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
python scripts to perform coin die clustering (performed on Riedones3D).

python scripts to perform coin die clustering (performed on Riedones3D).

Sofiane 2 Apr 29, 2022
Python NZ COVID Pass Verifier/Generator

Python NZ COVID Pass Verifier/Generator This is quick proof of concept verifier I coded up in a few hours using various libraries to parse and generat

NZ COVID Pass Community 12 Jan 03, 2023
Markov Chain Composer

Markov Chain Composer Using Markov Chain to represent relationships between words in song lyrics and then generating new lyrics.. ahem interpretive po

Kylie 85 Dec 09, 2022
A fancy and practical functional tools

Funcy A collection of fancy functional tools focused on practicality. Inspired by clojure, underscore and my own abstractions. Keep reading to get an

Alexander Schepanovski 2.9k Dec 29, 2022
Repository for DNN training, theory to practice, part of the Large Scale Machine Learning class at Mines Paritech

DNN Training, from theory to practice This repository is complementary to the deep learning training lesson given to les Mines ParisTech on the 11th o

Alexandre Défossez 6 Nov 14, 2022
chiarose(XCR) based on chia(XCH) source code fork, open source public chain

chia-rosechain 一个无耻的小活动 | A shameless little event 如果您喜欢这个项目,请点击star 将赠送您520朵玫瑰,可以去 facebook 留下您的(xcr)地址,和github用户名。 If you like this project, please

ddou123 376 Dec 14, 2022
A collection of tips for using MISP.

MISP Tip of the Week A collection of tips for using MISP. Published via BelgoMISP (todo) and this repository. Available in MD and JSON. Do you want to

Koen Van Impe 52 Jan 07, 2023
Helps to arrange nodes

Relax brush for nodes, helps to arrange nodes easier.

336 Dec 15, 2022
Simple GUI menu for micropython using a rotary encoder and basic display.

Micropython encoder based menu This is a simple menu system written in micropython. It uses a switch, a rotary encoder and an OLED display.

80 Jan 07, 2023
LinkML based SPARQL template library and execution engine

sparqlfun LinkML based SPARQL template library and execution engine modularized core library of SPARQL templates generic templates using common vocabs

Linked data Modeling Language 6 Oct 10, 2022
Python library to interact with Move Hub / PoweredUp Hubs

Python library to interact with Move Hub / PoweredUp Hubs Move Hub is central controller block of LEGO® Boost Robotics Set. In fact, Move Hub is just

Andrey Pokhilko 499 Jan 04, 2023
Data on Free Food at MIT

MIT Free Food Timing Procrastinating research by plotting data on how long it takes emails on the free-food at mit edu mailing list to go through. Dat

Peter Sharpe 2 Nov 01, 2021
Bionic is Python Framework for crafting beautiful, fast user experiences for web and is free and open source.

Bionic is Python Framework for crafting beautiful, fast user experiences for web and is free and open source. Getting Started This is an example of ho

14 Apr 10, 2022
Enjoy Discords Unlimited Storage

Discord Storage V.3.5 (Beta) Made by BoKa Enjoy Discords free and unlimited storage... Prepare: Clone this from Github, make sure there either a folde

0 Dec 16, 2021
Make your functions return something meaningful, typed, and safe!

Make your functions return something meaningful, typed, and safe! Features Brings functional programming to Python land Provides a bunch of primitives

dry-python 2.5k Jan 03, 2023
This Open-Source project is great for sensor capture and storage solutions.

Phase 1 This project helps developers in the creation of extended realities that communicate with Arduino and require the security of blockchain stora

Wolfberry, LLC 10 Dec 28, 2022
You will need to install a few python packages for this one.

Features Bait support Auto repair will repair every 10 catches Anti detection (still a work in progress) but using random times and click positions Pr

12 Sep 21, 2022
Материалы для курса VK Углубленный Python, весна 2022

VK Углубленный Python, весна 2022 Материалы для курса VK Углубленный Python, весна 2022 Лекции и материалы (слайды, домашки, код с занятий) Введение,

10 Nov 02, 2022
An addin for Autodesk Fusion 360 that lets you view your design in a Looking Glass Portrait 3D display

An addin for Autodesk Fusion 360 that lets you view your design in a Looking Glass Portrait 3D display

Brian Peiris 12 Nov 02, 2022
Recreate the joys of Office Assistant from the comfort of the Python interpreter

Recreate the joys of Office Assistant from the comfort of the Python interpreter.

Louis Sven Goulet 3 May 21, 2022