Object contribute to fluid sim animation start earlier than keyframe #29186

New Issue

ganesha adiguna · 2011-11-07T11:27:09+01:00

ganesha adiguna commented

2011-11-07 11:27:09 +01:00

%%%I tried to animate by inserting keyframe to the inflow option: "Object contribute to fluid sim" checkbox at frame 250. After baking the domain, the inflow start early at frame 243

My system spec:
Official Blender 2.60 r41226
Proc: Core i3 M370 @ 2.4ghz
Memory: 4gb
Vga: Ati Mobility Radeon 5470%%%

%%%I tried to animate by inserting keyframe to the inflow option: "Object contribute to fluid sim" checkbox at frame 250. After baking the domain, the inflow start early at frame 243 My system spec: Official Blender 2.60 r41226 Proc: Core i3 M370 @ 2.4ghz Memory: 4gb Vga: Ati Mobility Radeon 5470%%%

ganesha adiguna commented

2011-11-07 11:27:09 +01:00

Changed status to: 'Open'

ronan ducluzeau commented

2011-11-07 16:51:18 +01:00

%%%I confirm the problem with your .blend.

But I can't reproduce it with a .blend from scratch.%%%

%%%I confirm the problem with your .blend. But I can't reproduce it with a .blend from scratch.%%%

Lars Kohnert commented

2011-12-03 15:28:52 +01:00

%%%I do have the same Problem with blender 2.6. r41226
The amount of frames, wich the simulation starts to early, differs. It often changes after recalculation. From what i have seen so far, between 3 to 13 frames.
My system:
ATI Radeon HD5570
Intel Core i5-2300 2,8GHz
4gb Ram%%%

%%%I do have the same Problem with blender 2.6. r41226 The amount of frames, wich the simulation starts to early, differs. It often changes after recalculation. From what i have seen so far, between 3 to 13 frames. My system: ATI Radeon HD5570 Intel Core i5-2300 2,8GHz 4gb Ram%%%

Thomas Dinges commented

2011-12-30 01:55:48 +01:00

%%%Assigning to Daniel Genrich, our Fluid module owner. :) %%%

Thomas Dinges commented

2012-06-19 00:48:15 +02:00

%%%Any updates here? %%%

Daniel Genrich commented

2012-11-22 20:17:23 +01:00

%%%This is really weird.
I backtraced that the fluid start at "fluid time" 2.0 (50% of 4 seconds) which should equal frame 250 (50% of 500 frames) but it does end up activating on frame 246/247.

Interesting files:
a) solver_init.cpp line 1450
b) physics_fluid.c line 468

Still have no clue what's going on.%%%

%%%This is really weird. I backtraced that the fluid start at "fluid time" 2.0 (50% of 4 seconds) which should equal frame 250 (50% of 500 frames) but it does end up activating on frame 246/247. Interesting files: a) solver_init.cpp line 1450 b) physics_fluid.c line 468 Still have no clue what's going on.%%%

Daniel Genrich commented

2012-11-22 23:58:44 +01:00

%%%UPDATE:
Seems like frametimes are different from elbeem and blender side.

Elbeem is advancing more "simtime" per frame than blender is with the animated channels.
This means basicly:
We export a "start inflow on frame 250" which is translated by blender into a elbeem time of 2.0 (with 500 frames total and 4 seconds sim).
But within elbeem, the simulation is advanced by a "timestep" per frame (see mSimulationTime, ntl_world.cpp line 409), which is higher than the increasement of 0.008 (the number of this actual case).

Even after one frame, the difference is already visible:
Elbeem Blender
targetTime: 0.008000, frame: 0 targetTime: 0.008000, frame: 0
targetTime: 0.016159, frame: 1 targetTime: 0.016000, frame: 1
targetTime: 0.024319, frame: 2 targetTime: 0.024000, frame: 2

mSimulationTime%%%

%%%UPDATE: Seems like frametimes are different from elbeem and blender side. Elbeem is advancing more "simtime" per frame than blender is with the animated channels. This means basicly: We export a "start inflow on frame 250" which is translated by blender into a elbeem time of 2.0 (with 500 frames total and 4 seconds sim). But within elbeem, the simulation is advanced by a "timestep" per frame (see mSimulationTime, ntl_world.cpp line 409), which is higher than the increasement of 0.008 (the number of this actual case). Even after one frame, the difference is already visible: Elbeem Blender targetTime: 0.008000, frame: 0 targetTime: 0.008000, frame: 0 targetTime: 0.016159, frame: 1 targetTime: 0.016000, frame: 1 targetTime: 0.024319, frame: 2 targetTime: 0.024000, frame: 2 mSimulationTime%%%

Daniel Genrich commented

2012-11-23 02:11:54 +01:00

%%%Found a fix, still testing for sideeffects (and still 1 frame difference in loading the actuall fluidsim)%%%