SkeletonDiffusion_Demo/README.md

Generate gifs from sequences of obj files

Environment

Sorry, there is no environment file (the env I am currently using contains packages that are not necessary here). Here the version of a few necessary modules. You will have to install other modules as warning appears once you try running the code.

vedo                      2024.5.1     
vpython                   7.6.5 
python                    3.10.13 

Run the code

Prepare the data

Copy the directory '/usr/wiss/curreli/work/my_modules/Skeleton2Mesh/final_output/hmp/visuals_50samples/amass/SkeletonDiffusion/test' to this folder (I already gave you the rights - afterwards I will delete it). There are meshes for two test segments from AMASS, 9622_GRAB and 11949_HUMAN4D. The predictions are named according to the following convention:

• obs: Past obervation (input sequence)
• gt: GT future
• pred_closest_GT: prediction that is most similar to the GT
• most_far_: the remaining predictions are sorted accordng to their similarity to the GT future, or better to the similarity to the prediction closest to the GT. So most_far_0 is the prediction most diverse from the GT

In the demo we will not have a GT future. We will order the predictions according to a different criterion (maybe diversity).

Generate visuals

Now you can run

python plot_several_meshes.py --mesh_parent_folder=<put_correct_path_here/9622_GRAB>

The output should be in a parallel direction at lowest folder level.

If you are using a ssh connection

You probably have no screen, so the code terminates. You can avoid the problem by generating a screen first with our script setup_headless.bash:

setup_headless.bash python plot_several_meshes.py <...>

Debug via ssh connection

OPen the file of your VS Studio debug configurations, launch.json, and add this new configuration:

    { "name": "Python: Attach to Edge", 
    "request": "attach", "type": "python", 
    "connect": { "host": "localhost", "port": 5678 },
    "justMyCode": true,
    },

Now start the file in debugger mode with debugpy

./setup_headless.bash python -m debugpy --listen 5678 --wait-for-client plot_several_meshes.py

Now start your debug configuraiton ""Python: Attach to Edge" to attach your debugger and start debugging.

For the future

If you need to put more functions together consider using shell cripts

bash create_mesh_gifs.sh