Trait Weight Calculation
The rarity of tethered NFT assets in the metaverse is decided by numerous different traits; including colours, no. of accessories, types of accessories etc. We have established a new trait weighing system to randomize the traits.
When generating character and vehicle NFTs, traits are randomised but based on a weight system. Larger stakes will have a higher chance of acquiring rarer traits over common traits. Traits include things such as skin type, helmet, backpack, tattoos etc.
Letβs use skin traits as an example. In the below table, there are a total of 7 traits; these have been categorized into a tier based system (seen in the table). Traits are ranked based on their nominal weight and each weight has a corresponding nominal probability.
This method for calculating trait weights provides a simple and easy way to determine how rare is a trait compared to others. Rather than taking the rarest trait of each NFT or averaging their rarity, our algorithm calculates the trait weights.
If we take π€π€0ππ to be the nominal weight of ππ-th trait (ππ=1,2,β¦,7), then that corresponds to the case when the playerβs stake is equal to the average stake. Then we have Table 1 (above) of nominal weights π€π€0ππ (and nominal probabilities ππ0ππ) for the trait skin:
To further refine calculations we introduce another parameter β x β to capture and convey the value of both the playerβs stake and the average stake of players. If the playerβs stake is less than the average stake (π₯π₯<1), then most likely ordinary traits will be generated. If the playerβs stake is greater than the average stake (π₯π₯>1), then mostly rare traits will be generated.
Weights Traits Formula
The task is to find the functions π€π€ππ = ππ(π€π€0ππ, π₯π₯) and ππππ = ππ(π€π€0, π₯π₯), where π€π€0 = {π€π€01, π€π€02, β¦ , π€π€07} and
We are then presented with the following formula to calculate the trait weights:
where ππ > 0 and ππ > 0 are scaling parameters, πΌπΌππ and π½π½ππ are the same values from the Table 2. Using (7) for ππ = 4 and ππ = 0.5, we can build the plot Fig. 3. The behavior of the curves in Fig. 3 is similar to what was observed in Fig. 1 and 2. If we choose ππ = 2 and ππ = 0.25 we will get the plot shown in Fig. 4
Trait Weight calculations produce results that give enough emphasis to single rare traits and also include overall trait rarities in its calculation. And most importantly the results it gives match better with our human expectations.
While the calculation methods are the most core part of calculating the trait distribution, there are still many additional elements that are used to get the best final result.
Last updated