The Checkpoint #2 – Weekly AI art news, resources and inspiration
News and inspiration
KnownAI film festival call for entries – Open submissions of films created, or enhanced, by AI technology will open on Monday, April 11th, and will run until Thursday, April 28th 12:00pm CST. Get your film in now!
Midjourney release version 2 of their already amazingly good text-to-image model. Check it out or sign up for the beta.
Disco Diffusion 5.2 released – Now comes with VR mode, meaning it can create video suitable for a VR headset.
Proverbs - Shardcore tries out Latent Diffusion on hundreds of different proverbs to see how well it handles metaphor. Not well, it turns out. NSFW-ish and straight from the uncanny valley, but fascinating.
Puck – An AI game designer that can invent new game ideas, playtest them to try them out, and evaluate them using things it's been told about game design.
This bench does not exist – StyleGAN model trained on over 25000 photos of benches. The world always could do with more benches.
John Maeda creates 'Generative art by John Maeda' – Fascinating to think about an AI being able to replicate your own work.
Featured model: Stitch it in Time: GAN-Based Facial Editing of Real Videos
Models like StyleGAN are good at editing still faces - making them younger, older etc. - but it's not been so easy with video until now.
Released in February, this model can semantically edit video while creating extremely stable and coherent results. Check out the samples, they're amazing compared to the other approaches. Expect to see it in a major motion picture soon...
Try it out:
Thanks for reading! Please do send this newsletter to a friend if you like it. If you have any feedback or questions, find me on Twitter or send an email.