Reve Image is our first step towards world-class image generation — and you can experience it for free today 🌜
(🔊)

An image that speaks for itself.

It takes me about 2 hours to write a bespoke agent to perform difficult coding or semantic (prompting) tasks for me (with some creative tasks coming). Work is increasingly becoming meta-work. Rather than performing some tasks, we'll describe requirements, define metrics, and provide validation data.

I watched the last few minutes over and over. It's chilling, and one of the best payoffs I've ever seen.

I completely disagree with the backlash against Joker: Folie à Deux. It's a brilliant film, and definitely one of the best in the superhero/villain genre — a two-movie head fake that ends up not being about Joker at all, but the need for the oppressed to anoint a champion who will burn it all down.

I need to try this. Is it able to generate different voices for different characters? Can it do multi-character dialogue?

Generating text with DeepFloyd IF.

I realized today that I use a binary search algorithm to separate cheese slices cleanly (divide in half, then divide again, and keep going until you only have two slices left). It would be much more efficient if I could make my lunch semantically.

I guess it's time to rebuild.