🎲Generative Art Process

The Process for Selecting 1,500 Unique Ninjas

Creating exceptional generative art requires more than merely configuring layers and initiating the software. It demands substantial thought, not only in the initial planning stages but also in the meticulous customization of the rules and algorithms that shape the art.

With Ninjas, we aimed to craft designs that are both clean and minimalistic, yet distinctly unique. This approach ensures that each Ninja serves as a durable profile picture or brand identity. We wanted to make each Ninja PFP unmistakably recognizable, preventing any confusion between different Ninja PFPs as representing the same individual. The ideal scenario is where a single glance at a Ninja's PFP leaves a lasting impression of its uniqueness, while still maintaining a sense of unity, portraying all Ninjas as part of a singular, cohesive community and collection.

Artist Input

Eight artists and PFP enthusiasts helped create over 2971 hand curated Ninjas

To optimize the likelihood of blending traits in the most aesthetically pleasing manner, ensuring that even floor Ninjas appear extraordinary, we enlisted a group of eight artists. Their task was to meticulously select and amalgamate traits for our 300 Custom Ninja Presale. Throughout this phase, each artist crafted four bespoke Ninja options, tailored to the preferences of the presale winners.

This initiative also provided us with invaluable market insights, helping us to understand trait preferences and our audience better. Artists were also encouraged to go wild with the paint bucket tool and Photoshop, allowing us to generate several new color combinations of various Ninjas traits. In the course of this endeavor, our team curated close to 3,000 custom Ninjas. These custom Ninjas formed the starting point for our generative art algorithm, which we used to make sure no stone was left unturned.

Market Research Priorities

Our market research was conducted via our 300 Custom Ninja Presale, emphasizing the gathering of preferences in categories such as Colors, Types (including Classic, Monomi, Frog, Cat), Masculine vs Feminine, Hats, among others. From a statistical standpoint, a sample size of 306 individuals can accurately reflect the preferences of a larger group of 1,500 people, with a 95% confidence level and a margin of error of 5%. Although our presale sample wasn't perfectly random, it was as close as possible to this ideal. This approach is highly likely to enhance the probability that the traits we choose for our collection will align with our audience's tastes, thereby minimizing the risk of an oversupply of less desired traits.

Generative Art Algorithm Priorities

1. Make Ninjas as CLEAN as possible

   1. Max out minimal trait combinations

   2. Max out minimal color combinations

2. Make each Ninja distinctly unique to ensure durability as a personal or corporate brand

   • Three degrees of separation rule (All Ninjas must have minimum of three traits that are different from one another)

3. Make floor Ninjas extraordinary

   1. Incorporate as many artist customs as possible while achieving the first two priorities

   2. Leverage market research to ensure our distribution meets our audience's expectations

Make Ninjas as CLEAN as possible

Here we start with the first gen, then use that as the input for the 2nd gen, maintaining three degrees of separation between all Ninjas. The 2nd gen is the input for the 3rd gen, and so on.

Gen 1

• First gen removes back accessories, shinobi stripes and belts, limits hands weapon to thumbs up, and hats

  • Requires the head, men-yoroi, body color to be the same color (hand can be any color)

    1. This maxes out the 1 and 2 color Ninjas (second color coming from the face)

    2. Two degrees of separation between each minimal Ninja is sufficient

Gen 2

• Add hat accessories, allow 1 additional color across head, body, or men-yoroi

  • This maxes out 3 color Ninjas with hats

Gen 3

• Allow all traits and see what we can get with 1-2 color limitation (aside from face)

Putting It All Together: Creating the Final Gen

1. Start with all Ninjas chosen by Presale Winners, these are locked in

2. Add all Ninjas from Gen 1, removing any collisions based on degrees of separation rule

3. Add all Ninjas in the remaining set of 3,000 customs, removing any collisions

4. Add Ninjas from Gen 2 and then Gen 3 until we have 1,500 unique Ninjas

5. Compare the distribution of our final set to the preferences collected in market research, tweak as necessary