Here are some comparisons between the new @OpenAI DALL·E 2 image generation model vs VQGAN+CLIP "independent" model from last year... 🤯 https://t.co/qCQj6FJv3t