface Tension 0.5 Emission threshold at 15 / Stretch threshold at 1.25 / Surface Tension 1 Houdini default Flip with Surface Tension at 1 Crown test with 50 Emission threshold (problem with infinite expansion with liquid dropping from the collider) If someone has suggestion to improve it, feel free to tell me I'm still working on fixing the infinite expansion issue, but the emission threshold is doing great until 20-30 depending on the situation Cheers, thin_sheet_solver.hipnc 13 Quote Link to comment Share on other sites More sharing options...
dafa1011 Posted November 26, 2019 Share Posted November 26, 2019 The effcets are awesome! I tested the file you uploaded and followed your parameters, but I couldn't get the effect(rubbertoy)you showed. Do you have any Solver configuration tips? Quote Link to comment Share on other sites More sharing options...
DonRomano Posted November 26, 2019 Author Share Posted November 26, 2019 7 hours ago, dafa1011 said: The effcets are awesome! I tested the file you uploaded and followed your parameters, but I couldn't get the effect(rubbertoy)you showed. Do you have any Solver configuration tips? Try to add a gas surface tension and play with the amount ! Also, add the id per particle option (to be able to retime). And play with gravity/drag to get what you want ! Cheers, Quote Link to comment Share on other sites More sharing options...
Gianni333 Posted May 12, 2020 Share Posted May 12, 2020 Very nice setup! How would you adjust that if you had smaller scale and a particle resolution of say 0.003 or something like that? Which parameters would you scale up or down? Quote Link to comment Share on other sites More sharing options...
GlennimusPrime Posted May 26, 2020 Share Posted May 26, 2020 Hi @DonRomano Thanks for this! It's a very clever setup! I wanted to ask if the point wrangle on the source points is needed for the thin sheet solver to work inside of DOPs? Is that point wrangle setting some attributes that get referenced inside the DOP noetwork? int pts[] = pcfind(0, "P", @P, 1, 2); i[]@origpair = pts; Quote Link to comment Share on other sites More sharing options...
DonRomano Posted May 27, 2020 Author Share Posted May 27, 2020 On 12/05/2020 at 4:21 PM, Gianni333 said: Very nice setup! How would you adjust that if you had smaller scale and a particle resolution of say 0.003 or something like that? Which parameters would you scale up or down? Well, I didnt get my head into the tool for a long time, but i'll say the snap distance, a smaller distance threshold and a smaller emission threshold On 26/05/2020 at 8:28 AM, GlennimusPrime said: Hi @DonRomano Thanks for this! It's a very clever setup! I wanted to ask if the point wrangle on the source points is needed for the thin sheet solver to work inside of DOPs? Is that point wrangle setting some attributes that get referenced inside the DOP noetwork? int pts[] = pcfind(0, "P", @P, 1, 2); i[]@origpair = pts; Hey, it was for testing purpose but I ended up not using it because it was not optimal, and I forgot to delete it ! Cheers, 1 1 Quote Link to comment Share on other sites More sharing options...
scaduxx Posted September 25, 2021 Share Posted September 25, 2021 Do you still work on this solver? I have few improvements to mention! Quote Link to comment Share on other sites More sharing options...
DonRomano Posted September 26, 2021 Author Share Posted September 26, 2021 13 hours ago, scaduxx said: Do you still work on this solver? I have few improvements to mention! Well, it's been 2 years since I did it, and I've learned so many thing since then, that I'm likely to rewrite everything and improve it, but go ahead and tell me Quote Link to comment Share on other sites More sharing options...
scaduxx Posted September 26, 2021 Share Posted September 26, 2021 I did some test with higher substeps ( like 6-8) and the uncontrollable expansion seems to be resolved. Also you lower time scale. Also, a cool trick to split the tendrils and not have long streaks is to get the volume surface from your flip, apply a bit curvate/gradient smoothing, group particles based on that and use the sheeter only on these areas. This will prevent the filling on areas like droplets and let them detach from the main fluid.! I know this adds more time time to the calculations but I think it's worth it ! Last but not least, using a grid-scale of 1.5(instead of 2) is a key thing to get way more details on simulations with surface tension on! Quote Link to comment Share on other sites More sharing options...
