Matt Miesnieks is Back with Dejavu

It’s dejavu for Matt Miesnieks who is back with a freshly minted company that launched at GDC 2024 as part of a16z‘s prestigious Speedrun accelerator cohort. This accomplishment is already a feather in the cap of the startup. While I knew Matt was working on a new venture, as soon as I saw the pitch, I knew that the pain point Matt described at GDC was a really big, sticky problem. So I followed up to interview him on the approach and positioning (see full video below).

Backing up, Dejavu is focused on solving the fidelity, cost, and timelines of building high-quality photorealistic 3D content that shows its flaws on high-end devices like the Apple Vision Pro. From the moment I put the device on – and from what I had heard from companies that were pushing themselves to the limits for launch – the higher fidelity of the AVP made development teams see flaws they hadn’t seen before. The high resolution of the Vision Pro exposed a pain point with its 23 million pixels, 14K display; most existing content was built for lower fidelity existing headsets.

So while Apple launched with content curated to show off the potential of the device, and that IMAX Cinema display, content creators are having a hard time affording the jump in fidelity. Also, the price points and length of time it takes to create suitably rendered content are resulting in market friction and, in some cases, delays. But beyond porting existing content to the Apple Vision Pro, everyone from Hollywood to game studios is desperately looking for alternatives to costly visual effects production for creating realistic environments.

Dejavu aims to fill the gap in fidelity above Matterport’s reality capture system but below full-fledged visual effects teams. Miesnieks shares, “So what we’re doing next with dejavu is we want to build a pipeline sort of above Matterport, but below a visual effects team, or basically try and automate that work that the visual effects team has done. And in that sense, we’re going to replicate.”

https://youtu.be/IYTbdQeVE-U

Matt Miesnieks is Back with Dejavu (https://youtu.be/IYTbdQeVE-U)

Once deployed, and for a price point perhaps in the tens of thousands, not hundreds of thousands, they have developed a rendering engine that can take high-quality 3D scenes and make them work at full photo-real resolution on any device, and in a fraction of the time. Says Miesnieks about the current state of the art, “You need to hire a team of tech artists, you know, three, four, five artists working for a few months to, you know, optimize that scene down to a smaller mesh, lower quality textures, baked in lighting, and it all needs to be done by hand just so that it looks good. And that ends up costing hundreds of thousands of dollars. What we’ve built with dejavu is a way for us to take that scene and its full raw file form, run it through our engine and give it back to you as a run time that will work at that full photo-real resolution and run on any device.”

For those who have been following Matt’s journey as a serial entrepreneur from Layar (acquired by Blippar in 2014) to 6D.ai (acquired by Niantic in 2020), and now dejavu, there are certainly elements that do feel like Déjà vu as he seeks to build another foundational element for spatial computing. And, especially for those of us who remember how AR Cloud brought so many brilliant people together to build the foundations of the eventual “Mirrorworld”,

Miesnieks sees this more as a continuation of that thread, rather than a move away from it. Investors (and the rest of us) are curious to know where this all leads, and what that Mirrorworld of 2025 might look like, where real-world locations are captured and transformed into interactive, multiplayer experiences. This is where the vision to create a platform hosting millions of virtual places with various practical utilities emerges:

“Like I could run a, you know, SEAL team six mission training, how to rescue hostages from the same house. You’re dropping an Ikea couch and wondering whether to buy it. But all of it is from our point of view, you know, all of these are 3d data of real places that we’re hosting and making them available to you that feel photo realistic.” – Matt Miesnieks, CEO, dejavu

So as dejavu bridges this gap, they’ll be able to provide everyone from indie creators and agencies to Hollywood studios with a more affordable and efficient pipeline for creating high-quality 3D models. While we wait for the Mirrorworld to emerge, it is obvious that the pieces are coming together, and that capturing photorealistic real-world locations and turning them into interactive, multiplayer experiences is a huge piece of the puzzle.

Listen or watch the full interview on the Spatial Intelligentsia video podcast (use the timestamps below to jump to a topic) or watch above.

00:00 Introduction and Background
00:51 The Fidelity Issue with Apple Vision Pro
08:21 Capturing High-Resolution 3D Scenes
10:17 Differentiating 360 Look-Around and 3D Environments
14:02 Challenges of Creating High-Quality 3D Models
22:34 The Gap Between Matterport and Visual Effects Teams
26:41 The Mirror World and the Future of Deja Vu
30:04 Interactive Destinations and Applications
31:11 The Limitations of Traditional Property Viewing
32:18 The Importance of Time of Day and Lighting
33:15 The Potential for Multiple Uses of Locations
34:11 The Value of Real Estate and Virtual Real Estate
35:32 Creating a Catalog of Places for Various Use Cases
36:24 The Power of Presence and Emotional Reactions
37:08 The Potential of Virtual and Real Places
38:06 Simplifying the Experience and Focusing on Quality
39:05 The Quest for Indistinguishable Virtual and Real Experiences
40:05 Exploring the Emotional Impact of Virtual Places
41:13 The Resurgence of Virtual Real Estate
43:07 The Potential of Spatial Video and Audio