real world vs houdini scale help PLEASE!

   3130   4   1
User Avatar
Member
355 posts
Joined: 11月 2015
Offline
hi guys, I think maybe I’m maybe overthinking this scale think but I’m just having a difficult time. They say model stuff to real-world scale and I do my modeling in houdini. So let’s say the real-world scale of my model is 10inches (about 0.25m), There is pretty much no solver in houdini that behaves the way I’d expect, they behave volatile with these kinds of scale even though thats the real-world scale of the object. Most times, adjusting the forces does not yield much better results.

So what would be like a standard factor to scale these objects by for simulation?

Most of the scale stuff i’ve seen is about what to do after bring scenes from maya or other DCC but I just want some kind of consistency in working with smaller objects such as these.

Thanks.
Edited by traileverse - 2021年4月20日 21:52:27
hou.f*ckatdskmaya().forever()
User Avatar
Member
2529 posts
Joined: 6月 2008
Offline
The working rule is all pyro should be simulated as close to the size of the pig head as possible (according to the joke crew). After simulating, use a transform node to size it to fit your scene.

FLIP is really hard to work with at real world units (due to physical memory requirements), so try to scale things down before you begin.

When round tripping from Houdini to other DCCs, you generally have to scale up by 100x or down by 0.01.

Not sure about RBD, but I don't think I have ever seen any real world unit RBD simulations either.
Edited by Enivob - 2021年4月21日 08:55:46
Using Houdini Indie 20.0
Ubuntu 64GB Ryzen 16 core.
nVidia 3050RTX 8BG RAM.
User Avatar
Member
355 posts
Joined: 11月 2015
Offline
Enivob
The working rule is all pyro should be simulated as close to the size of the pig head as possible (according to the joke crew). After simulating, use a transform node to size it to fit your scene.

FLIP is really hard to work with at real world units (due to physical memory requirements), so try to scale things down before you begin.

When round tripping from Houdini to other DCCs, you generally have to scale up by 100x or down by 0.01.

Not sure about RBD, but I don't think I have ever seen any real world unit RBD simulations either.

lol pig head as frame of reference, funny indeed. So, in essence, small scale models really need to be scaled up if modeled in houdini to match real-world size of objects and once the size suits your needs, do your simulation and scale back down? that's the tried and true of dealing with sim and object scale in DOPS? Thanks.
hou.f*ckatdskmaya().forever()
User Avatar
Member
21 posts
Joined: 2月 2020
Offline
Adding to Enivob's comments regarding scale on Pyro/FLIP, I have always found it necessary to scale Bullet RBD sims up 10x to have them behave in a stable manner (especially with regards to jittering/collisions) and then scale the result back down.

I've found it easiest to make a reference copy (rt-click->Actions->Create Reference Copy) of one Transform node to then copy and paste anywhere I need to scale up for simulating. Afterwards, one more reference copy with the "Invert Transformation" key/override removed and toggled on for scaling things back after the simulation.
User Avatar
Member
355 posts
Joined: 11月 2015
Offline
Arvan
Adding to Enivob's comments regarding scale on Pyro/FLIP, I have always found it necessary to scale Bullet RBD sims up 10x to have them behave in a stable manner (especially with regards to jittering/collisions) and then scale the result back down.

I've found it easiest to make a reference copy (rt-click->Actions->Create Reference Copy) of one Transform node to then copy and paste anywhere I need to scale up for simulating. Afterwards, one more reference copy with the "Invert Transformation" key/override removed and toggled on for scaling things back after the simulation.
Thanks Avran, this was helpful, scaling by 10x seems to be a value that works out well for bullet. I.ve been doing some soft bullet constraints testing lately (shown in gif images) and had to go up with scale and had similar issues with the vellum solver.

Attachments:
bullet_soft_RBD_test_01.gif (10.4 MB)

hou.f*ckatdskmaya().forever()
  • Quick Links