Submit a Project

ML x ART

A showcase with creative machine learning experiments

👋 I'm Emil, a resident at the Google Arts & Culture Lab. I've curated this collection.

Year
Artist
Technology
Theme
Medium
Default order

Sofia Crespo

CNN generated blend of corals

CNN

Fabin Rasheed

Use GPT-3 to describe a painting step-by-step and paint it according to the instructions

GPT

Merzmensch

A montage with several 3D Ken Burns Effect generated videos from images

3D Ken Burns Effect

Robbie Barrat, Acne

Train a GAN on Acne clothing, generate images with the GAN, and create physical products according to the GAN images

GAN

Synflux, Hatara

A GAN trained on 2 million birds and converted to double Jacquard knit format

GAN

Fadi Al-Ghawanmeh and Kamel Smaïli

Translating voice to RNN generated MIDI files to improvise Arab maqamic music in real-time

CNN, RNN

Mariel Pettee, Chase Shimmin, Douglas Duhaime, Ilya Vidrin

Generated choreography with VAEs trained on high-quality motion capture data of improvisational dance

VAE, RNN

Scott Eaton

Use Pose Detect and a custom pix2pix HD to turn a dancer into sculptures and then applying Optical Flow

pix2pix, Optical Flow, GAN

Nadine Lessio

Use Alexa voice assistants and add heuristics for it to disagree

Voice Detection, IoT, Hardware

Refik Anadol

A StyleGAN trained on 100 million pictures of New York; the GAN-generated images are then animated with Point Advection in Houdini

GAN, Houdini, ESR-GAN, StyleGAN

Xander Steenbrugge

Visualizing sound by interpolating a GAN trained on paintings

GAN

Bård Ionson

art-DCGAN trained on 20K images of oscilloscope art to mimic the effect when you close your eyes and rub them

art-DCGAN, GAN

Jürgen Schmidhuber

Create an artwork with the shortest algorithm while at the same time captures the essence of an object, using Kolmogorov Complexity

Occam's razor, Kolmogorov complexity

Bjørn Karmann, Google Creative Lab

A CNN model in a device that collects data and is auto trained to make an action

CNN, Hardware

Mario Klingemann

A blog post made with Markov Chains starting with religion and ending up discussing Adobe After Effects

Markov Chains

Billie Eilish, Google Creative Lab, YouTube Music, IYOIYO, Plan8

An interactive, gradient-based audio and video alignment of thousands of covers of Billie Eilish’s “Bad Guy”

Joel Simon

Use a genetic algorithm to generate faces and the dlib face classifier for the objective function

Face Recognition, CNN

Robbie Barrat

GPT-2 to generate a set of rules describing a drawing, inspired by Sol LeWitt

Transformer, GPT

Alex J. Champandard

Extract patches from an image to generate a texture with a GAN discriminator

GAN, Discriminator, Inpainting

Alexander Mordvintsev

Run an image through a pre-trained CNN, and then upscale the image backward. For each layer, adjust the image according to the difference between the activations and the image

CNN, Activations, DeepDream, Gradient Ascent

Jason Salavon

Reconstruct images by introploating a collage of GANs and guiding it with a GAN discriminator

GAN, Discriminator

Helena Sarin

SNGAN trained on doodles and upscaled and colorized with CycleGAN

GAN, CycleGAN, SNGAN

Derrick Schultz

Style Transfer with content from waves footage and style from StyleGAN

StyleGAN, Style Transfer, Superresolution

Guillaume Slizewicz

A mobile broadcasting robot using MobileNet V2 trained on COCO, OCR, and text to speech to amplify messages in public spaces

OCR, Text-to-speech, COCO, MobileNetV2

Submit projects here 💅

☞ ML x ART on Twitter

Copyright © 2021 Emil Wallner

All artworks © of their respective owners