Last active
February 14, 2022 05:48
-
-
Save marimeireles/b976fdde03982deb54bffefc8389f5f2 to your computer and use it in GitHub Desktop.
stylegan tattoo gen
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
## Examples of colabs using stylegan | |
1. stylegan3 https://colab.research.google.com/drive/1Nal3M-wjv6BeIgyTgvxhaccPFGEP6cbk?usp=sharing#scrollTo=WbCGexOebBGi | |
More resource on the colab: https://levelup.gitconnected.com/training-a-stylegan3-in-colab-gan-create-nft-6dd119774644 | |
2. stylegan3 https://github.com/PDillis/stylegan3-fun | |
3. stylegan2 w/ code to train your own model https://colab.research.google.com/github/dvschultz/ml-art-colabs/blob/master/Stylegan2_ada_Custom_Training.ipynb#scrollTo=jOftFoyiDU3s | |
4. original repo for stylegan3 https://github.com/NVlabs/stylegan3 | |
5. according to the author is a better default to train on top of https://github.com/skyflynil/stylegan2 | |
## Scrapper | |
1. Instagram scrapper https://github.com/arc298/instagram-scraper | |
Doesn't provide classification by color, etc. | |
2. Pinterest https://github.com/SwatiModi/pinterest-web-scraper | |
Pretty basic, only classifies by tag, but pinterest has way cleaner data than instagram | |
To create a cleaner dataset we can normalize the imgs somehow: | |
1. All black and white | |
2. Same size | |
3. Run some classifier on top (this is a lot of work, I'm guessing, | |
but maybe there's some easy thing, but classifying it by hand will probably be worse.) | |
## Resources | |
~~1. pre-trained sets https://pythonrepo.com/repo/edstoica-lucid_stylegan3_datasets_models~~ | |
2. convert low resolution outputs from stylegan to higher https://github.com/aiXander/StyleGAN-Resolution-Convertor | |
3. for whatever reason, the inspiration dude is using https://app.runwayml.com instead of colab. but i think they're very | |
different platforms | |
4. anime: https://mega.nz/file/PeIi2ayb#xoRtjTXyXuvgDxSsSMn-cOh-Zux9493zqdxwVMaAzp4 | |
5. abstract art: https://drive.google.com/uc?id=1YzZemZAp7BVW701_BZ7uabJWJJaS2g7v |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment