Combining Solvers With a Multi-Solver in Houdini

A cool thing that you can do in Houdini is to stack solvers with the help of a multi-solver. In this tutorial I'm showing a simple example of this by using a pop solver together with a bullet solver.

[@edschiffer]

THANKS! That's exactly what I was looking for.

[@quentinrossetto1899]

5 years later and this video still helps me a lot, thanks !

[@quentinrossetto1899]

Still the simplest and most usefull ressource on multisolver as well, congrats !

[@AnsonSavage]

I keep coming back to this video, so good!

[@abstractwaves6166]

very good tutorial, short and straight to the point

[@aminebenafla]

simple and sweet ! i like that sometimes we can get soo lost inside Houdini about how we was build ( complexity ); but to get the job done can be more easier than we thought i dont mean i dont love the level of access to the data with it thats why i I threw my 8 years in 3ds max but i use to watch alot of tutorials ! that one its pretty sharp about specific question beside the fact to get the background ! you do a great job dude thanks !

[@aeko]

AMAZING video! Thanks so much for the super quick explanation. Simple and perfect! New subscriber. Please make more about DOPs like this!

[@TheKitneys]

Great stuff. Thanks for sharing your insight.

[@ParticleSkull]

Cool stuff! Thank you

[@Dominic-du4ow]

Simple and powerful. Thanks for sharing.

[@mohanpugaz]

Oh Man Very very useful tricks!. Thanks a lot!

[@rickedeckard2006]

Many thanks for you time. Excellent tip.

[@Sobieski526]

So simple! Thank you

[@muhammetmertturan]

I LOVE YOU, thanksss. U R Amazinggg

[@PhilippPavlov]

Oh, so thats how it could be done... Thanks, very useful!

[@rahulbisht396]

you always surprise me sir

[@riccardobrunello2248]

really cool!!

[@saadomar1101]

i gotta thank you man !

[@traceurAlex]

super cool!

[@anthonymcgrath]

loving these short tuts for you thanks :) any chance you could do one for destruction? I know there's a few on yt but maybe a 3 vid series.. propagate cracks.. controlling simulated pieces to a final rest position.. stuff like that?

[@ShortSweet3D]

Happy you like them! :) Sure, that's not a bad idea. I have a few others tuts I need to finish up first, but then I'll have go at it.

[@massimobaita7178]

Thank You very much!

[@rileyman360]

quick question, after adding the bullet rdb solver into the multisolver, and making sure they're in the correct order shown here, I found that the objects don't follow the curve force I had drawn, however when I remove it, they follower the curve force exactly as I want them to. Any idea on what might be happening? The tutorial you've made is otherwise perfect and greatly appreciated.

[@ahahlex]

thanks !

[@tomjerry84]

I wanna give this a hundred Like

[@ChrisGray3d]

Very cool. Doing this in 17.5, the packed primitive 0 that comes out of the dopnet doesn't seem to be affected by the pop curve force. It just floats off. I'm deleting it with a blast, but I wouldn't mind understanding why this first cube is different. It's even created when I turn the particle generation off.

[@ShortSweet3D]

Hmmm, odd. I opened up my old project file and it seems like it still work in 17.5 for me

[@HangYuriYangFX]

I actually have a question... How to make pop interactive with simulated RBDs?

[@mrmaxim77726]

Це дивовижно! Дякую!

[@Koba_K24]

This is wonderful but they're still intersecting because the geometry isn't following the rotation of the bullet solver's geometry representation. I haven't found a way to solve this yet and this is happening also in this video if you look closely... they're only partially avoiding each others.

[@dannyanalogue]

I have the exact same problem. Did you ever solve it?

[@angosalvo5734]

I hope it's not too late to thank you for this :) If you're still around. Is there a way to modify the scale of objects along the curve based on their age ?

[@098765432qwertyuiop]

Entagma did a video on this some time ago, basically you need to trick the solver into thinking it's new objects @every frame (also this crashes a lot so beware) kzbin.info/www/bejne/oYvYoIOvgcidbJY And i;m not really sure of the usefulness of the popsource in OP's videos as you could just as well create a packedobject on everyframe and applying the popforce in the 3rd input of the Rigid Body Solver

[@angosalvo5734]

@@098765432qwertyuiop Thanks. Will check it out I've got I work around though for what I was doing

[@sketchydreams5097]

Hi! I tryed on Houdini 18.5, 19.0 and 19.5, on 19.5 the box colision meshes apears far from the geometry, in the other versions works fine, anyone faced that, knows how to fix? or if it`s a bug ?

[@archiltsutskiridze9605]

Hello man. How to you fix this bug?

[@sketchydreams5097]

@@archiltsutskiridze9605 I only could use in the 18.5 and 19.0, I don't know what happened on 19.5

[@archiltsutskiridze9605]

@@sketchydreams5097 Thank you Man! 👍😎

[@ichtreffnixtv8605]

Is there a way to scale the instances inside the dopnet? I'd like to decrease the scale over time based on @nage.

[@ShortSweet3D]

yes, set the packed transform with setpackedtransform() (or setprimintrinsic if you are on an really old version of Houdini)

[@ichtreffnixtv8605]

@@ShortSweet3D Thx for your reply, should I use a pointwrangle ? And where exatcly should I place it?

[@ShortSweet3D]

@@ichtreffnixtv8605 depends, I usually put it in a SOP solver and an attribute wrangle. I think Entagma might have done a tutorial about this a few years back that you prob still can find online

[@ichtreffnixtv8605]

@@ShortSweet3D Alright. Thanks for the answers ! :)

[@ichtreffnixtv8605]

Ok, I've managed to change the scale of the instances, even over time , however I can't control it. I actually have some sort of nage, which would let me map the scale with chramp(), but the problem is that there are always new instances being generated, and I permanently want to get the packed transform of the creation frame. This would be easy if all the packed instances were created at one frame, then I could just branch of with a timeshift and set it to creationframe, then wire it in my dop and read the packed transform, which would always be the initial packed transform. Then I could multiply that with my chramp() and I would get a clean control of my scale in the sim. I'm not sure if this is possible. Long story short; is there a way to store the value of the packed transform on its creationframe in an attribute? So something like this: matrix3 trn = primintrinsic(...).valueAtTime(@creationtime)

[@mathef]

Very cool tutorial! Thank you :) Can anybody help me with randomizing particle scales please?

[@ShortSweet3D]

Just put a wrangle right after the scatter node and type: "f@pscale=fit01(rand(@Frame),0.5,1);" The last two values are the scale range.

[@DommageCollateral]

hello!

[@christianbohm6118]

This is completely unnecessary. The POP Curve Force works with packed RBD directly.

[@BaconBlocks]

not completely, in this setup you're able to continuously emit rbd objects

[@christianbohm6118]

@@BaconBlocks You can continuously emit RBD objects. Not very hard.