Philippe Lewicki 🛡

This is a 𝟯𝗗𝗚𝗦 of 𝗟𝗲 𝗟𝗼𝘂𝘃𝗿𝗲. Would love to hear perspectives from people in: - film / VFX / VP - real estate - architecture - tourism / culture - gaming / XR - insurance / built environment #GaussianSplatting #3DGS #VirtualProduction #XR #DigitalCapture

I’ve started building a library of 𝟯𝗗 𝗚𝗮𝘂𝘀𝘀𝗶𝗮𝗻 𝗦𝗽𝗹𝗮𝘁 𝘀𝗰𝗮𝗻𝘀 across Paris, Los Angeles, and private properties. Not just “cool 3D” — trying to validate where this creates real business value. What would you want to do with scans like these? #VirtualProduction

Be honest: Mentra glasses don't sound exciting at first glance.No display. Just a camera and mic. The magic is under the hood. They're built on a rock-solid, tested Open Source framework. This makes them the most customizable and versatile AI glasses on the market.

The G1 set the bar for lightness and style. Now, how much better is the G2? Even Realities is raising the stakes again. Check out the unboxing of G2 and R1 below to see what's new. Follow along for the full review coming soon!

The future is on your face.Over a dozen AI/AR glasses launched last year. I didn't just unbox them. I wore them for weeks.Why? Because you can't review a pair of glasses for just one day..Follow me if you want the honest verdict before you drop your cash.

@VRDesktop On my last LBE retail experience using mixed reality I started with the openXR / unity 6 but had to move to the Meta sdk because of shader not rendering right in passthrough

VR Developers, don’t use the Meta XR SDK to develop your games. Whether you use Unity or Unreal, you should use the built-in OpenXR support in the engines. The Meta XR SDK only works with Quest & Link; it is a buggy mess hardcoded to not work with other runtimes and headsets.

@SadlyItsBradley Yes, you can test the difference with Even reality or the new rokid glasses. They compromised the color for binoculars

I’m now convinced that smart glasses with only a monocular display cannot be mass market anytime soon I think with current technology: people are way better off without a display rather than the tradeoffs of adding a single one and I don’t expect Apple or Google to solve it

@mayankja1n I have a couple of those

Real ones know the origin story.

@Roblox is developing an open‑source AI that creates 3D models from text or images. The new v0.5 update boosts fidelity and adds bounding‑box controls. corp.roblox.com/newsroom/2025/… (GitHub link in comments)

@Snosixtytwo Are the players animated with paise detections and 3d rig avatars? And the performance gain is on the parallel processing of multiple players?

SIGGRAPH is over, but the papers remain! If you missed it, our team presented Echoes of the Coliseum: Towards 3D Live Streaming of Sports Events Check it out here: humansensinglab.github.io/basket-multivi… doi.org/10.1145/3731214

#AWE2025 lots of AR glasses to try and develop on!!

About the Snap spectacles

Excited to return at AWE this year. I will be part of a panel about the award winning TMobile Retail experience for Formula 1. We will how we built and deliver a successful immersive location base retail experience: awexr.com/usa-2025/agend…

@_LucasRizzotto Maybe, is reality still a thing?

The AI slop world we're headed into will spark a new golden age for documentaries. If everything around you is fake, suddenly anything grounded in reality will have 100x the weight.

@marc_habermann This looks great. I have a question: All the renders have a specific focal length and camera angle. Is it just for consistency, or did you experience artifacts in lower view angles of the rendered avatars?

Want to have explicit control over skeletal pose, facial expressions, and hand gestures for your photorealistic virtual avatar that can be learned from multi-view video? Check out our #Siggraph2025 work, EVA, which presents an expressive, full-body, and photorealistic avatar.

@marc_habermann @janusch_patas Looking forward to test it

@janusch_patas We will soon release source code and data :) concerning skeleton and mesh: our representation comes with both.

[SIGGRAPH '25] EVA: Expressive Virtual Avatars from Multi-view Videos Contributions: 1. We introduce EVA, a novel method enabling full-body control with real-time, photo-realistic renderings, robustly handling loose clothing dynamics and various facial expressions. 2. We develop an expressive deformable template that generates a deformable human template mesh and employs a multi-stage tracking algorithm to faithfully capture facial expressions, body motions, and non-rigid deformations from multi-view videos. 3. We propose a disentangled 3D Gaussian appearance module that models the body and face independently, ensuring separated control and high-quality renderings.

@janusch_patas Separating the body and face is clever

@benz145 Yes, the same. It's a weird video. I was hoping to get details on the tech they will do, but they just talked about themselves.

Anyone else find this ‘make our own documentary about ourselves’ bromantic marketing… weird?