DALL·E 2 upcoming launch
Sam Altman mentioned the OpenAI team is hopeful for a public product launch this summer. Watching makers play with GPT-3 over the last 2 years has been exciting but I wonder how much riskier this will be from a deepfakes and social bias standpoint. What upsides / downsides do you see and what are you excited to build with it?
Looks like something cool to use and explore. Send it soon, please Sam.
I am beyond excited to build with Dalle2. I think with some tinkering one can connect and idea with its visual counterpart. Initial use cases surrounding illustration, animation, and automated storyboarding from scripts
A presentation generator where a bot asks you questions about the objectives of your presentation and auto generates compelling images based on your answers! The trick will be to ask the right questions in the right order and massage the answers to, hopefully, get back compelling images from DALL-E 2!
@gastonhummel ooo like true next gen PPT wizard, that massaging could be accomplished in a automated fashion too with the right NLP(GPT3 fine tune) middle ware
@gastonhummel I am working on that right now. Mostly text generation, but images are coming. Check out slidekick.ai
@victor_metelskiy thanks for sharing slidekick.ai! I just tried it and it works well. My initial idea is to leverage the concept in Dan Roam's "Draw to Win" book in which he describes how to create visually compelling stories that can be told with just 6 slides, irrespective of the topic. 1) What it is or who will be impacted, 2) What the impact will be, 3) Where it fits with what your audience already knows, 4) Steps for getting there, 5) Causes and effects, 6) Summary. I don't know if I have the resources to actually build this but it's something I'm interested in experimenting with more than anything else.
Creativity: "Make an image of in the style of . Or a combination of Productivity: NLP image and HTML editing. This might actually replace powerpoint consultant interns.
Upsides: more cool graphic content. Downsides: Skynet is coming.
I read that the new image generation model has a feature that lets you upload a starting image and get multiple images with good variations.
Hey Christina, Am currently using GPT-3 and its a game changer! How can we get access to DALL-E 2? Thank you, Salvatore