AI: stands for Artificial Intelligence, demonstrated and built by machines, different from the human one
Metadata: the simplest way to represent NFTs by highlighting their own main characteristics
Rarity: NFT’s collections have different type of rarities within their single items. The more a NFT is rare, the higher (usually) is the value of itself
AI will help us create the very foundations of our NFTs (or at least the foundations for those ones unable to draw as me).

 There are many different types of AI and I’m definitely not the person who can go in details about that but, for the sake of our needs, we will focus on just one: prompt-to-image AI.
What does it mean? It’s simple:
You think about something by leveraging the most powerful tool you have —> you brain
You write this somewhere on a laptop
A software will do its job
Will give back to you an image (or a set of images) representing your text
Stop. That’s it. Easy, uh?
Explaining this a bit better, an AI model that is prompt-to-speech is a model that is trained by natural language, where developers make sure the model is pre-trained by some words and that then it continuously retrains everytime users use it (by asking if the information provided is a good one or a bad one). By introducing a neural network behind, the system make sure to generate images that represents your prompt.
What is fundamental for our use case then?
Having great creativity
Being as specific as possible
Waiting few seconds before the machine give you the information
So far, there are many different tools we can leverage to do this, and all of them allow us to create images that can be used even for commercial rights! This is wildly important when it comes to Intellectual Property, as it allows us to use them everywhere. There will be plenty of time to talk about IP, in the meantime look at this image created by A16Z about this environment and read their amazing new categorization of this important topic.
So getting back to AI Tools, the ones I’ve tested (and they work pretty well) are:
You can try all of them (just click on the link posted above), but remember that MidJourney is currently running on Discord.
For our use-case, I’ll use DreamStudio because:
It’s very easy to log into it
It’s the best one for creating art-based AI
It helps you create more than 1 image ad a time
Last but not least, it’s from Stable Diffusion, on of the first company into the prompt-to-image environment
Ready? Let’s try it!
First of all, go to https://stability.ai/ and go try the Beta Access of Dream Studio.
You can try to log in anyway you want, I do think that Google is the best fit!
Once logged, start dreaming! There are few things you can keep in mind:
In the box in the bottom, you can write whatever you want! I suggest you to give the input “by [name of the artist]” to make the image looks as good as possible
You can adjust the height or the width of your images, making them bigger or smaller! For our purpose, let’s stay with 512x512 (minimum settings)
By increasing the CFG scale, you’ll make the image you want closer and closer to your inputs
The more STEPS you add, the more layers and details your image will have
Increasing the Number of Images, well… produce more images! (9 tops)
Now you can start playing with your AI to create images! It will be very fun, and I suggest you to shape your creations more and more by testing all the possible options at the beginning!
If you want to create an NFT collection, you have to keep in mind that:
NFT collections have limited supply
NFT collections have metadata
NFT collections have rarity
Once we have decided to create our images, we will focus on this!

 Now, for the next 24 hours, go there and start experimenting!
Let’s say we want to create now a collection of 99 NFTs divided in:
72 Common images (9 images x 8 iterations)
18 Rare Images (9 images x 2 iterations)
9 Epic images (9 images)
Going back to our tool, let’s use a bit of imagination and start creating our images! My suggestion is to start from the common images and give an input that contains what are called “traits”. These can be whatever, but say that we use the following:
a background
an element of distinction
an full image
an art style
You can do, as an example, something like this:
By approaching the creation in this way, you can repeat over and over some information to make “common” images aka those ones that have some common and repeated traits.
These will be fundamental for us later on.
So, if you have understood the assignment, you can start creating the overall collection by repeating this over and over and creating your Epic and Rare images.
IMPORTANT: for the rare ones, instead of creating 9 images all together, try to create 1 at the time. We are looking to create our 1/1, as called in the NFT term.
Have fun, and we’ll catch up tomorrow to start experiencing how to mint our collection!
LFGGGGGG 🚀🚀🚀🚀🚀