Python script to preprocess images of all Pokémon to finetune ruDALL-E

Overview

ai-generated-pokemon-rudalle

Python script to preprocess images of all Pokémon (the "official artwork" of each Pokémon via PokéAPI) into a format such that it can be used to finetune ruDALL-E using the finetuning example Colab Notebook linked in that repo. This workflow was used to create a model that resulted in AI-Generated Pokemon that went viral (10k+ retweets on Twitter + 30k+ upvotes on Reddit)

My modified Colab Notebook that I used to finetune the model on Pokémon is here: this Notebook's release is purely for demonstration/authentication purposes and no support will be given on how to use it because it is incredibly messy and embarrassing, but there may be a few ideas there that are useful for future generation. Some notes on how the process works are included below, with oppertunity to reproduce/improve it.

The script outputs two things: an images folder with all the preprocessed images plus a data_desc.csv file which contains the image path and Russian caption pairs for finetuning. Some examples of the preprocessed input images are present in the images folder, plus the final data_desc.csv.

The model used is not included in this repo because it's currently too large (~3GB) to distribute (will add the model to Hugging Face at some point).

Preprocessing Script Notes

  • The GraphQL interface to PokéAPI is used as it allows to retrieve the type information plus IDs of all Pokémon in a single request. As a bonus, the returned IDs include the alternate forms of Pokémon (e.g. Mega) which would not otherwise be present just by incrementing IDs.
  • ruDALL-E requires 256x256px, RGB input images. In this case the source input images from PokéAPI are conveiently both square and larger than 256x256 so they downsample nicely. Since the images have transparency (RGBA), they are composited onto a white background.
  • The translation service used is Yandex, which apparently has decent rate limits, plus as a Russian company the translations from English to Russian should theoetically be better.
  • The captions (which are later translated into Russian) are determined by type. For example, a Grass/Poison type will have the caption A Grass-type and Poison-type Pokémon, which is then translated into Russian. In theory, this improves the finetuning process by allowing ruDALL-E to notice trends, plus in theory this can be leveraged at generation-time to control the generation (e.g. prompt with A Grass-type Pokémon and have ruDALL-E generate only Grass-type Pokémon)
  • Due to potential rate limits on translation, translations are cached at runtime by Pokémon type(s) so the API is pinged only once.

Finetuning and Generation Notes

  • The model used above was trained for 12 epochs (4.5 hours on a P100), at a max learning rate of 1e-5. The pct_start param of the OneCycleLR scheudler was set to 0.1 so that learning rate decay happens faster. Despite that, the model converged quickly.

  • The parameters for finetuning ruDALL-E are very difficult to get the expected results. Too little training and the output images will be too incoherent; too much training and the model will overfit and output the source images, and also ignore any text prompts. In the social media posts above, the model is slightly overfit and attempts at using text prompts to control generation failed. But overfitting is not necessairly a bad thing as long as it avoids verbatim output.

Usage

You can install the dependences via:

pip3 install Pillow requests translatepy tqdm

Then run build_image_dataset.py

Getting the images into the ruDALL-E finetuning Colab Notebook is up to the user, but the recommended way to do so is to ZIP the generated images folder (~42 MB!), upload it to Colab (or upload to Google Drive and copy it into the Notebook from there), and unzip the folder in Colab itself via !unzip.

Maintainer/Creator

Max Woolf (@minimaxir)

Max's open-source projects are supported by his Patreon and GitHub Sponsors. If you found this project helpful, any monetary contributions to the Patreon are appreciated and will be put to good creative use.

License

MIT

Owner
Max Woolf
Data Scientist @buzzfeed. Plotter of pretty charts.
Max Woolf
A Python library to simulate a Zoom H6 recorder remote control

H6 A Python library to emulate a Zoom H6 recorder remote control Introduction This library allows you to control your Zoom H6 recorder from your compu

Matias Godoy 68 Nov 02, 2022
SECRET SANTA / KRIS KINGLE

SECRET SANTA / KRIS KINGLE Note: Before executing the script, make sure to turn

DEV_FINWIZ 10 Dec 06, 2022
ALSPAC data analysis studying links between screen-usage and mental health issues in children. Provided data has been synthesised.

