Jump to content

Houdini 15 Sneak Peek


Recommended Posts

Guest mantragora

I would have guessed that every tool in the 3d market takes a few years to mature and be built properly. Everything requires a few versions to iron out the bugs and workflow, which is fine as long as the bugs are ironed out- Iirc the h14 selection order issues that affected your tools were fixed within a reasonable amount of builds of being reported.

Overall selection workflow wasn't the slickest in H14. It looked like if no one used it for real work and just assumed that it's very good.

SESI have problem with over-complicating things, and PolyBridge from SneakPeek looks like it have overcompicated UI. I think that reason for this may be because no modeler was using it extensively, just like selections, so SESI devs didn't get proper response what should be tweaked.

  • Like 1
Link to comment
Share on other sites

The sneak peek isn't meant as comprehensive feature list for H15, it's more of a highlight reel. Some features don't translate well to being shown in that way depsite being immensely useful, and others just didn't make the cut (or it'd be 20 minutes long and take months to make). You did see the enhancements to the deform SOP in the Animation section, you just didn't know it ;)

 

So true, so true.

 

 

I've just read the documentation. This is insane amount of enhancements, changes and new features. Madness!

  • Like 1
Link to comment
Share on other sites

  • 3 weeks later...

Some Siggraph notes of OpenVDB in H15

 

http://www.openvdb.org/download/openvdb_houdini_2015.pdf

 

And here's the PowerPoint file with videos (minus the new Siggraph demo videos which are replaced with stills since they're embargoed until after gold release): 

 
 
All the 2015 (and 2013) OpenVDB course slides are pretty interesting (and often Houdini-centric):
Edited by johner
Link to comment
Share on other sites

 

And here's the PowerPoint file with videos (minus the new Siggraph demo videos which are replaced with stills since they're embargoed until after gold release): 

 
 
All the 2015 (and 2013) OpenVDB course slides are pretty interesting (and often Houdini-centric):

 

 

Does this mean H15 replaced most of the standard volume usage inside Houdini with VDBs? It seems like that from the docs but I am not sure.

Link to comment
Share on other sites

Does it really make sense to support VDBs in DOPs when you can already use clustering? As far as I understand it, since fluid simulations most likely produce dense volumes, then using VDB could potentially be less efficient because of its structure being optimized for sparse volumes.

Link to comment
Share on other sites

Does it really make sense to support VDBs in DOPs when you can already use clustering? As far as I understand it, since fluid simulations most likely produce dense volumes, then using VDB could potentially be less efficient because of its structure being optimized for sparse volumes.

I mean sparse field not support in DOP

Link to comment
Share on other sites

So am I. Waiting H15 test result.

 

But what is the point really... anyone could blab about seeing hearing about research on GPU super threaded FEM solver that would be unreal, makes the H14 FEM solver 'useless'  The endless quest of features that 3d users always want is congruent with the prose of wandering through a  kids toy store.

Link to comment
Share on other sites

It would be much more efficient for most cases. I think it's worth noting that DreamWorks has sparse gas and fluid solvers which are mentioned in the 2015 VDB siggraph slides.  The Dneg flip solver also mentioned in the VDB particle slides supports sparse simulations.

 

So why would these companies invest so much dev into making sparse solvers if it doesn't increase performance?

 

Clustering is one possible solution but has it's problems of it's own trying to hide the seams and stitch the containers together.

  • Like 1
Link to comment
Share on other sites

Don't we have the software to implement these sparse solvers if one really wants it? For example you can dig into the PBD solver and see all the code that is used and if it's good enough for SideFx then what is the exact reason why someone can't implement a new solver...  Below is a DreamWorks paper for those whomever wants to 'fix' the built-in one and make a sparse solver... II'l bet $100 that when people try to code a production ready version they end up crying at how hard it is to make these darn things and come running back. Any takers?

 

' Performance Optimization Study for the DreamWorks Animation Fluid Solver '

https://software.intel.com/sites/default/files/m/d/4/1/d/8/Fluid_Solver_Performance_Optimization.pdf

 

 

Edit: It's worth noting that a new voxel based fluid simulation tool was released today for Softimage and will be most likely ported to Fabric... and surprise surprise it's not sparse too. 

 

http://www.blackcore.technology/softimage_efx

Edited by tar
Link to comment
Share on other sites

It would be much more efficient for most cases. I think it's worth noting that DreamWorks has sparse gas and fluid solvers which are mentioned in the 2015 VDB siggraph slides.  The Dneg flip solver also mentioned in the VDB particle slides supports sparse simulations.

 

So why would these companies invest so much dev into making sparse solvers if it doesn't increase performance?

 

Clustering is one possible solution but has it's problems of it's own trying to hide the seams and stitch the containers together.

 

Intuitively I would have think that VDBs wouldn't make DOPs much more efficient but then actual performance measurements often prove intuitions to be wrong! :)

 

I'd be curious to know more but I couldn't find much from the slides that you pointed out?

Link to comment
Share on other sites

Anyways this is all very distracting from what has been done in H15 - in the PPT Johner linked to I'm seeing seamless distributed simulation with great smooth meshing, variable density meshing and some very significant performance metrics!

 

 

BTW is that Nuclear explosion HDA available. Looks good.

 

 

Does this mean H15 replaced most of the standard volume usage inside Houdini with VDBs? It seems like that from the docs but I am not sure.

 

I don't read that in the slides.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...