Fluid disappears at frame 650
AlphaZero666 opened this issue · 9 comments
System Information
Blender Version (including hash):3.1.2
FLIP Fluids Version:1.3.0
**Operating System:Windows 11
**CPU:i5
**GFX:what is gfx? Graphics card? GTX 1060 6GB
**RAM: 32GB
Describe the bug
Provide a clear and concise description of what the actual bug/issue is that you are experiencing.
At frame 650 the fluid glitched and kinda went underground and i tried rebaking and still the same so i tried rebaking from frame 600 and now its from frame 600
To Reproduce
Provide descriptive instructions for how to reproduce the issue or how to use your .blend file.
Well since this happened on frame 650 and now 600 i have no idea i just want to fix it
Expected Behaviour
A description of what you expected to happen.
It is supposed to be a river and it should just flow down
Actual Behaviour
A description of what actually happened.
The water suddenly disappears and goes into the mesh, i will attach
Screenshots
If applicable, add screenshots to help explain your problem.
Im writing this on phone and i dont see any button so ill just send links to the image on blender discord server
https://media.discordapp.net/attachments/869803682231906314/980237195107512380/unknown.png
https://media.discordapp.net/attachments/869803682231906314/980237252376539247/unknown.png
You can also contact me on discord [D3ATH] ;)#7798
Hi, thanks for the report! Please attach or link to a .blend file that reproduces the issue, which is required for bug reports (See guidelines here).
For this issue, please also attach a .zip of the simulation log files which can be found in the cache_directory/logs folder. This will help us better understand how the simulation was running on your system.
If you are able to attach a render or a quick viewport animation to show how the liquid is disappearing into the mesh, we may be able to resolve this issue more quickly.
okay, here it is
the blend file: https://www.transfernow.net/dl/20220530vOn3FaDN
logs.zip
this log?
https://user-images.githubusercontent.com/106467397/171026783-be716d6b-8c12-4e56-9a6f-e189a93b5dee.mp4
it is really short because it takes really long to preview render
Thanks for the update! I can see what is causing the issue from the attachments.
There are two Landscape objects set as obstacles. One is set as a regular obstacle, and the other is set as an Inverse type obstacle. Inverse obstacles contain liquid inside of the object rather than causing liquid to flow around it (documentation) and is what is causing the liquid to collapse into the mesh.
In addition to this issue, there looks to be geometry problems with the Landscape obstacles. The obstacles contain many intersecting faces which could explain how the simulation became unstable and flipped from running normally to suddenly containing fluid inside of one obstacle. Intersecting faces are considered non-manifold and can cause problems. See this documentation for more info and tips on detecting and debugging issues related to geometry: https://github.com/rlguy/Blender-FLIP-Fluids/wiki/Manifold-Meshes
For example, geometry issues can be detected with the built-in 3D-print Toolbox addon:
And one more tip: If you have obstacles that are much larger than the domain, you may want to trim them down to improve performance and resource usage: https://github.com/rlguy/Blender-FLIP-Fluids/wiki/Obstacle-Object-Settings#does-my-obstacle-object-need-to-be-completely-contained-by-the-domain
Hope this info helps!
okay, thank you so much, one more question, if i somehow remove the geometry issues, do i have to rebake the whole simulation?
I think you should be able to resume the simulation from a point before the failure (such as from 600 instead of the beginning). When resuming a simulation, the obstacle geometry will be updated and re-exported to the simulator by default (More Info).
yes i managed to resume, but i got another question :D why cant i choose from which frame i want to resume but only from 50 100 150 etc or the latest one?
The savestate interval can be set in the FLIP Fluid Simulation > More Bake Settings section: https://github.com/rlguy/Blender-FLIP-Fluids/wiki/Domain-Simulation-Settings#savestates
The interval value must be set before baking. Increased frequency of savestates will increase cache size.
so i might have another question, do you have any idea how to sudden appearance of the whitewater might have happened?
https://www.transfernow.net/dl/20220530LGms34rz
My guess would be that instabilities due to earlier geometry issues affected whitewater generation in the first part of the simulation. If this is the case, you would need to re-bake the entire simulation so that whitewater generation will be consistent throughout the animation.