A16Z GAMES reposted this
Really excited to announce that we're joining forces with A16Z GAMES 🚀 It's been a blast working with Jonathan Lai, Jack Soslow and the rest of the stellar a16z team. We also just launched our AI-powered 3D marketplace, sharing with our community 10,000+ ready-to-use 3D assets! Check it out --> https://lnkd.in/ervAG9i3 Thank you to everyone who has supported us in our journey so far, from our customers to our angel investors, who all backed our vision in the early days. Let's change the game - on to the next challenge 💪
We’re thrilled to announce A16Z GAMES investment in Kaedim 🚀🚀 As costs soar and studios seek solutions, Kaedim's approach stands out. By empowering an internal art team with AI, Kaedim meets studios' high requirements, reduces costs, and generates valuable data in the process. Games have a cost problem. Today, >70% of games are unprofitable and it's getting worse. Since the pandemic, studio costs rose 15%-20% but end consumer game spend increased only 1%. 3D production is a major cost driver. Each AAA game requires 1000s of high-quality 3D art pieces, each of which can take days or weeks to complete. To understand the scale, The Last of Us: Part II had 2,335 people credited for the game, of which only 2% were engineers. Against this backdrop, Generative AI is poised to democratize a complex 3D creation process. However, game studios have stringent requirements around topology, performance, and control that pure AI generation isn't ready to meet... yet. One stringent requirement is topology, which is the constellation of polygons that form a 3d mesh. Studios require their assets to have optimized, flowing topology to ensure performant, artifact-free animation. Today, generative 3D struggles with topology due to architectural limitations. Most models use score-distillation-sampling (SDS) or its derivatives. SDS leverages neural rendering not surface geometry. Extracting clean meshes from NeRFs remains an open research problem. Data limitations also impact generative 3D. Models often rely on 2D priors supplemented with 3D assets. 3D datasets like Objaverse (10M assets) are relatively small, and unfortunately creative efforts like IRL object capture face the NeRF to geometry problem again. While Gen 3D can't meet all studio needs yet, it is a disruptive innovation: >100x faster & cheaper than humans, but with worse topology. As it improves, it will unlock new use cases & grow the pie. A classic case of Christensen's disruption theory. But nothing bridges the gap for today's burning market need, except Kaedim. With a human-in-the-loop, Kaedim generates clean topology. Simultaneously, they are well positioned for Gen 3D's disruption as they are being paid (rather than paying) for a data moat. Kaedim's AI-powered tools turn 3D artists into super-humans. A 3D asset that would otherwise take an artist hours and hundreds of dollars to complete can be done with Kaedim in 20 minutes for $20. Studios who use Kaedim often report 80% speed ups in their production process. This has led to rapid adoption. Kaedim has >250 studio customers and >7 figures in revenue. They are working with the largest AAA studios, film production companies, and e-commerce platforms. We're excited to support Konstantina Psoma and Kaedim on their mission to revolutionize 3D production. Together, we'll empower artists, enable studios, and unlock new possibilities for immersive content! Read more about our investment here: https://lnkd.in/eCkqJfPx cc: Jonathan Lai
Let's go!! 🚀 🚀
Well done Konstantina! Very impressive.
🚀🚀🚀
💪 💪 💪
Let’s gooo! 🔥
Bravo! Konstantina Psoma
Congrats, Konstantina and team! 👏 Super-excited for what comes next!
SO impressed Konstantina Psoma! Well done 🙂
Bravo Konstantina Psoma, well done!
andrewchen.substack.com / a16z
5moCongrats and excited to be working together!!