adobe is cooking a new inpainting method that understands context extremely well! ELI5: it can insert stuff with the correct perspective and add/remove connected reflections/shadows 👀

the thing is called OmniPaint project page: code (coming soon):

The secret to kids learning faster and better has been known since 1960. But it’s been impossible to implement. Technology is just starting to catch up, so we can finally use what we’ve known for years to radically change how we educate kids. Want to learn how? Read this:

Adobe is so slow to the draw. Any idea when they'll release this? ChatGPT 4o might have already made this outdated.

i wouldn't take my chances and wait for it

Looks promising 👀

indeed!

Is this available yet to use?

it's adobe 😅

openai and gemini already did it. cheaper.

Would like to see tests of this with OpenAI and Gemini xD bet they could do it

yeah, multimodality is a different beast