• Real Creative
  • Posts
  • From Pompeii to Sushi Houses: This Week's AI Creations

From Pompeii to Sushi Houses: This Week's AI Creations

Meta's AI Animation Breakthrough + Legal Update on AI Art

August 16, 2024

Happy Friday!

Here’s what we’re covering today:

  • A reimagined Sesame Street-style music video

  • Behind the curtain look– how Meta creates animation at scale

  • Legal update: what this new court case means for AI artists

As always these links and over 500 other AI projects are available at realcreative.ai.

Did a friend forward you this email? Sign up for free here!

WATCH 👀

  • This Midjourney-created animated series is our historial pick of the week.

  • Created in less than 24 hours: this creator went from idea to finished project in 9 hours.

  • Tools used: lyrics: ChatGPT, song: Suno, images: Midjourney & ComfyUI Video: Runway Gen-3

  • New dream unlocked- using AI tools Midjourney, Magnific AI, Immersity to create a habitable sushi structure beyond imagination.

  • Prompt used: Midjourney v 6.1 award-winning photography of witchfanartarticles, elegantly dressed in haute couture, dresden walldiff, in the style of dark fantasy creatures, soggy, angura kei, mysterious beauty, dark gray, post-painterly, Halloween --sref 181

TRY ✍️

  • Meta developed an AI feature to animate generated images, focusing on deploying it at scale to serve billions of users quickly and efficiently.

  • They implemented several optimization techniques to reduce latency, including halving floating-point precision, improving temporal-attention expansion, leveraging DPM-Solver to reduce sampling steps, and combining guidance and step distillation.

  • To handle global traffic at scale, Meta used a traffic management system that keeps requests in the same region as their requester when possible, and implemented strategies like optimized retry settings and gradual execution delays to reduce errors and maintain high availability.

READ 🤓

  • A federal judge has allowed copyright infringement and trademark claims to move forward in a lawsuit against AI art generators, including Stable Diffusion, Midjourney, and DeviantArt.

  • The court found that Stable Diffusion may have been "built to a significant extent on copyrighted works" and created with the intent to "facilitate" infringement, which could implicate other AI companies that used this model.

  • The ruling is considered a major win for artists, as it allows the case to proceed to discovery, where plaintiffs could uncover information about how AI firms harvested copyrighted materials to train their models.

Thanks for reading!

We’re so glad you tuned in. Please share this with fellow creatives, AI aficionados, and everyone in between!


Have any suggestions, questions, or additional feedback?
Email us! [email protected]