Deep Lagrangian Fluids

Back to Index

Finnaly, the prize.

Let's use GIT once again:

• git clone https://github.com/intel-isl/DeepLagrangianFluids.git

Install other dependencies, as the manual says:

• pip3 install --upgrade git+https://github.com/tensorpack/dataflow.git
• pip3 install zstandard plyfile msgpack msgpack-numpy

Go to dataset/splishsplash_config.py, and set the path of the simulator:

• SIMULATOR_BIN = "/home/liliane/miniconda/envs/DeepFluid/GIT/SPlisHSPlasH/bin/DynamicBoundarySimulator"

Show time! Go to DeepLagrangianFluids/scripts:

• python3 run_network.py --num_steps 1 --weights pretrained_model_weights.pt --scene example_scene.json --output example_out --write-ply train_network_torch.py

Note: the number of steps was set to 1 just to quickly see if it works.

Recreating part of the data for the cover video:

• cd DeepLagrangianFluids/scenes
• python3 ../scripts/run_network.py --weights ../scripts/pretrained_model_weights.pt --scene canyon_scene.json  --output canyon_out --num_steps 1500 ../scripts/train_network_torch.py

<- Back

Next ->