简体中文
EnglishWhat is justinpinkney.com?
Updated
justinpinkney.com is the personal website and blog of Justin Pinkney, a place where he writes about creative uses of machine learning. It is an occasionally updated blog rather than a product or service: visitors can browse posts, read an about page, find links to his presence elsewhere on the web, and subscribe to an RSS feed.
What you will find there
The site is organized around Justin's experiments with machine learning, especially generative models applied to images. The homepage lists recent posts, and a "See more posts" link leads to the full archive.
Recent posts listed on the homepage
| Post | What it covers | Date |
|---|---|---|
| Trailer Faces HQ Dataset | A dataset of 187k high-resolution images of faces with varied expressions taken from movie trailers | August 2024 |
| Experiments with Swapping Autoencoder | Things you can do with a model that separates images into style and content components | May 2024 |
| Image Mixer Diffusion | Blending and mixing images by fine-tuning an Image Variation Diffusion model | April 2024 |
| Face Mixer Diffusion | Experiments training face-conditioned diffusion models | January 2024 |
| Würstchen v2 Pokémon | A quick experiment fine-tuning Würstchen v2 for text-to-Pokémon generation | January 2024 |
Who it is for
The content suits readers interested in applied generative modeling: dataset releases, fine-tuning experiments, and write-ups of what worked with models like Swapping Autoencoder and diffusion variants. If you are looking for tutorials, documentation, or a commercial tool, this is not that kind of site; it is a record of personal experiments.
How to use the site
- Read the blog: start from the latest posts on the homepage, or follow "See more posts" for the archive.
- Learn the background: the about page describes who Justin is and what he works on.
- Follow updates: subscribe to the RSS feed if you want new posts without checking back.
- Find him elsewhere: the site links out to other places on the web where he publishes.
One practical note: the posts are written as experiment logs, so expect observations and results rather than step-by-step instructions you can copy directly. The dates on the homepage are the most reliable signal of how current each write-up is.