Blockade Labs is releasing an update for its Skybox AI as the premier 360-degree AI seamless environment art creation platform. And now the quality is much better than before, with the ability to create and download in 8K visual quality.
That makes it suitable to automatically generate imagery that can be displayed by the Apple Vision Pro mixed reality headset, said Marguerite deCourcelle, CEO of Blockade Labs, in an interview with GamesBeat. deCourcelle will be a speaker on an AI and games panel at GamesBeat Summit 2024. (You can use this code for a 25% discount: gbs24dean25).
The company first released Skybox AI a year ago. Since then, the company has increased the utility of Skybox AI with many features such as remix, sketch-to-skybox, depth maps and HDRIs, skybox-to-3D-mesh, new style options and more.
So far, more 1.5 million users to date from the entertainment, gaming, VR, medical, real estate, mental health, and education sectors, as well as hobbyists. And over 10 million Skyboxes have been created. Companies like NVIDIA, Accenture, and T-Mobile are using Skybox AI for their product demonstrations.
And there are more than 5,000 paying customers using Skybox AI.
And today Blockade Labs is releasing new updates that improve the quality of development with its Skybox AI technology. The new Model 3 update drastically improves the quality and resolution of all Skyboxes.
“Our big release is the 8K resolution released with our Model 3, which has 77% more pixels than the Model 2,” deCourcelle said. “It’s a major upgrade and increase in fidelity. Our Apple Visoin pro developers are creating apps with us and we’re very excited to see that product in action.”
Now with the ability to create and download in 8K, developers for the Apple Vision Pro have a way to easily create for the platform. Also being released in the new feature, Remix Your 360 — which allows users to upload any 360 images to be remixed by Skybox AI.
Origins
deCourcelle has been in the crypto scene for a decade.
She started in fine art as a painter. In 2013, she started getting into crypto and hiding private keys into physical artwork using converting the data into visual design. She created the first crypto-puzzle in 2014, in a painting called The Dark Wallet.
“So essentially, the Bitcoin was hiding in plain sight. And, and then I continued with puzzle mechanics. Then the public will try to figure out how that was designed. And these became very viral,” deCourcelle said. “We were mining Bitcoin at the time, and it was a really fun like medium. The idea of combining money with art just seems like a sick gross idea, really. But I was fascinated by how it just grabbed everyone’s attention.”
In 2014, she did her most famous work and it took three years for the public to solve it. That happened during the Bitcoin bull run. Her occupation was inventing crypto puzzles.
“That puzzle design interest in creative work and working with experimental emergent technologies, I made a lot of friends with very talented developers,” she said. “These puzzles almost acted like a CIA recruiting tool. All of a sudden, they had the best of the best talent that wanted to make these crazy wild experiments with me.”
She went on to found Blockade Games in January 2018, and that team created a game called Plasma Bears and she went on to “more weird crazy ideas.” Then she teamed up with her friend Adam B. Levine, who had previously launched an AI company. They decided to build an art creation platform by merging the two companies. She started Blockade Labs with Levine in December 2022.
Then they brought Skybox AI to market in the fastest way they could, progressing from 2D art to 3D art. And now they are releasing updates for it. It can generate art assets that are normally really hard to make and require the work of technical artists who don’t really enjoy making them, she said. This can help companies deal with a shortage of artists, she said.
Over time, they want to integrate with other platforms like Unity to reach more developers. She foresees metaverse companies using such art creation programs because they don’t have enough artists to build out their immersive and interconnected worlds.
Skybox AI is focused on generating 3D environment art using generative AI and large language models. And it is creating apps that empower creators to transform their imagination into digital realities through immersive design.
One of the interesting challenges is coming up with art for mixed reality spaces where the generated art has to match with the physical environment of the real world that is visible via the mixed reality passthrough cameras.
“We have demos of that in action and it’s something we’re really excited about working on with the Apple Vision Pro folks,” she said.
She said that her company was featured by Jensen Huang, CEO of Nvidia, last year during his keynote talk at Siggraph. She foresees technologies coming where you can automatically generate an environment image and then say, “Delete all the trees” and it magically happens.
The company raised money from Animoca Brands, Spatial Capital, 6th Man Ventures, Galaxy Interactive,
Baller Ventures, Red Beard Ventures, Polygon Ventures, Digital XYZ and Yunt Capital.