ADSMH - Mental Health and Screen Time Group coursework for Applied Data Science at the University of Bristol. Overview The data set that you have was

Kai 1 Jan 13, 2022
This is a simple python script for checking A/L Examination results of srilankan students

AL-Result-Checker This is a simple python script for checking A/L Examination results of srilankan students INSTALLATION [Termux] [Linux] : apt-get up

Razor Kenway 8 Oct 24, 2022
This is a Docker-based pipeline for preparing sextractor-ready multiwavelength images

Pipeline for creating NB422-detected (ODI) catalog The repository contains a Docker-based pipeline for preprocessing observational data. The pipeline

1 Sep 01, 2022
Generates images with semantic content from distribution A in the style of distribution B

A2B Generates images with semantic content from distribution A in the style of d

Richard Herbert 2 Dec 27, 2021
Fluxos de captura e subida de dados no datalake da Prefeitura do Rio de Janeiro.

Pipelines Este repositório contém fluxos de captura e subida de dados no datalake da Prefeitura do Rio de Janeiro. O repositório é gerido pelo Escritó

Prefeitura do Rio de Janeiro 19 Dec 15, 2022
Scalene: a high-performance, high-precision CPU, GPU, and memory profiler for Python

Scalene: a high-performance CPU, GPU and memory profiler for Python by Emery Berger, Sam Stern, and Juan Altmayer Pizzorno. Scalene community Slack Ab

PLASMA @ UMass 7k Dec 30, 2022
Python-Roadmap - Дорожная карта по изучению Python

Python Roadmap Я решил сделать что-то вроде дорожной карты (Roadmap) для изучения языка Python. Возможно, если найдутся желающие дополнять ее, модифиц

Ruslan Prokhorov 1.2k Dec 28, 2022
Allows you to purge all reply comments left by a user on a YouTube channel or video.

YouTube Spammer Purge Allows you to purge all reply comments left by a user on a YouTube channel or video. Purpose Recently, there has been a massive

4.3k Jan 09, 2023
This is a modified variation of abhiTronix's vidgear. In this variation, it is possible to write the output file anywhere regardless the permissions.

Info In order to download this package: Windows 10: Press Windows+S, Type PowerShell (cmd in older versions) and hit enter, Type pip install vidgear_n

Ege Akman 3 Jan 30, 2022
Some ideas and tools to develop Python 3.8 plugins for GIMP 2.99.4

gimp-python-development Some ideas and tools to develop Python 3.8 plugins for GIMP 2.99.4. GIMP 2.99.4 is the latest unstable pre-release of GIMP 3.

Ismael Benito 53 Sep 25, 2022
[draft] tools for schnetpack

schnetkit some tooling for schnetpack EXPERIMENTAL/IN DEVELOPMENT DO NOT USE This is an early draft of some infrastructure built around schnetpack. In

Marcel 1 Nov 08, 2021
It is a personal assistant chatbot, capable to perform many tasks same as Google Assistant plus more extra features...

PersonalAssistant It is an Personal Assistant, capable to perform many tasks with some unique features, that you haven'e seen yet.... Features / Tasks

Roshan Kumar 95 Dec 21, 2022
Source for the Fedora Silverblue and Kinoite variants.

Source for the Fedora Silverblue and Kinoite variants.

Fedora Kinoite 7 Aug 20, 2022
A napari plugin to inspect data within a cisTEM project

napari-cistem A plugin to inspect data within a cisTEM project This napari plugin was generated with Cookiecutter using with @napari's cookiecutter-na

Johannes Elferich 1 Nov 07, 2021
A parallel branch-and-bound engine for Python.

pybnb A parallel branch-and-bound engine for Python. This software is copyright (c) by Gabriel A. Hackebeil (gabe.hacke

Gabriel Hackebeil 52 Nov 12, 2022
Dyson Sphere Program Blueprint Toolkit

dspbptk This is dspbptk, the Dyson Sphere Program Blueprint toolkit. Dyson Sphere Program is an amazing factory-building game by the incredibly talent

Johannes Bauer 22 Nov 15, 2022
A Python utility belt containing simple tools, a stdlib like feel, and extra batteries. Hashing, Caching, Timing, Progress, and more made easy!

Ubelt is a small library of robust, tested, documented, and simple functions that extend the Python standard library. It has a flat API that all behav

Jon Crall 638 Dec 13, 2022
Service for working with open data of the State Duma of the Russian Federation

Сервис для работы с открытыми данными Госдумы РФ Исходные данные из API Госдумы РФ извлекаются с помощью Apache Nifi и приземляются в хранилище Clickh

Aleksandr Sergeenko 2 Feb 14, 2022