[MastaMinds]

Greetings everyone,

I am not new to Paraview, but I also do not consider myself an expert. I have been trying to create a movie using ParaView. I made simulations in Fluent and exported the results to ParaView. The grid size is 8 million cells and it is an unsteady flow.
I was able to successfully export the results I wanted to ParaView and visualize them. However, I suffered from some issues. ParaView is very slow when visualizing the results or even importing them. I tried running it on a computer with excellent specifications, but the problem is still there.
The PC I am using has a core i7 processor with 8 logical cores, 64 GB RAM, and a 4 GB dedicated NVIDIA GeForce graphics card.
After visualizing the results, I am able to rotate the view and save an image without issues, but that's not the only thing I need.

I am trying to create a movie of iso-surfaces of Q-criterion at different time instants, and it appeared to be impossible and I don't know where the issue lies. I remember a professor showing us in the class a movie of 3D Lambda2 iso-surfaces that show the transition from laminar to turbulent and it ran smoothly without issues (it was OpenFOAM and the grid size was probably slightly smaller). Nevertheless, the movie he showed us was on his personal laptop, which has specifications by far less than the machine I am using.

Can anyone help?
Thanks.

[flotus1]

There are two factors that determine how fast Paraview can create an animation from multiple time steps:
1) File I/O. Paraview needs to load the values for each time step from disk, which can take some time depending on file size and file system throughput. There is probably an option to load all time steps at once into memory, which takes longer initially, but makes switching between time steps much faster. Not 100% sure about that, I never used it.
2) The amount of computing that has to be done. If your file only contains the velocity vector field, you need to apply a filter in PV that computes e.g. the the Q-criterion. This step takes some time to compute, but can be sped up using PV in parallel. By default, PV performs these computations on a single core. If on the other hand your data already contains the field for the value you need in your animation, processing times are much shorter.

I don't know what your professor did to create an animation in real-time on his laptop in Paraview. Maybe a combination of all of the above, or maybe he just showed a pre-recorded animation?
I am used to animations taking between a few seconds up to a few minutes per time step in PV, depending on mesh size and the amount of processing necessary in PV.

[MastaMinds]

Originally Posted by flotus1

There are two factors that determine how fast Paraview can create an animation from multiple time steps:
1) File I/O. Paraview needs to load the values for each time step from disk, which can take some time depending on file size and file system throughput. There is probably an option to load all time steps at once into memory, which takes longer initially, but makes switching between time steps much faster. Not 100% sure about that, I never used it.
2) The amount of computing that has to be done. If your file only contains the velocity vector field, you need to apply a filter in PV that computes e.g. the the Q-criterion. This step takes some time to compute, but can be sped up using PV in parallel. By default, PV performs these computations on a single core. If on the other hand your data already contains the field for the value you need in your animation, processing times are much shorter.

I don't know what your professor did to create an animation in real-time on his laptop in Paraview. Maybe a combination of all of the above, or maybe he just showed a pre-recorded animation?
I am used to animations taking between a few seconds up to a few minutes per time step in PV, depending on mesh size and the amount of processing necessary in PV.

Thank you for your reply. The file I exported to ParaView contains Q-criterion values and they do not need to be computed from the velocity field. With regard to the animation my professor played in class, no it wasn't pre-recorded, It was real-time and he ran the animation in ParaView itself. One thing I am sure of is that the results he showed us came from an in-house DNS code.
I think I will skip viewing the animation in ParaView and just save the animation as a movie.
Thanks again.

[flotus1]

Found an option to "cache geometry for animation". That should speed things up after an initial load cycle.
Another possibility might be that your professor did not animate an iso-surface extracted from a full field on the fly. Instead animating the iso-surface itself, which had previously been exported.