Legs in the metaverse and news on Generative AI. A letter from the CEO | Autumn

The long-awaited delivery of tech news

Legs in the metaverse and news on Generative AI. A letter from the CEO | Autumn

Emotional swings which Elon Musk creates for Ukrainians during these days get on the nerves. First, he tweets that Ukraine should give up its lands to the terrorist state of russia, then he threatens to stop providing Starlink services for Ukrainian defenders. Finally, he agrees to continue supplying them and even plans to enable the donation option for Starlinks. You can't sit on two chairs in this war, Elon. We continue to fight for Ukraine's victory and we won’t give up neither our citizens nor our territories.

In today’s letter:

⚡️ Reface Breaking: attending the Web Summit November 1-4 and reducing stress with Reface’s ASMR tab
💸 Number: for $1,499, Meta facilitates mixed reality remote work, full-time if you can afford it. Is Quest Pro worth the money?
🤜 Point: generative AI as a way to not only generate pictures but to create new imagery and stylize them (DreamBooth case)
🤫 Smoking room: GTA 6 spoilers leaked, AlphaTensor in the game, neural 3D rendering scaling issue, and more
🦵Oh, My Meme: thanks for the legs, Mark‌‌


‌Web Summit 2022

I was lucky to join the speaker line-up at key tech events this year, and I’m very excited to participate in the biggest one — the second year running! I’ll be onstage at Lisbon’s Web Summit this November, talking about the next step in instant AI content creation — neural 3D rendering and creating our photorealistic 3D selves.

Top speakers such as The Sandbox and OpenSea co-founders, DeepMind and Apple executives, startupers, entrepreneurs, and many other experts are scheduled. I bet you’ll be surprised. But no more spoilers; see you at the Web Summit’s Growth and Central Stages on November 3.

One of the coolest things about Reface is that every day we fill the app with something interesting for you — from new trending video collections and TikTok challenges to the most fried memes. Our new ASMR tab means total relaxation and an opportunity to feel like an ASMR blogger, including celebs. Swap, share, and crinkle chill stuff into the mike.

Enjoyable toast scratching as Margot Robbie

💸  number


Quest Pro by Meta, which generated a lot of buzz this year, was announced last week as a promising alternative to your workspace and even more in terms of VR use cases. Partly, Quest Pro is an upgraded Meta Quest 2 with additional options such as emotion tracking, adding virtual objects, and rendering optimization. However, its application can be even more comprehensive if you choose to use it professionally.

A mixed reality headset retails at $1,499 — around three times the price of Quest 2.

Meta is also working on a feature called Instant Avatars, which will allow you to scan your face or other real-life objects using your phone, upload the data to the headset, and say hello to your virtual self.

Here’s Mark’s impressive hyperrealistic demo:

This was an unusual partnership between Meta and Microsoft, bringing its most extensive services — Teams, Office, Windows, and even Xbox Cloud Gaming — to Meta’s Quest VR headsets. Well, if you want to enjoy your all-inclusive virtual Windows-driven office via Quest Pro, you’ll get it.

🤜 point

Generative AI is still very new. But good

AI text-to-image generators can forever change entertainment, filming, and CGI industries, but there’s still one big thing to deal with. Despite breakthrough technological development, machine learning is only beginning to improve its abilities to create lifelike and beautiful things. This stage is called Generative AI, which means that machines can create something new rather than analyze and process existing items. This process will soon become not only faster but in some ways better than what people make manually — in many fields.

Recently developers from Google Research team introduced text-to-image model DreamBooth AI which allows to train Stable Diffusion AI with different objects. This algorithm can associate forms and visual patterns with keywords, and personalize the result.

Let's say AI doesn’t know who or what I am, and even if I try to describe myself in detail, the algorithm's results will not reflect me very accurately, to say the least. DreamBooth helps correct this issue by enabling Subject-Driven Generation — by downloading my photos and selfies to generate a more realistic synthetic image of me.

But even more incredible is the fact that by working with Stable Diffusion, DreamBooth can put me in any given environment or create a new one, and stylize me as any character. I could be Batman, Barbie, Ken, or your crush in Twilight.

Can we call Generative AI art, or is it still a copy of something? The question remains open just like the question of the uniqueness of any new works of art — our experiences are based on the experiences other creators have already gained. But I like the point that AI can run with my creative ideas, fulfilling my dream of being an astronaut, for example. Find more inspiration and info about generative AI and its applications in this piece by Sonya Huang, Pat Grady, and GPT-3.

🤫 smoking room

Overheard at the water cooler‌‌

  • The biggest leak in game dev history: the upcoming Grand Theft Auto 6, which is still in development, has been hit by a savvy hacker. It's time to take more action to protect your corporate Slack channels. Smart-ass!
  • 53M downloads and just 9% DAU, and they are still stressed out by the imperfections of the world — Gen Z's fave app BeReal is no longer a cool story as it falls victim to the very problems it's trying to solve.
  • If you're passionate about neural 3D rendering, here's a bright article by my colleague and Reface's co-founder, Oles Petriv, on why this breakthrough technology is moving faster than the market.
  • DeepMind developed a complex 3D multiplication game that turns into an algorithm and had their AI mega brain solve it faster than ever, breaking a 50-year-old record in computer science.
  • Wanna step into the mind of a fully autonomous AI artist and try out escape rooms and an AI-driven haunted house-like world on the eve of Halloween? Here you go.

🦵Oh, My Meme

As you know, Meta has spent $10 bln to finally develop…legs for its avatars, demonstrating that Horizon Worlds still looks like a visual misunderstanding from the 2000s’. Users have been doing what they could the whole week to deal with this breakthrough invention.

Stay chill, Refacers. See you at the Web Summit soon and after in our following tech recap, as there will be even more breaking news to share.

Dima Shvets‌