borisb2 Posted September 22, 2014 Share Posted September 22, 2014 Hi. Coming from naiad and synapse (weta) I am finally starting to use houdini for water. Unfortunately I started with H12.5 and are now trying to adapt my setups to H13. What looked more natural in H12.5 now looks a lot more stringy/tendrily in 13. Is that a common issue? Sure, when I just open a H12.5 scene in 13 and resim it looks similar - but starting a new scene and using exactly the saem values for source / flipObject / flipSolver, the fluid looks A LOT different .. hmmm Have a look here: About 5mil particles 23mil voxel count (.007 particle separation) 12x5x5m fluid Houdini 12.5 H13 a lot more tendrils/clumping flipbook: www.ambientfx.de/files/riverbed_125.mov www.ambientfx.de/files/riverbed_13a.mov So I guess there are different settings in the subnets that would need to be matched .. which is close to impossible of course? Any advise what I could improve here? Thanks a lot Quote Link to comment Share on other sites More sharing options...
borisb2 Posted September 22, 2014 Author Share Posted September 22, 2014 (edited) Here's another attempt to get similar look. Left: the cache from Houdini 12.5 Right: same setup (started in a new scene), using EXACTLY the same settings for fluid-source, Flip object, collision and Flip solver The fluid from H13 looks AWEFUL! ... What's going on here? Edited September 22, 2014 by borisb2 Quote Link to comment Share on other sites More sharing options...
eetu Posted September 22, 2014 Share Posted September 22, 2014 If you could attach both scenes it would make it easier to assess what the difference is. You could just methodically go through all the settings and see if any of the defaults have changed, my first guess would be maybe something in the reseeding. Quote Link to comment Share on other sites More sharing options...
CeeGee Posted September 22, 2014 Share Posted September 22, 2014 some options are missing h12.5 compare to h13, they can look same but resault is diffrent Try to turn off automatic extrapolation by default - solver tab on flipsolver, but when you open 12.5 in h13 auto extraplotion is on by default. chk image and video, 12.5 and h13 with extraplotion off look similar flipCompare.mov Quote Link to comment Share on other sites More sharing options...
borisb2 Posted September 22, 2014 Author Share Posted September 22, 2014 Also, when comparing to H12.5 sim, I'm still loosing a lot of volume. That must be the reseeding? Even with similar values? Here's my test with disabling the "automatic extrapolation" It's cutting off the front edge... and I also can see the volume loss (feathering out on right edge) .. similar amount of particles and voxels in both sims Quote Link to comment Share on other sites More sharing options...
Skybar Posted September 22, 2014 Share Posted September 22, 2014 I think I recall they changed reseeding to have a random seed based on $T (time) by default in H13. So if you are using reseeding it will probably look quite different just from that. Quote Link to comment Share on other sites More sharing options...
johner Posted September 22, 2014 Share Posted September 22, 2014 If you could attach both scenes it would make it easier to assess what the difference is. I'll just echo eetu. There were a fair number of changes in FLIP from 12.5 to 13, but obviously simulation quality is always supposed to be improving. If you can post a simple example we can look at what might have changed. Quote Link to comment Share on other sites More sharing options...
borisb2 Posted September 22, 2014 Author Share Posted September 22, 2014 .. looks like I overlooked one mismatch in substepping (only at scene level DOP node) .. I was focusing on all values and nodes INSIDE DOPS and totally overlooked the DOP-node itself.. still not identical sims but a lot closer.. Thanks for your help guys! Quote Link to comment Share on other sites More sharing options...
johner Posted September 22, 2014 Share Posted September 22, 2014 (edited) .. looks like I overlooked one mismatch in substepping (only at scene level DOP node) .. I was focusing on all values and nodes INSIDE DOPS and totally overlooked the DOP-node itself.. Ah, that would make sense. You have pretty fast-moving water, so the additional substeps would definitely help with the collisions and volume preservation. Just note that substepping at the DOPnet level is a bit more expensive than at the FLIP solver level. Some operations like Reseeding occur once per DOPnet solve so it will change the look a bit to go from 2 DOPnet substeps and 2 FLIP substeps to just 1 DOPNet substep and 4 FLIP substeps, for example. But it should solve a bit quicker. Edited September 22, 2014 by johner Quote Link to comment Share on other sites More sharing options...
borisb2 Posted September 23, 2014 Author Share Posted September 23, 2014 (edited) ..moving the foam-question into a new thread.. Edited September 24, 2014 by borisb2 Quote Link to comment Share on other sites More sharing options...
borisb2 Posted September 23, 2014 Author Share Posted September 23, 2014 (edited) . Edited September 24, 2014 by borisb2 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.