Meta’s long-promised photorealistic avatars are set to release in the next few months, which the company is calling ‘holograms’. I went hands-on the annual Connect developer conference to see just how Meta plans to take on Apple.
Meta announced it’s finally rolling out its immersive avatars, and they’re coming first to its Meta Ray-Ban Display smart glasses, which includes a single see-through display in the right eye.
Previously called ‘Codec Avatars’, Holograms are coming to the newly announced Meta VR Glasses too, but those are slated to arrive sometime in Spring 2027. You can learn more about Meta VR Glasses in our hands-on from Connect.
I got an opportunity to go hands-on with Holograms on both devices, which proved to be entirely different chatting experiences based on just how different the devices are in the first place.
Note: We have stylized Meta’s holograms with a capital ‘H’ to differentiate its avatars from true holograms.
Creating a ‘Gram
The Hologram creation process… okay. Let’s back up a second. These aren’t holograms by the text book definition, as they aren’t 3D light fields, but rather AI-infused immersive avatars that have more of a 2.5D quality to them. More on that below.
Anyway, the Hologram creation process isn’t the same as creating an Apple Persona, which uses Vision Pro’s built-in depth sensors and RGB camera sensors to scan your face. Instead, you simply use Meta’s iOS or Android mobile app and the phone’s selfie camera as you follow a few on-screen prompts, including improvising an answer to the question “What’s your favorite kind of day?”, reciting a short statement and laughing after it, smiling, and turning your head side to side.
All of that is important since Holograms on Display are driven by voice only, meaning the AI is doing a fair amount of work to make your digital copy act as natural as you’d expect a human to. It’s rolling out first to Display in early access in the coming months, but only in the US for now, and only on WhatsApp video chats.
Holograms on Display
Chatting with another Display-clad user, I noticed fairly quickly my simulacrum wasn’t exactly me. Still, it was close enough that if you didn’t know me, you might actually be fooled.
That said, if you did know me, you’d know I try not to look as consistently annoyed as my avatar does in the video below—I was visibly jetlagged during the capture, and probably feeling a little weird telling a room full of Meta execs my favorite day… which is April 25th by the way, because it’s not too hot, not too cold. All you need is a light jacket.
Sofien Bouaziz, Senior Director of Research and Engineering at Meta Reality Labs, had a much more convincing Hologram however, which he admitted took a few tries to get to his liking. His was, without exaggeration, indistinguishable from the very live version of him I saw just moments before he ducked into another room for the call. If I wanted to get mine that good, I was told I could try up to 10 times per day to capture and generate an avatar.
Notably, in the lower left-hand corner of the chat, you’ll see a little ghostly Hologram icon so you know when you’re talking to an avatar or the genuine article.
All told, it’s a flatscreen call experience that I think is more for smartphone users than Display users since the visible display in your right eye is so small. Still, it’s a promising bit of tech to help bridge the odd video chat gap inherent to Display.
Holograms on VR Glasses
Now this is where things get interesting. VR Glasses have both eye and face tracking via the headset’s downward-facing sensors. That last part didn’t make the keynote, but it was confirmed in my Hologram demo.
The result: a very realistic full-bodied avatar, which I was told was generated in a similar way to the smartphone face scan, albeit capturing the user head-to-toe. While not confirmed, I suspect the user will also need to do specific moves to give the AI some gesticulations to mimic, such as a neutral hand position as well as broader, more demonstrative gestures.
When asked whether users would be able to skip the mobile app and use the headset itself (Vision Pro-style) to generate an avatar, I was told the company was sticking with the mobile app since it was less physically awkward.
The avatars I saw at Connect had a distinct 2.5D quality to them. They feature stereoscopy, and feel 3D when looking straight on from a respectable distance, but notably lack any sort of real depth, making them more like some sort of 2D projection than a proper 3D avatar, which you’d expect to be able to walk around the room, fist bump, or hand you a virtual object. These can’t do that.
Still, the AI is doing a fair bit of legwork here to make avatars consistently lifelike. Those hand poses, her hair moving, the flapping of her shirt hem when she shifts from one pose to another is all AI.
It’s still early days, so there’s no telling how things can change between now and VR Glasses’ release next year, although the emphasis on not having fully 3D avatars feels like a conscious decision. It’s fairly clear Meta wants to narratively distance itself from its failed metaverse ambitions by not only offering a grownup VR headset, but also grownup VR avatars to match that will probably be good enough for one-on-one chats.
Now, are Holograms better than Apple’s Personas? It depends on what you mean by better. Personas are still floating heads, albeit very realistic floating heads with actual 3D depth. Vision Pro’s face tracking is also uncannily good.
Putting aside Holograms’ lack of 3D depth, what I saw at Connect admittedly looks very good. Still, I’m curious to see how VR Glasses interpret my own micro-expressions before judging whether its AI model is doing too much in service of simply looking like a person.
Disclosure: Meta provided lodging for one Road to VR reporter to attend Meta Connect 2026 where information for this article was gathered.
This article was originally published on roadtovr