What is the relationship between APCA and phi?

[cgiosy]

Hello.
While playing around with APCA, I discovered that (bgY^0.618 - txtY^0.618) * 1.14 ± 0.027 is somewhat similar to APCA's Lc. Of course, the error is not negligible, and especially as Y approaches 0 or 1, the error with Lc increases to about 3-5. Since it lacks APCA's various core and valuable processing steps, this is an expected result.

Anyway, considering that, unlike APCA, this is effectively a single-variable function, it seems to be a relatively optimal approximation. However, it happens that 0.618 is close to 1/phi, and considering the existence of DeltaPhiStar, I started to wonder if there's a possibility that APCA might also be based on phi.

So, what I'm curious about is whether this is merely a coincidence, or if the appearance of phi is actually somewhat expected. Alternatively, are there any explanations or links I could refer to about how DeltaPhiStar derived this magic constant to predict approximate color contrast?

Hmm, I'm running a few more tests, and it seems more likely that it's just a coincidence due to its similarity to existing constants. The optimal value changes quite significantly with minor modifications like adding multiplication, changing the criterion from the sum of squares to something else, or applying weights.

[Myndex]

Hi @cgiosy

APCA is not based on phi, but you will notice that values in the midrange may seem to have a relationship. This is because phi in association with certain other parameters can make a close approximation to visual lightness (perception) curves.