Year: 2022

Another favourite startup of mine, making huge progress. I still vividly remember the TED Talk I saw in person in Tanzania at TEDGlobal 2017. >

@Zipline is now the national drone service provider for Rwanda https://t.co/5qieVSexgV https://t.co/eWJwkJvUuz

RT @mattdesl: It would be *extremely* unfit to use IP and copyright as a kludge to block AI training on real-world data and force these mod…

New feature in Dream Textures 0.0.9: Texture Projection – texture an entire scene at once using the depth to image model.

The video here is better than my screenshots:
https://t.co/SfQmHCUG7o

Phenomenal job by @CarsonKatri https://t.co/PeaoSnRWtP

Did I ever mention that @replit is one of my favourite startups ever? Ever? https://t.co/a5uWMgbQMr

Funny: On one side, #NFT speculators count on media scarsity. On the other, the #stablediffusion community are unlocking infinite media abundance.

#stablediffusion used to generate music (via spectrogram generation). Holy cow.

Riffusion https://t.co/l6NsuJUReX https://t.co/6FSFDJyS3A

@mattdesl Absolutely incredible, Matt. Thanks for sharing this with the community.

@philbarden It’s not active yet in the UK. Often, the rollout is US first. It would be interesting to hear your thoughts on this once you have access to it.

I’ll guess: the story provides a more relatable context. Maybe the implementation will be frictionless, in Amazon style.

@philbarden Interestingly, via #GenerativeAI, a seller like Amazon might automatically generate a story based on past purchases, increasing the relatability even more.

Let’s say that I purchased cat food for years. Amazon could use a large language model to summarize purchases, then

@philbarden feed that summary to a diffusion model like #stablediffusion, to generate a very relatable background image of the story + the item that they want to encourage buyers to purchase.