Hands-On Experience with Gracia’s Volumetric Scenes: A Remarkable Impression

During my time wandering the AWE Europe show floor, one standout moment wasn’t even part of the official lineup. It was a spontaneous demo of Gracia’s moving volumetric scenes, right in the corridor of the event. Trust me, it was fascinating enough that I just have to share it with you.

Gracia

Gracia is a budding company with grand ambitions to be the YouTube for volumetric content, using the magic of Gaussian Splats. The goal is for creators to upload 3D scenes—likely captured on their phones—to Gracia’s servers, allowing users to immerse themselves in these virtual worlds, either alone or with friends. Thanks to Gaussian Splats, these scenes aren’t just something you view; you step inside them and live the experience as the creator intended.

Right now, Gracia is just getting started, offering an app on Quest and Steam VR where you can explore a handful of scenes their team scanned. On Quest, your options are limited to default scenes due to the hardware restrictions. The Steam VR version, however, includes community-scanned scenes. It’s technical; the Quest’s mobile hardware requires a simpler version of the splats, compromising visual quality. Despite this, running such an experience on Quest is a feat, and Gracia’s team is proud of their optimization efforts.

Some time ago, I tried Gracia on Quest and noted its promise alongside some shortcomings, like noticeable scan artifacts and a clunky interface. Unfortunately, I couldn’t test the PC version because my setup wasn’t quite powerful enough.

Thankfully, during AWE, an investor named Tipatat Chennavasin had my back. He gave me a little impromptu demo of Gracia on his high-powered laptop right then and there, showcasing how well it performs on PC and highlighting a brand-new feature: moving volumetric scenes.

Moving Volumetric Scenes

What Gracia has done is digitize several small animated scenes—typically involving people—that last between 10 and 30 seconds. By entering the Gracia app on SteamVR, you can experience short volumetric videos unfolding right before your eyes, still rendered with Gaussian Splats. This is a big leap forward; most similar programs, like Meta Horizon Hyperscape, have stuck to static scenes. That said, moving scenes are no small task: they require intense scanning and are heavy on resources, marking them as an experimental feature for now.

In fact, as reported by Upload VR, these clips involve expensive 3D scanning and extensive processing time on AWS. A staggering six minutes is needed to train just one frame, translating to weeks for a complete minute of footage, unless optimized through parallel processing. Adding to the challenge, downloading these clips would theoretically demand an internet connection of 2.4 gigabits per second, a speed rarely available to most users. However, Gracia is optimistic, aiming to significantly enhance compression within a year, which should allow streaming of this content on modern fast internet connections, and they’re also looking into cloud VR streaming.

Hands-on with Moving Volumetric Scenes on Gracia PC

Tipatat queued up Gracia and handed me a stylish Quest headset, complete with an orange headband and a matching Link cable that connected it to the PC. Sliding on the headset, I was first greeted by Gracia’s familiar—not-so-great—interface, which I quickly swatted away. Focused on the scene, I was captivated by a blond girl dancing right in front of me. And, believe me, it left me in awe. (Yes, I went there with the pun—I was at AWE, after all!)

The first thing to note was the dramatically improved visual fidelity on the PC compared to the Quest. On Quest, there’s a cyberpunk scene where the character’s quality is commendably high for mobile hardware, but the necessary compression leaves faint artifacts, lending a fur-like texture to her skin.

On PC, those issues vanished—her appearance was strikingly believable. Not completely real, as our brains are adept at spotting minor imperfections. For instance, her skin occasionally looked brush-stroked rather than natural, yet the likeness was impressive, surpassing even Metahuman models in certain aspects.

The most astonishing feature wasn’t her lifelike appearance; it was that she was moving, dancing fluidly across the scene. Because it was a volumetric scene, I could maneuver around her, viewing her performance from any angle, amplifying the realism as I approached her, much like I would with someone actually there. At one point, I knelt down to see things from a different perspective, and Tipatat jokingly quipped, "You perv, checking below her skirt?" I quickly retorted, "Just exploring different angles," though I doubt anyone bought that.

This light-hearted moment got me thinking about the potential in the adult entertainment industry. Realistic, moveable digital figures, visible from all angles and perhaps in AR within your own space… it certainly seems like the next step forward for adult content, which often leads tech innovation. This could indeed boost the advancement of Gaussian Splat technology.

A particularly nifty feature was using the controllers to shrink the dancer, placing her in my palm like a miniature figure. In this scaled-down form, the visible artifacts diminished, and her appearance became even more life-like.

Each video could be paused, allowing me to examine her features in detail, including the semi-transparent black dress she wore.

Following the dance, I explored two additional scenes. One captured a joyful moment between parents dancing with their little one, while the second involved a chef demonstrating a recipe. The latter was the least impressive, with more scanning flaws and a somewhat impersonal tone.

Final Considerations

The moving volumetric scenes on Gracia genuinely exceeded my expectations. Their potential as the future of content sharing is undeniable—imagine capturing life moments as 3D scenes and preserving them as moving Gaussian Splats to relive whenever you choose. It’s a thrilling concept.

Of course, this isn’t happening overnight. While the visual quality is consumer-ready, the capturing technology remains overpriced and cumbersome, and the data is hefty to download and render. The vision is to one day use phones and mobile headsets for both scanning and rendering, ushering in a new age of 3D memories.

For now, if you have the PCVR gear to support it, I highly recommend trying these moving scenes on Gracia—they’re truly captivating.

(Header image by Gracia)

Disclaimer: This blog contains ads and affiliate links. Clicking on an affiliate link means I may earn a small commission, which makes me a happy camper. You can read my full disclosure here.