Skip to content

Point Cloud Generation Networks Applied To Eeveelution Meshes.

Notifications You must be signed in to change notification settings

VKudlay/EeveePCG

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Starter Project: Generating Eeveelution Point Clouds With Point Set Generation Network

Repo By Vadim Kudlay

eeveelution pics

A full tour of the results and thought process behind the implementation can be found in Eeveelutions.ipynb and its associated html export.

This is a brief implementation of a point set generation network to attempt to generate different eeveelutions (i.e. the Pokémon evolutions). Specifically, the network is trained to map from an origin point cloud to a destination point cloud based on a 'transformation vector' which specifies which class the cloud should progress towards. The implementation largely utilizes an basic autoencoder architecture quite similar to the one mentioned in the main reference paper.


Requirements

If you would like to replicate the experiments without modifying the code much, the experiment and associated visualizations were implemented with the following libraries:

  • Tensorflow 2.5
  • Python (~3.9.6)
  • PyMesh (~0.3)
  • Matplotlib (~3.4.2)
  • Scipy (~1.7.0)
  • Numpy (~1.19.5)
  • Jupyterlab (~3.0.16)

Major References


Note

A subset of the utility functions (specifically the loss function implementations) may have been depricated if unused in the experiment; as such, they may require minor tuning and checking over if used in further analysis. These functions were used at some stage of the exploration and may have use in certain contexts but were not significant enough to include in the final analysis.

About

Point Cloud Generation Networks Applied To Eeveelution Meshes.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published