Infrastructural Layers of Meta and Generative AI for NFTs

Knowledge sharing is important to us @ Ginkgo Analytics. Each Friday, we have a 12 minutes style exchange format: 12 minutes presentation/demonstration and 12 minutes of questions. This month, we had Alina to talk about Infrastructural Layers of Meta and Generative AI for NFTs:

In this article we will try to explain in simple terms what the metaverse is and the technologies behind the scenes. We will also try to go through buzzwords like DAOs, ICOs, smart contracts, NFTs, blockchain and more so after reading this article you have an idea about them. At the end we will show, how we used generative AI to create NFTs for Ginkgo Analytics.

What Is Metaverse?

The term “Metaverse” is not new. It has been first introduced in 1992 in the book “Snow Crash” by Neal Stephenson. In this book, people interact with each other through avatars in high-quality virtual real estate observing the word through virtual reality goggles. It is also not the first project involving interaction of humans in virtual reality: other projects entered the market earlier, e. g. “Second Life” from Linden Lab.

What is it that makes Metaverse so special and why millions of users and investors believe in the future of it? For me, the answer is simple: Metaverse is much more than a virtual exchange platform or virtual place to interact. Metaverse is a very smart combination of four technologies: gaming elements, mixed/virtual reality, Web 3.0 and blockchain and this is the combination that makes it so unique and valuable and promises a brilliant future. Let’s shortly go through those “building blocks” of Metaverse one by one:

Gaming and Virtual Reality building blocks are probably the easiest to grasp and self-explanatory. The point to add here is that the VR equipment is getting cheaper and cheaper each year making it more accessible to a broader range of consumers. At the moment of writing this article you can buy the VR glasses under 500 dollars. Furthermore, the technology is developing quickly: Teslasuit allows to capture full body motion. It is also important to keep in mind that those technologies might be used not only for online games but also for training purposes e. g. education of engineers and medics or for creation of more realistic meeting rooms for business purposes.

Web 3.0 is the easiest to understand by comparing it to Web 1.0 and Web 2.0. At the very beginning we had mostly static text in the internet and could only read it (information economy). Later, with Web 2.0, users were enabled to read and write, therefore creating content interactively (platform economy). Still, everything was hold on-premises and centrally controlled. In Web 3.0 you can read, write and own something in the internet (token economy). Virtual environments are not centrally controlled but rather governed by DAOs (Decentralized Autonomous Organizations) and based on blockchain technology and IPFS (InterPlanetary File System – distributed file storage)

Blockchain technology is basically a list of mutually dependent entries, which relate to each other through a factoring mechanism. It has four core characteristics: blockchain is disintermediated (you can carry out a transaction directly to another party), distributed, decentralized and trustless (no need of third party to verify the transaction). It allows faster transactions (no need for man in the middle), pseudonymization, additional level of security through cryptography/consensus mechanism and easy access from everywhere in the world. With the second generation of blockchain supporting token ids and metadata (e. g. URLs) smart contracts, scripting functionalities, ICOs (Initial Coin Offering), NFTs (Non-fungible tokens) and DeFis (decentralized finances) can be deployed on top

Infrastructural Layers of Meta

There is another approach one can take to decompose Metaverse into constituent components: infrastructural layers. So-called Protocol Layer is based on the second generation blockchain technology with Ethereum heavily dominating the market. On top is Token Layer: exchangeable (fungible) assets which are used by metaverse residents for payments. Further there is a Land Layer which is an already non-fungible (non-exchangeable) token which basically secures an ownership of virtual land in e. g. Sandbox. At the very top, there are other assets or what most of the people mean under NFTs.

What are NFTs and Why Are People Paying Insane Amount of Money for Them?

As we just discussed, NFTs are non-fungible assets which you can use in virtual land. They are cryptographically unique and are described in smart contracts we already mentioned. What is a smart contract? It is basically a piece of code to execute a contractual agreement. Solidity is the programming language used for Ethereum and it is Turing-complete: you can e. g. program your NFTs in such a way that you get some tokens every time the NFT is resold. If programmed and deployed, they will be autonomously executed. Smart Contracts are fully auditable and transparent although contract parties remain pseudo anonymous.

This “underwater layer” of smart contracts is what makes NFTs valuable. Their uniqueness and ownership are guaranteed through blockchain technology, they are very reliable because the information about NFTs is stored on blockchain and media on IPFS (storing information directly on blockchain is very expensive), the scarcity of NFTs is secured through auditability and transparency of smart contracts and blockchain technology makes counterfeit of NFTs very difficult.

What can be represented with NFTs? Actually, nearly everything! Of course, it is avatars and gaming items for virtual reality. You can also connect virtual land itself (Barbados was the first country to open an embassy in the Metaverse). Real-life moments might be represented as NFTs (famous example is the NBA Topshot). Real-world assets and domain names might be also linked to NFTs to add a level of security. Collectible items (avatars, profile pictures with CryptoKitties and CryptoPunks being famous examples) with provable scarcity are also a typical example of NFTs. Digital arts dominate the market of NFTs with e. g. famous example of Deeple “The First 5000 Days” ($69 million at the Christies Auction). NFTs can be items of virtual fashion (interestingly, luxury brands traditionally reluctant to digitalization were first to enter the metaverse). You can also represent everyday usecases as NFTs: licensing of software or events tickets are some examples

Why Netflix and Chill If You Can NFT and Chill? Generative AI in Action

In Ginkgo we used generative AI to create NFTs. For that a simple text-to-picture generator and picture-to-text controller working against each other were repetitively applied to the sentence and generated picture. The motto of the company “Ginkgo Analytics creates bespoke data-driven solutions to transform our clients into AI -powered enterprises” was used as a base sentence. You can see the process below in the video. Interestingly, final results resemble the wall of our Hamburger office as well as some dashboards and visualizations we deliver to a client. That was a lot of fun!

The second experiment which we carried out with the same settings was an attempt to create a NFT collection with generative AI. For that, a sentence “magic world in the mountains” was enhanced with random characteristics such as “child painting” or “depth of field”. You can see some of the results below generated by AI without any editing, nice, isn’t it?

On your way to become NFT millionaire with generative AI it is important to be aware of very high required calculational capacities: it might be difficult to generate even low-resolution images on your local machine. We would recommend deploying it directly in cloud environment.


I hope now you are familiar with Metaverse and its technologies and may be even motivated to create NFTs with AI (don’t forget to formulate the smart contract behind the way you are being paid each time your NFT is sold, cash multiple times, #SMILE)! We at Ginkgo love cutting edge-technologies and challenging problems. Feel free to connect!


Comments are closed.