retime simulations

   44088   29   3
User Avatar
Member
390 posts
Joined: Jan. 2012
Offline
Actually I think I know why an isolated piece is strobing when it hits the ground. Your simmed geo is spinning chaotically from frame to frame. When it tries to interpolate from each of those positions it strobes.

If you increase the rotational stiffness of the debris to 100, and set collision padding to 0, it helps alot. This isn't a problem with the timeblend timewarp setup, just the sim.
.
User Avatar
Member
230 posts
Joined: Oct. 2009
Offline
zdimaria
Not yet. The strobing is strange to me –it seems like they remain the same shape just scale down through the interpolation. I'll keep looking at it though. Although I would still like to figure this out, is there a reason you can't change your sim to give you the results you want, instead of fixing it after?

is quite complex simulation with hundreds of pieces. It is something like the Paris scene in the Insception but all CG (in the movie was not). The physics should look real but also sureal, like being underwater. I am already using a lot of masks with forces like drag to slow them down and push them around. But at the top of that I want it to run for 2 seconds and then gradually slow down and keep going like this few more seconds and then come back to simulation's original time.

I use a VOP with ramp which remaps the time of the animation.
I don't want just to slow down all the simulation as the example file.

Also it is much easier to fix the final artistic look of the time with this way and avoid resimulate everything again and again. So far it works with the main fractured pieces because they don't change over time and works with fluids (dust and liquids). It doesn't works with debris as you can see.

thanks you for looking on this.
User Avatar
Member
230 posts
Joined: Oct. 2009
Offline
zdimaria
Actually I think I know why an isolated piece is strobing when it hits the ground. Your simmed geo is spinning chaotically from frame to frame. When it tries to interpolate from each of those positions it strobes.

If you increase the rotational stiffness of the debris to 100, and set collision padding to 0, it helps alot. This isn't a problem with the timeblend timewarp setup, just the sim.

where is collision padding parameter? If is not problem of the timeblend timewarp setup, then why by disabling them it goes fine? Its odd…

UPDATE:
I fount padding, but it does not fix anything. Do I have to use something else in the SOP level before the retime setup?
User Avatar
Member
390 posts
Joined: Jan. 2012
Offline
is quite complex simulation with hundreds of pieces. It is something like the Paris scene in the Insception but all CG (in the movie was not). The physics should look real but also sureal, like being underwater. I am already using a lot of masks with forces like drag to slow them down and push them around. But at the top of that I want it to run for 2 seconds and then gradually slow down and keep going like this few more seconds and then come back to simulation's original time.

The real solution imho to this would be to up your frame rate. Changing your FPS to something lik 1000 or 2000 will allow you to keep the simulation the same, and then slow it down to ultra slow motion afterwords. (This is how they would have shot the scene in Inception)

I fount padding, but it does not fix anything. Do I have to use something else in the SOP level before the retime setup?

If you watch your simulation, when the pieces hit the ground, they spin out of control. There is no way to interpolate between these frames that I can think of. You would need to slow them down somehow. I cant check your file right now but, ways to do this would be to make sure they all have mass, increase the rotational stiffness and imo (not everyone would agree on this) but don't use bullet. Again, instead of trying to solve all of this, you could just increase your frame rate.

zak
.
User Avatar
Member
230 posts
Joined: Oct. 2009
Offline
zdimaria
The real solution imho to this would be to up your frame rate. Changing your FPS to something lik 1000 or 2000 will allow you to keep the simulation the same, and then slow it down to ultra slow motion afterwords. (This is how they would have shot the scene in Inception)

well I tried every possible way to have the slow motion but by increasing the pfs change the simulation, it does not just slow it down. It like there is no gravity at all, for example in the this file the torus will never touch the ground, all the pieces fling around, the same with debris, its out of control. The same is with the scale time parameter, it changes the simulation. Only timeblend-timeshoft SOPs or CHOPs can do it right. That's why I want to get a simulation to looks real and then slow it down. Also I will save simulation time.
User Avatar
Member
230 posts
Joined: Oct. 2009
Offline
I kinda fixed it, if you change the solver from bullet to RBD just for the debris, you cant really see the problem. There is much less scaling when they hit the ground, if there is any, but because they rotating, it seems like part of the rotation.

But still will be nice if anyone ever come with a solution or explenation why bullet creates this problem.

Finally, I just realized that changing the fps, only effects simulations with bullet solver, it works fine with RBD solver. I hope in a fix for this from SESI.
User Avatar
Member
230 posts
Joined: Oct. 2009
Offline
hello,

I am back on the same topic. The director wanted to have real slow motion smoke in three shots. By using the time scale can be very tricky as it effects the simulation itself not just the time. By increasing the fps also effect the simulation. So I used a combination of time scale and Jeff's technice as you can see it in his post here. But I am getting a flickering and I cant understand why is that. I cant attach a file but I have attached few rendered frames. What can be this? The container is fixed and everything is as Jeff's setup for the retiming.

Attachments:
retime.mov (156.1 KB)

User Avatar
Member
53 posts
Joined: May 2010
Offline
Do not change timescale,it will change simulation resault.
Do not use timeblend/timeshift, it perform linear blend between two voxels.(ie. a smoke rapidly move ,you will see : fluid isn't moving and at somewhere smoke gradually appear)

Read back you sim cache,
for field data,advect fields use vel field,
the scale of gasadvect can be the “scaletime” factor.
and particle data , move it by this formula P + (v*newtimestep).

but bcz pyro advect maybe maccormark or BFECC (they are second order)
particle moving maybe also not linear,
you maybe still need a little blend between your retime resault and every original frame.
User Avatar
Member
3 posts
Joined: Feb. 2014
Offline
Hey guys

I have the same problem as Christos Parliaros. I used the scene file of jeff but there is a strobing effect. If any one has found working solution to this issue. please post the sample file. I am new to houdini so does not know alot.


thanks

APoorva
User Avatar
Member
129 posts
Joined: Jan. 2013
Offline
https://vimeo.com/112151619 [vimeo.com]
  • Quick Links