What goes into creating collectible NFT sets like Bored Apes?

Call them what you will — combinatorial art, a "random item lottery,” or a bubble poised to burst. 

Collectible sets of character-based NFTs, a genre exemplified by the Bored Ape Yacht Club — a set of 10,000 images of disinterested cartoon apes, each with a unique set of characteristics — have exploded in popularity this year. The Bored Apes themselves have seen nearly $500 million in sales since April when they were first minted.

With NFT sales reaching historic highs in recent months, many other projects have cropped up with a similar mix of hand-drawn art and computerized randomization to try to cash in on the boom. 

From the outside, the newer projects may seem like copycat cash grabs — and some of them may be. But high-quality combinatorial NFT collectible sets aren’t easy to create, and making them requires both technical and artistic skill. 

The Block spoke with the creators of three generative NFT collections to learn about the process — which typically involves drawing the images, creating the randomization algorithm, drawing up the smart contract for minting and refining the artwork into thousands of unique pieces. 

Creature World

The kind of digital art featured in these collectible sets requires the creation of thousands of layers. The artist makes these layers by hand, usually through a digital art program such as Apple’s Procreate on the iPad or Adobe’s Photoshop. 

Danny Cole prefers the iPad. He is a digital artist and the founder of Creature World NFTs, a project that like Bored Apes is composed of 10,000 images of the same basic character: in this case, a bald man with cartoonishly round eyes and broad nose.

The Creature World NFTs were minted on  August 28, 2021. They have already amassed $80 million in sales, according to the NFT sales tracker CryptoSlam. 

Cole first came up with the creature figure as a child, often experimenting with the man’s features using  paint on canvas. It wasn’t until a friend showed him some randomly generated NFT projects in September of 2020 that Cole thought he could put his own beloved creatures on the blockchain. 

“He was like, ‘I think [generative art projects] would lend itself really well to the creatures. And I think that you would be able to design something really, really crazy,” Cole recalls his friend saying. 

Except Cole had a problem: he knew how to make art, but he didn’t have technical chops to algorithmically randomize the images and create thousands of unique final pieces.  

So Cole asked Kevin Sims, a friend from high school, to help with the computer legwork. “Kevin said, ‘If you designed a bunch of artwork, I could build you an algorithm. I also know how to write smart contracts to turn these into NFTs.” 

The Creature World algorithm required instructions for how to retrieve artwork and know how many images to make. Then it had to create 10,000 unique JPEG images without any repeats. 

The algorithm also included a rarity scale, which allowed some features to be less common than others. For example, creatures with an open-mouthed “happy” smile constitute 95% of the collection, whereas the remaining five percent feature a closed mouth or smirk. “Hiding” creatures, which blend into their surroundings, comprise 0.17%. There is also a range of different color themes, from rainbow, which is extremely rare, to the more common gold. 

However, Cole stresses that Creature World is more than its 10,000 collection or the rarer, valuable items within it. “A rare creature to me is one that I look at and say, ‘that is so me,’” he says. “That is not something numbers can describe.”

Sad Girls Bar

Sad Girls Bar also features 10,000 images, which feature somber women in various outfits, hairstyles, emotional states, and items in their upraised hand. 

Glam Beckett, the artist behind the collection, had been drawing images of sad girls for eight years before deciding to launch an NFT collection in August.

“Back then, there were very little female-based NFT projects,” Beckett, who hails from Russia, said with the help of a translator. “That's why we decided, ’Well, let's make it happen like this with these sad girls.’”

Beckett started drawing the art for Sad Girls Bar in July, using Procreate. Creating the art for an NFT project didn’t really differ from the digital art she had created before, she says. The process involves turning every single element, such as a hairstyle or piece of jewelry, into its own layer. Each image has about 15 layers, and it took Beckett a month to create all of them. 

Sad Girl Bar’s developer Andrew then built the algorithm on Python, which took about two weeks, and the NFT smart contract for minting the NFTs in a program called Solidity, which took about a week, Andrew told The Block through a translator.

However, the Sad Girls Bar team found that the workflow was more complicated than just drawing the images, writing the algorithm, and then minting. The most challenging part was making sure that certain elements did not appear together in the same image. For instance, Beckett notes that Sad Girls with bunny hats could not also have any hairstyles, glasses or masks. 

This meant that the team had to redo some art and add exclusions, or commands to keep certain traits separate, into the algorithm.

Every new exception slowed down the algorithm, requiring Andrew to further tweak the coding so that the algorithm could work quickly enough before it came time to mint. 

The team reviewed all 10,000 images in the collection to ensure the exclusions worked and all the traits combined as intended. After the final check was done, the team put their NFTs up for mint on August 23. Buyers paid 0.03 ETH to own a Sad Girls Bar NFT. 

“​​I noticed that a lot of girls were buying it as a gift for their friends or for their girl friends to get them into the NFTs,” Beckett said. 

Still, it took a month to sell all the images. The project finally ended its mint on September 26, 2021. 

FoxyFam

Like Creature World and Sad Girls Bar, the team behind an NFT project FoxyFam, a collection of cartoon foxes, also set out to create 10,000 NFTs. That didn’t end up going according to plan, though. 

The team’s artist, who goes by JetPunch, made the art in five days using Photoshop and Clip Studio Paint, a tool geared toward creating manga and comics.

Ncover, FoxyFam’s pseudonymous developer, then made the algorithm using Visual Studio Code using NodeJS as the main programming language. To create rarity, he had to program a "weight" for each layer that determined the probability that the algorithm would choose that element. 

For example, if Hat One has a weight of one, and Hat Two has a weight of 30, the chances of Hat Two appearing is 30 times greater than Hat One — which would be the rarest. 

Like the Sad Girls Bar team, the FoxyFam team had to revise the art and algorithm to prevent certain elements from clashing, such as rare parrots sitting on the shoulder of Foxies with masks.

Then, Ethereum’s high fees caused the team to re-evaluate the scale of the project, explains FoxyFam’s founder Mmwalk. “Few projects were selling out at that point, and we saw it as an opportunity to scale back, limit supply, and begin starting on the project that our owners were excited about,” he adds. 

Thus, the team decided to mint 3,333 NFTs instead of 10,000. “Looking back on it, we would have scaled back less, but it doesn’t matter because that was the momentum we needed,” says Mmwalk. 

Having fewer NFTs makes the existing ones more scarce and thus more valuable. And since the team is building a Temple Run-esque running game that will feature the FoxyFam NFTs, the beta testing went much faster with a smaller group of participants. 

“It's easier to make moves — and make them quick,” says Mmwalk.