Generate abstract 3d model using only 2d image
Python JavaScript HTML Makefile CSS
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
controller
monitor
qualifier
renderer
.gitignore
Makefile
README.md
docker-compose.yml
kubernetes-common.yml
kubernetes-volumes.local.yml
main.blend

README.md

GenGen

This project aims to build 3d model using only 2d grayscale image as reference. It is impossible to make a real, accurate 3d model, what I'm trying to achieve is an abstract model. The result of an algorithm will be a model consiting of many triangles, looking at it from specified direction and with proper lighting should reveal base image. From any other direction model will look like randomly placed triangles.

Current status

At this moment algorithm is able to generate models for very simple shapes, and is very, very slow.

How?

  • Using genetic algorithm with tournament selection.

Example

  • base image
  • result image
  • 3d model screen
  • time, generations, renders, config

Requirements

sudo apt-get install blender pip install pyyaml