Thanks! The dual rest blend is the out-of-box solution, really - you'll find it tucked inside the default mantra liquid shaders (and the blend approach there is a bit more sophisticated than mine). But if you want to do the same thing with a different renderer, it's good to be able to do it 'by hand'!
Wow. Just... thank you so much for this! I've been blasting my head with Houdini and though much of it is still beyond me, particularly the VEX stuff, the way you explain everything is actually making sense to me (albeit slowly). This tool is just ridiculous but its like some insane enigma cube that feels almost within grasp and then slips away. I'll keep at it, and absorb all this valuable info as much as possible. Thanks again! FYI, I need to make some nice dripping slime from a creature's mouth so this has come at the right time :)
Houdini's depths are nearly limitless. That can be daunting, or you can just take the zen approach and say "I will never know everything, so it's enough to just know more" and gradually the pieces will click. Happy to know I could help a little :)
You're doing an awesome work here, I hit like before starting each video, thank you :) Also, instead of vdb in a loop, which is relatively expensive, you can in a wrangle use 'xyzdist', which will return the @depth, and then use its returned 'prim num' to find the name from the fractured geo, but since you can't sample a string attrib, in the vornoi fracture enable 'primitive piece', and a litte hack here is to put a 'exploded view' with so tiny scale, between your vornoi fracture and wrangle.
That's definitely an option! I find the vdb reshaping options a bit more reliable for erosion than something like peak sop, so that's the main motivation for using volumes. And although it's definitely expensive, it's only a one-time setup computation, so I tend not to worry much about optimizing there. Using a small exploded separation for the xyzdist test is a great tip, though!
Great tut! Got some great ideas out of it. Just curious if it would be lighter to create a point at the centroid of each piece to calculate distance from surface as opposed to the vbd workflow?
Thanks - glad to hear it sparked some ideas! As for the point sampling, depending on the effect you want you could get away with it - and it would definitely be faster. However, your level-set in that case (the imaginary surface where every point has the same depth value) will always be a sphere. So you'd be losing the shapes of the fractured pieces. You'd also be getting distance-from-middle rather than distance-from-surface, which would add a bit of complication for variable-size chunks. You could try using xyzdist to get your closest surface point, but the lookups involved aren't much different from calculating the VDB in the first place, and you'd also lose the sign (i.e., inside v outside). Ultimately, since the volume processing is just initialization (i.e., it doesn't have to cook every frame), I wouldn't worry too much about optimizing at that stage!
amazing tutorial, I have a question about the intro with the 2 spheres colliding. It looks fantastic and I wanted to recreate it myself but how did you do the shell like fracturing that keeps the slime in it? rbdmaterialfracture?
thanks! that particular clip is using the 'alternative workflow' I cover near the end of the video (1:05:35). So it's using a standard RBD sim for the collision dynamics, then using that info to drive / bias the flip sim.
Thank you, this has been extremely helpful. I am trying to create a simulation of a tart crumpling and a liquid pouring out. I was wondering if there was a way to only fracture the surface and have the interior completely liquid so i can make a solid container for my liquid to pour out of. Any help would be very much appreciated.
Off the cuff, assuming you're ok with the fracturing part, what I might try is setting the exterior points to 'stopped' (and int attrib you can specify on the points) - which will prevent them from moving, but still collide with the other particles. Then you could have a release trigger (e.g., some other collision volume) that would switch the contained points over to their desired viscosity values. That way only the region of the tart you wanted would collapse per the tutorial's method... But I could be misunderstanding what you're after!
@@tomstrickland6290 I'd just do it with an attribute wrangle in SOPs when you're setting up the source particles - a one-liner: i@stopped=1; for the exterior group. You'd then switch them back to an active state (i@stopped=0) inside DOPs probably using a POP wrangle and whatever activation logic you're thinking of!
Beautiful, Sir! Very clear explanation for a super cool effect! May I ask if you know some way to melt objects keeping the proper uv mapping? Would like to melt 3d scans of human heads, but Attribute Transfer would always show uv seams, do you know any workaround? Thanks a lot!
Thanks! Depending on the effect, this can be more or less difficult. One of the biggest issues is when you have a lot of internal points coming to the surface - as these would never have had original uvs. But if you keep things relatively viscous / rigid, then the surface should transfer ok. To deal with seams, you want to look for prims where the verts have uvs that jump from the high end (near 1) to the low end (near 0), then adjust the problem verts (e.g., if your verts go from .9 to 0.05 on a uv dimension, either switch the 0.9 to -.1 or the 0.05 to 1.05). It's worth making an HDA for this if you do a lot of fluid UV'ing.
Note - the above solution assumes wrapped textures, which I'm realizing isn't applicable for your head example. In those cases, if you give your original geometry an attribute defining its uv islands, you can use that attribute to restrict your attrib transfer. It's not always perfect, but it will get things a lot closer...
@@ArgoBeats Hope it works out! I might do a little 'quick-tip' on seam fixing some day. My approach isn't perfect, but it works well enough for certain cases that it could be useful for others!
