Koji コウジ 高治@kojikanao_nyc05:20 AM - 20 Jul 2021
The steps you need are very simple.
Step1. clone repo
Step2. install packages
Step3. download pre-trained model
Step4. run a script
StarGAN2 for practice
This version of StarGAN2 (coined as 'Post-modern Style Transfer') is intended mostly for fellow artists, who rarely look at scientific metrics, but rather need a working creative tool. At least, this is what I use nearly daily myself.
Here are few pieces, made with it: Terminal Blink, Ghosts, Occurro, etc.
Tested on Pytorch 1.4-1.8. Sequence-to-video conversions require FFMPEG. For more explicit details refer to the original implementation.
- streamlined workflow, focused on practical tasks
- cleaned up and simplified code for better readability
- stricter memory management to fit bigger batches on consumer GPUs
- models mixing (SWA) for better stability
Presumed file structure
|├ _in||input data for processing|
|├ _out||generation output (sequences & videos)|
|├ data||datasets for training|
|│ └ afhq||[example] some dataset|
|│ ├ cats||[example] images for training|
|│ │ └ test||[example] images for validation|
Top comments (0)