Just a reminder, I made a whole series last summer about what something like DALL-E 2 could mean for the future of creativity
GPT-X, DALL-E, and our Multimodal Future
1. Transcribe your online lecture with OpenAI whisper
2. Summarize your transcription with GPT-3 in simple terms
3. Ask GPT-3 follow up questions to improve your understanding
4. Have GPT-3 generate homework questions to test and evaluate your understanding of the topic at hand
"Photo of an oversized grizzly bear dressed like a hipster wearing sunglasses during summer and getting a brain freeze from eating too much ice cream in front of a marina"
#dalle2
#dalle
"Photo of a confident oversized grizzly bear dressed in an abstract high fashion outfit wearing sunglasses on the fashion runway at Paris fashion week"
#dalle2
#dalle
It didn't paint but whatever "Photo of a grizzly bear, wearing a denim jacket, on the roof of a building painting the new York Skyline at sunset"
#dalle2
#dalle
Stable diffusion is moving fast and making refreshing moves like no other group in the industry. As someone just drawn to energy and momentum, you simply cannot ignore this
OpenAI literally swapped out the "DALL-E-2-no-faces-allowed" model for "DALL-E-2-faces-allowed-very-high-res" ... how long has this been in the back pocket?????
My hunch is that by the end of the year or so you will not be able to find the weights for Stable Diffusion anywhere. I won't be surprised if GitHub and hugging face will be forced to take them down.
OpenAI casually updating DALL-E 2 with photo realistic face support at commercial scale shows us how much they're toying with the entire industry at this point