scaduxx Posted September 26, 2021 Share Posted September 26, 2021 I noticed that in some cases your refine_droplet node tends to delete/snap some droplets, which is unnatural. Quote Link to comment Share on other sites More sharing options...
scaduxx Posted September 26, 2021 Share Posted September 26, 2021 v01_mesh.mp4 Quote Link to comment Share on other sites More sharing options...
scaduxx Posted September 26, 2021 Share Posted September 26, 2021 Quote Link to comment Share on other sites More sharing options...
virghost Posted October 9, 2021 Share Posted October 9, 2021 awesome！ THX！ Quote Link to comment Share on other sites More sharing options...
vinyvince Posted October 9, 2021 Share Posted October 9, 2021 On 26/09/2021 at 4:40 PM, scaduxx said: I did some test with higher substeps ( like 6-8) and the uncontrollable expansion seems to be resolved. Also you lower time scale. Also, a cool trick to split the tendrils and not have long streaks is to get the volume surface from your flip, apply a bit curvate/gradient smoothing, group particles based on that and use the sheeter only on these areas. This will prevent the filling on areas like droplets and let them detach from the main fluid.! I know this adds more time time to the calculations but I think it's worth it ! Last but not least, using a grid-scale of 1.5(instead of 2) is a key thing to get way more details on simulations with surface tension on! Might I ask what exactly you mean here, please? "apply a bit curvate/gradient smoothing, group particles based on that and use the sheeter only on these areas." Quote Link to comment Share on other sites More sharing options...
scaduxx Posted October 13, 2021 Share Posted October 13, 2021 On 10/9/2021 at 2:46 PM, vinyvince said: Might I ask what exactly you mean here, please? "apply a bit curvate/gradient smoothing, group particles based on that and use the sheeter only on these areas." inside the Sheeter sop, where you get the particles from the flip solver and apply the filling process, you can do VDBfromParticles, then use VdbSmoothSDF to smooth a bit the surface, and then use a Group SOP where you group only the particles that are inside the smoothed surface, AND THEN apply the filling process only on that group. This causes the Hole Filling to be applied only on areas that are the "body" of the simulation and not the egdes. 3 Quote Link to comment Share on other sites More sharing options...
scaduxx Posted October 13, 2021 Share Posted October 13, 2021 Quote Link to comment Share on other sites More sharing options...
DonRomano Posted October 14, 2021 Author Share Posted October 14, 2021 On 13/10/2021 at 3:45 PM, scaduxx said: inside the Sheeter sop, where you get the particles from the flip solver and apply the filling process, you can do VDBfromParticles, then use VdbSmoothSDF to smooth a bit the surface, and then use a Group SOP where you group only the particles that are inside the smoothed surface, AND THEN apply the filling process only on that group. This causes the Hole Filling to be applied only on areas that are the "body" of the simulation and not the egdes. Very clever ! Thanks for your work, I'm probably going to rewrite it from scratch this year and make something better this time ! 1 Quote Link to comment Share on other sites More sharing options...
Atom Posted October 14, 2021 Share Posted October 14, 2021 (edited) I was playing around with trying to implement Billy's suggestion. Do you mean that all those wrangles inside the solver should only be run on the inverse of the group of selected points? ap_dr_thin_sheet_solver.hipnc Edited October 14, 2021 by Atom 1 Quote Link to comment Share on other sites More sharing options...
scaduxx Posted October 18, 2021 Share Posted October 18, 2021 Not all of them, point rearrangement should run on all of the points. To be honest I'm not 100% sure which sops should be applied to all particles. The fact that still some droplets get deleted is an issue, and also the fact that this solver works only on specific grid scale and particles scale is an issue. I'll try to implement something more procedural, but keep in mind that hole filling works really depending on substeps and surface tesnion Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.