To date, the remote-work company has raised $2.5 million, and it plans to raise more soon. The company has about 13 full-time people. They include artists like Random Ghost, a famous cyberpunk curator and brand developer. He helps with the model training.
Making better generative art
Blockade Labs’ Skybox AI is an innovative text-to-360 8K image generator tool with API support, enabling
rapid creation of high-resolution content through user-guided text and input. Developed in-house with
minimal third-party dependencies, it offers the fastest, easiest, and most cost-effective solution for
creating customized professional-grade high-definition 3D immersive environments and worlds, deCourcelle said.
With advanced AI and ML capabilities, Skybox AI empowers users of all levels to generate and export on-
demand seamless customized 360-degree spaces and HDRI’s for various applications including gaming,
entertainment, architecture, education, and more.
One of the problems of computer-generated art is that it may be easy to create a prompt to generate something new and imaginative. But for an artist, it may be a lot of work or an impossible task to refine that first image into something that looks exactly like what the artist wants. It might also be hard to generate one scene and then move on to the second or third scenes.
To address that, she said Blockade Labs has always designed for usability. You can use a tool to sketch something and the computer vision will recognize what you are sketching. You could create a horizon and a tree in your sketch and then it can generate the tree or the property line a needed.
“That introduces an element of control, and then we have something we call remix,” she said. “We have various preset styles that are all curated by our team and numerous styles. And so that allows you to have a consistent control of and try out different looks with the same scene.”
Skybox AI benefits
deCourcelle said that her tools save money and time, as world building is costly and time-intensive. Skybox AI allows creatives to rapidly ideate, prototype, customize, and integrate their creations.
With Skybox’s instantaneous creation to generate and test concepts, the development process is streamlined, improving time to market, and lowering costs. Unity’s recent gaming report cited 54% of
developers are using AI to build one-of-a-kind experiences and, that developers only have one to three
months to prototype. That’s not always enough time to get a product or game right.
Skybox AI is also an interoperable API that can be used with all major VR, gaming, and interactive development platforms and engines. It’s a Unity Verified Solution and part of NVIDIA’s Inception plan, easily integrated into Unreal Engine and the Omniverse, and creations can be exported in GLB. Delivering 8K resolution, Skybox AI is ideal for developing for the Apple Vision Pro.
It’s also completely customizable. Skybox AI generates seamless scalable 360-degree 8K environments/skyboxes, stunning HDRI lighting, and 3D world meshes that can be edited, stylized, customized, and imported for use in preferred software.
These 24-bit color depth images automatically adjust to the right proportions for equirectangular skyboxes. Features such as Sketch to Skybox, Remix, and Skybox-to-3D-mesh allow for full customization of every environment and layer.
“We’re aiming to be the top of the funnel from everything 360 with the workflow and the 3D workflow and all the API engine integrations that make us the go to for that solution. It’s not a matter of just creating a 360 image. They don’t have solutions for fixing the poles or for fixing the seams and making it seamless, which we all already do,” she said. “This makes us defensible in the way that we are very much focusing on that particular use case. We’re not trying to be all things generative AI, but just helping you establish your framework for the world.”
Blockade Labs argues that it is years ahead of its rivals since it has ML experts creating the tech in-house, making the Skybox AI product years further along in solutions for generative services than competitors.
She added, “One thing that Blockade Labs has done that’s pretty different than a lot of your other generative AI companies, even in the realm of world building, is we’ve developed interoperable APIs for an open framework that any type of developer can build wherever they’d like, on whatever 3D engine on whatever platform they would like to be building.”
And she said, “We’re not solving the hard game dev problems of building the metaverse itself as far as infrastructure. Those are problems that game devs are already really good at and have solutions for. We solve for their content pipeline and building agent systems that can interact with their existing infrastructure.”
In creating art for augmented reality, deCourcelle said Blockade Labs can “develop the world in one shot, where you have the entire world and then you use computer vision to then understand everything that was just generated in a lightweight way.” That means it doesn’t use tons of computational resources, and that the work can be done in under 10 seconds.
Right now, the company is focused on generative AI and art.
“We want to do one thing really well and then continue to open up usability for further development that might be considered more computationally intensive,” she said.