Gen-1 Now Available to All

Runaway ML Gen-1 announced general availability

Super cool tech, thanks for sharing.

Here’s the paper describing their approach for anyone who can (or wants to attempt to) understand it:

I don’t understand most of it but I was curious what approach they were taking. It sounds like it’s a latent diffusion model, the same as Stable Diffusion, but trained with video and image-text pairs that are somewhat different in structure than what SD uses.

“We propose a controllable structure and content-aware video diffusion model trained on a large-scale dataset of un-captioned videos and paired text-image data. We opt to represent structure with monocular depth estimates and content with embeddings predicted by a pre-trained neural network.”

Their paid plan only includes one trainable custom AI generator, and they don’t provide pricing on additional ones. Weird

this is super fascinating! Does anyone know how would one try out their 2.0 version?



1 Like

Pretty sure it’s in Beta so pretty random when it comes to access as far as I can tell. This is what I here from other users as well.

1 Like

Thanks Paul, let me know if you hear otherwise - I’ve joined their Discord as well. Here’s the video that I saw that sparked my interest - Amazing AI Filmmaking is Here: Gen-2 The Ultimate Cinematic AI - YouTube.

1 Like