Text to Image in 5 minutes: Parti, Dall-E 2, Imagen

The key ideas and intuition for how these AI image generation systems work.
Part 2: • Text to Image: Part 2...
Text + Image Generation Playlist: • Text + Image Generation

[@matveyshishov]

The most intuitive explanation I've seen. Why just 2.5k views?!

[@xuanluo6997]

Really like how the explanation is not very mathy but very intuitive!

[@mohammadyasser785]

I really like your work. Only thing i will ask for is that you add (part 1) to this video. I was a bit confused when i found part 2 first and kept looking for a part 1. Looking forward to watching more of your work in the future 🙂

[@saikatnextd]

This is awesome ❤ thanks for explaining, have you considered doing AI 101 for all the modern AI ?

[@azwaabrasid]

thank you for this. but how LLM can suddenly knows how to build a multiline triangle?

[@Dron008]

Does Midjourney have some differences?

[@AshrayMalhotra]

At 4:28 , it should be a 32*32 section instead of 8*8 (which is what was said in the video) or did I miss something?

[@g5min]

The 256x256 image is represented as a 32x32 grid of patches, where each patch is 8x8 -- hopefully that helps!