Big thanks for an awesome tutorial. I rebuild this using the toy instead of a box but for some reason ended up with voro_vdb_sample depth >0 for some points. I used a beak to ensure the VDB volume is bigger than the are covered by the particles. It seems that during the point wrangle I might be getting racing condition if Houdini is trying to be smart and run it in parallel somehow. To address this, I made a detail attrib and my own point for loop. But still end up with positive depth somehow. Any ideas why?
Hey there. I think with a complex shape like the toy, the step of scaling up by 1.5 to cover the edges isn't going to work the same way. That's a good catch. You'll want to expand before the voronoi fracture using a peak sop, or by converting to vdb, dilating with reshape sdf, then converting back to polys before fracturing. The latter approach will probably be more reliable, since peak can generate interpenetrating geo. I tested quickly and it seems to work. Hope that helps!
@@DiffuseFX Yes, I caught that one first and did already use a peak. Will try the other method now. Also considered dilating the VDB pieces in the loop. Will continue to experiment. Thanks. PS: A combination of pre fracture dilating and in loop dilating fixed it in the end. Might have worked with less particle separation.
Would you be so kind as to quickly note down the setup in Redshift? Is it just two materials thrown into a Material Blender with restratio as the Blend Color?
Hi! How are you? I've got a problem, when I put the v in the att wrangle after the for each, it doesn't work. Make a crash in the code. Do someone know what could be? Thanks!
Another one great tutorial! Good explanation and lot of knowledge! Thanks a lot!
Thank you ! Very interesting tutorial.
Super in-depth tutorial and very clear and precise. Very good
Amazing tutorial.
What a legend. Thanks for putting this up for us.
That was amazing. Thanks a lot!
Absolutely legendary tutorial! Jeez. Thanks for this.
Wow!!! Such a great tutorial! Thänx a lot!
Fascinating tutorial Grayden! Your approach to UVing a liquid is great!
Thanks! The dual rest blend is the out-of-box solution, really - you'll find it tucked inside the default mantra liquid shaders (and the blend approach there is a bit more sophisticated than mine). But if you want to do the same thing with a different renderer, it's good to be able to do it 'by hand'!
Thank you for taking your time to do this! You are great at explaining - very interesting!
My pleasure - thanks for the kind words :)
Thank you very much for this tutorial. Please keep uploading!
Thanks! And don't worry - there's more to come!
Fantastic tutorials. I would like to encorage you to continue doing them. Many thanks.
Thanks so much! And not to worry - I'm not planning to stop any time soon :)
Thank you so much man!
Thank You very much!
Wow ! Awesome!
I wish I could give you all the likes I could give. This is such a helpful tutorial!
Thanks so much! Really appreciate it :)
Loving this one. My only comment is the volume is set really low in all your tutorials.
Thankyou so much for this tutorial ! Learned a lot :D
Glad to hear it - thanks for watching!
awesome sauce.
Wow. Just... thank you so much for this! I've been blasting my head with Houdini and though much of it is still beyond me, particularly the VEX stuff, the way you explain everything is actually making sense to me (albeit slowly). This tool is just ridiculous but its like some insane enigma cube that feels almost within grasp and then slips away. I'll keep at it, and absorb all this valuable info as much as possible. Thanks again!
FYI, I need to make some nice dripping slime from a creature's mouth so this has come at the right time :)
Houdini's depths are nearly limitless. That can be daunting, or you can just take the zen approach and say "I will never know everything, so it's enough to just know more" and gradually the pieces will click. Happy to know I could help a little :)
thank you so much
nice tutorial
Really awesome tutorial. I do not comment much, but this tutorial saved my ass and was very thorough.
Thanks! If you used it for a project, I'd love to take a look!
I'm not really using Sims, but there is some super useful tips and tricks. Really good tutorial. Thank you very much for doing them !!
Thanks! And all good sims start with careful SOPs work :)
The rest ratio is on the details
Great video. Thanks.
I've looked there, but I swear I never see it... Wouldn't be the first time I've missed something completely obvious, though!
Love the tutorial! You can get rest ratio using the dop import with the flip preset. Dont import using the same dop node.
aha! I knew there had to be something obvious I was missing - thank you!
You're doing an awesome work here, I hit like before starting each video, thank you :)
Also, instead of vdb in a loop, which is relatively expensive, you can in a wrangle use 'xyzdist', which will return the @depth, and then use its returned 'prim num' to find the name from the fractured geo, but since you can't sample a string attrib, in the vornoi fracture enable 'primitive piece', and a litte hack here is to put a 'exploded view' with so tiny scale, between your vornoi fracture and wrangle.
That's definitely an option! I find the vdb reshaping options a bit more reliable for erosion than something like peak sop, so that's the main motivation for using volumes. And although it's definitely expensive, it's only a one-time setup computation, so I tend not to worry much about optimizing there. Using a small exploded separation for the xyzdist test is a great tip, though!
Amazing tutorial! Thank you very much!!!! Indeed
Let me buy you a coffee! :)
Great tut! Got some great ideas out of it. Just curious if it would be lighter to create a point at the centroid of each piece to calculate distance from surface as opposed to the vbd workflow?
Thanks - glad to hear it sparked some ideas!
