D
Deleted member 23558
God make my neurotransmitters great inc
- Joined
- Nov 3, 2022
- Posts
- 64,847
- Reputation
- 91,074
The mistake people make here is treating SMV like a pokemon game where you can just subtract and minus attributes. Using this method, this guy who has objectively 0 flaws
would mog him (hooked nose)
The real way to calculate would probably involve the principle of weights used in neural network. Each attribute (say height/ cheekbone width / PFL) would have some weight attatched to it wi(fxi) where wi is variable and f(x)i is the attribute. Run it in a computer with million faces and therell be an appropriate weight attatched to each attribute.
Here x1 and x2 can be height and pfl resp. Multiply the summation of those with a specific bias (or do it individually if u have enough computation power) and the output would be most acc. Of course you will also have to program that after a specific value (threshold) or even continous gradient the higher PSL you go the less things like phenotype, Height, frame etc matters but Itll probably can be made as self adjusting mechanism
would mog him (hooked nose)
The real way to calculate would probably involve the principle of weights used in neural network. Each attribute (say height/ cheekbone width / PFL) would have some weight attatched to it wi(fxi) where wi is variable and f(x)i is the attribute. Run it in a computer with million faces and therell be an appropriate weight attatched to each attribute.
Here x1 and x2 can be height and pfl resp. Multiply the summation of those with a specific bias (or do it individually if u have enough computation power) and the output would be most acc. Of course you will also have to program that after a specific value (threshold) or even continous gradient the higher PSL you go the less things like phenotype, Height, frame etc matters but Itll probably can be made as self adjusting mechanism
Last edited: