Unbelievable development in the text-to-image AI world. I can't describe well enough how this new version of DALL-E 2.0 is so amazing. But I can list what I understood: 1. It has a free version model. Here's the link (also in the video's description by TwoMinutePapers): https://huggingface.co/spaces/stabilityai/stable-diffusion 2. Diffusion! Customization! and other montage tricks are possible (not in the free version probably because of hardware limitations). Diffusion means: https://youtu.be/nVhmFski3vg?t=245 Animation trick in the cropped picture (don't miss this 10 second wonder): https://youtu.be/nVhmFski3vg?t=343