A downloadable tool for Windows, macOS, and Linux

This Work is for a Bachelor Thesis in Game Design, and may only be used non-commercially

This work is a trained pix2pixHD Model, with the purpose of coloring and shading creature sketches (1024x1024) to be used as concept art.

The training was done on a Dataset consisting of approximately 200 different "World of Warcraft" creatures for 500 Epoches.

To function properly it requires input which is or resembles a canny map. Blurry lines will hinder the generators ability to make usable output.

Status	In development
Category	Tool
Platforms	Windows, macOS, Linux
Author	Elendayer
Tags	2D, artificial-intelligence, concept-art, pix2pixhd

Download

latest_net_G.pth 695 MB

Downlaod and install Anaconda @ https://www.anaconda.com/products/individual
Download pix2pixHD @ https://github.com/NVIDIA/pix2pixHD
Install pix2pixHD by following the installation instructions give on the site
(On windows replace the train.py with the one from https://programmer.help/blogs/pix2pixhd-train-your-own-dataset-win10.html)
Add a Folder in the ../pix2pixHD-master\checkpoints for the project
Add "latest_net_G.pth" to the checkpoints/"project-folder"
Add a Folder in the ../pix2pixHD-master\datasets for the project
Add a folder called "test_A" to the dataset/"project-folder"
Add 1024x1024 pixel sketches to the "test_A" folder
Open the pix2pixHD Folder in Anaconda ("cd ../pix2pixHD-master")
Replace the "project-folder" with the name of the project-folder and execute the command:

python test.py --name "project-folder" --ngf 64 --label_nc 0 --no_instance --how_many 464