As for the point sampling, depending on the effect you want you could get away with it - and it would definitely be faster. However, your level-set in that case (the imaginary surface where every point has the same depth value) will always be a sphere. So you'd be losing the shapes of the fractured pieces. You'd also be getting distance-from-middle rather than distance-from-surface, which would add a bit of complication for variable-size chunks. You could try using xyzdist to get your closest surface point, but the lookups involved aren't much different from calculating the VDB in the first place, and you'd also lose the sign (i.e., inside v outside). Ultimately, since the volume processing is just initialization (i.e., it doesn't have to cook every frame), I wouldn't worry too much about optimizing at that stage!
Featured on CGRecord!
Much appreciated!
amazing tutorial, I have a question about the intro with the 2 spheres colliding. It looks fantastic and I wanted to recreate it myself but how did you do the shell like fracturing that keeps the slime in it? rbdmaterialfracture?
thanks! that particular clip is using the 'alternative workflow' I cover near the end of the video (1:05:35). So it's using a standard RBD sim for the collision dynamics, then using that info to drive / bias the flip sim.
Thank you, this has been extremely helpful. I am trying to create a simulation of a tart crumpling and a liquid pouring out. I was wondering if there was a way to only fracture the surface and have the interior completely liquid so i can make a solid container for my liquid to pour out of. Any help would be very much appreciated.
Off the cuff, assuming you're ok with the fracturing part, what I might try is setting the exterior points to 'stopped' (and int attrib you can specify on the points) - which will prevent them from moving, but still collide with the other particles. Then you could have a release trigger (e.g., some other collision volume) that would switch the contained points over to their desired viscosity values. That way only the region of the tart you wanted would collapse per the tutorial's method... But I could be misunderstanding what you're after!
@@DiffuseFX Thanks for you answer!!!
@@DiffuseFX thank you so much for getting back to me. Would you be able to specify which node I would change the exterior points to stopped on?
@@tomstrickland6290 I'd just do it with an attribute wrangle in SOPs when you're setting up the source particles - a one-liner: i@stopped=1; for the exterior group. You'd then switch them back to an active state (i@stopped=0) inside DOPs probably using a POP wrangle and whatever activation logic you're thinking of!
Beautiful, Sir! Very clear explanation for a super cool effect! May I ask if you know some way to melt objects keeping the proper uv mapping? Would like to melt 3d scans of human heads, but Attribute Transfer would always show uv seams, do you know any workaround? Thanks a lot!
Thanks! Depending on the effect, this can be more or less difficult. One of the biggest issues is when you have a lot of internal points coming to the surface - as these would never have had original uvs. But if you keep things relatively viscous / rigid, then the surface should transfer ok. To deal with seams, you want to look for prims where the verts have uvs that jump from the high end (near 1) to the low end (near 0), then adjust the problem verts (e.g., if your verts go from .9 to 0.05 on a uv dimension, either switch the 0.9 to -.1 or the 0.05 to 1.05). It's worth making an HDA for this if you do a lot of fluid UV'ing.
Note - the above solution assumes wrapped textures, which I'm realizing isn't applicable for your head example. In those cases, if you give your original geometry an attribute defining its uv islands, you can use that attribute to restrict your attrib transfer. It's not always perfect, but it will get things a lot closer...
@@DiffuseFX thank you very much, I'll try to apply your precious suggestions, I'm at an intermediate level, hope I'll make it!
@@ArgoBeats Hope it works out! I might do a little 'quick-tip' on seam fixing some day. My approach isn't perfect, but it works well enough for certain cases that it could be useful for others!
@@DiffuseFX that would be great, thank you again!
Big thanks for an awesome tutorial. I rebuild this using the toy instead of a box but for some reason ended up with voro_vdb_sample depth >0 for some points. I used a beak to ensure the VDB volume is bigger than the are covered by the particles.
It seems that during the point wrangle I might be getting racing condition if Houdini is trying to be smart and run it in parallel somehow. To address this, I made a detail attrib and my own point for loop. But still end up with positive depth somehow. Any ideas why?
Hey there. I think with a complex shape like the toy, the step of scaling up by 1.5 to cover the edges isn't going to work the same way. That's a good catch. You'll want to expand before the voronoi fracture using a peak sop, or by converting to vdb, dilating with reshape sdf, then converting back to polys before fracturing. The latter approach will probably be more reliable, since peak can generate interpenetrating geo. I tested quickly and it seems to work. Hope that helps!
@@DiffuseFX Yes, I caught that one first and did already use a peak. Will try the other method now. Also considered dilating the VDB pieces in the loop. Will continue to experiment. Thanks.
PS: A combination of pre fracture dilating and in loop dilating fixed it in the end. Might have worked with less particle separation.
Would you be so kind as to quickly note down the setup in Redshift? Is it just two materials thrown into a Material Blender with restratio as the Blend Color?
Hi! How are you? I've got a problem, when I put the v in the att wrangle after the for each, it doesn't work. Make a crash in the code. Do someone know what could be? Thanks!
Please, say how to mix colours in this sort of liquids? I can't find any option for it
any chance for vellum with flip interaction?
what do you have in mind? there's a handful of ways I could imagine approaching that, depending on the specific effect...