Playing with interesting ways of using the Stable Diffusion Image Variations model
I’m Justin Pinkney and this site is my home on the web.
My day job is as a machine learning researcher at Lambda Labs. I spend most of my time writing code, developing algorithms, training models, and encouraging people to embrace good software development practices.
Most recently updated pages:
Fine tuning Stable Diffusion to generate Pokemon
A runway app to make silly face edits using the StyleSpace of StyleGAN
A collection of the weird and wonderful uses of an exciting new vision and text model.
A dataset of thousands of high resolution aligned faces from uikyo-e prints
The story of deploying a deep neural network to the cloud and Toonifying tens of thousands of images an hour.
Use StyleGAN model blending to find your Pixar self
Mixing GAN models for weird and wonderful effects
A list of ideas I probably wont ever have time to try out.
An article looking at the connection between input space and feature space in deep neural networks and how various novel methods have been invented by generalising techniques between the two.
Baking imaginary cakes with StyleGAN
Incorporating big zoomable images into an MDX blog with Leaflet
A repository of pre-trained models for StyleGAN 1 and 2
An implementation of StyleGAN in MATLAB
Transforming normal faces to Ukiyo-e style portraits using Generative Adversarial Networks
Where to find me on the web
Image to image translation with GANs
Takes a picture and rearranges the pixels to make another
Fast and accurate face detection and alignment in MATLAB with neural networks
Command line progress indicators for MATLAB
My former life as a physicist