hugoaboud / fluidtrack

Open Source library for tracking objects on terrain fluid flows
GNU General Public License v3.0
4 stars 0 forks source link

Divergencies between readme tutorial and simulation #1

Open victorhb opened 5 years ago

victorhb commented 5 years ago

Hello, I am not able to reproduce the tutorial found in the readme.

One issue I found is on step 3 that suggests the following command line: python scripts/parse_cache.py sim/brumadinho/cache_fluid sim/brumadinho/fluid.flow 50 [0, 0, 0.203386] [19, 18.5, 3]

To execute that line I had to make a few modifications:

1. I had to explicit the use of Python 3.5 (of course it depends on the default python used in each system, but it would be easier to make it clear that you're using python 3); 2. I had to add a "/" at the end of "sim/brumadinho/cache_fluid" otherwise the script tries to look for files such as "cache_fluidfluidsurface_final_0000.bobj.gz" that don't exist; 3. the list parameters must not have space between the items.

At the end, to get my command line working I had to run: python3.5 scripts/parse_cache.py sim/brumadinho/cache_fluid/ sim/brumadinho/fluid.flow 50 [0,0,0.203386] [19,18.5,3]

Still, I couldn't get it working properly because the mud on my simulation is running under the terrain as shown on the two figures below.

It looks right on the first frame. lama1

However, as the animation starts the mud runs to unexpected places lama3

I noticed that on your tutorial, there are different objects on Blender, such as an "emitter," that does not exist on the brumadinho.blend. I'm wondering if I am doing something wrong or if there are important divergences between the tutorial and the brumadinho simulation files. Could you give me some help on what is possibly happening? I'm using Debian operational system with Blender 2.79.

Thank you for the important work you're doing

hugoaboud commented 5 years ago

@victorhb thanks for the interest! also thanks for the tips on the readme, I'll fix it now.

The images on the tutorial are from mixed older versions, the "Emitter" object shouldn't be there really (I started simulating with an emitter inflow, but then moved to a 12,5e6m3 solid modeled according to the dam topology). I'll record the GIFs again with the current version to avoid confusion.

I'm also updating the project, adding volume to the terrain, to avoid the mud getting through it. That's how I've simulated it in the first place (just select all faces and extrude by 0,3m). I was trying to migrate the simulation to OpenFOAM (so I've deleted the terrain volume to export just the planar grid), but it might need even more resources (CPU/RAM) than Blender to present a quick solution.

I should push an update to this real soon. If you could pull it and run the simulation with a grater resolution it'd be amazing (I'm currently working on the road, so I can't leave my computer processing). The results that are used by fluidtrack itself are interchangeable with OpenFOAM (once I write a "parse_cache" for OpenFOAM).

The updateFlow method from fluidtrack is what deserves more attention right now. It's just gimmick code, basically, for now.