The Checkpoint #1 – Weekly AI art news, resources and inspiration

The Arcology

News and articles

Video diffusion – Researchers from Google demonstrate creating high-quality video by extending image-based diffusion techniques. Read the paper at arXiv.

Topaz Labs launch Gigapixel AI version 6 – Big update to Topaz's excellent upscaler with native M1 support, redesigned UI and a 50% performance boost.

Using initial images with VQGAN+CLIP – What initial images are, how they work, and how to use them effectively to create better art. Disclaimer: written by me.

Genovese Generatives (24 Styles by Vito Genovese) @DMTFL_AI

you'd think I'd get used to how fun these colors that come outta VQGAN+CLIP are by now. I am 363 days into my adventure of AI image synthesis and it continues to get better & more exciting by the day. what the fuck https://t.co/acZNT26jNv https://t.co/ZuYIsn4jop

Featured notebook – Latent Diffusion

THE CHEECKTONT (A 1970s sci-fi poster for 'The Checkpoint')

New text-to-image notebook from @multimodalart using the Latent Diffusion model from CompVis and the LAION 400M dataset, with improvements by @rivershavewings. Not only is it capable of impressive results, it's one of the first models to create readable text.

It's worth trying for the text generation alone, though be aware the LAION dataset is unfiltered so may contain biases or produce NSFW results.

Try it out:

• Colab notebook

• Hugging Face

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.