You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
At least in my experience, the output of a curve function is very sensitive to the absolute value of its derivative, not by the continuity of its derivative.
If a curve is too slanted, thus separating appart two near RGB values, artifacts will start to form, since compressed subtle noise will become more evident due to this separation.
So if you want to, say, white-balance an extremely tinted input, you want to achieve the right blacks, greys and whites. But when you force them to be balanced, the curve might be too slanted if you have different strong tints at darks and brights, for example. A linear interpolation in this case is the best option, since it's the interpolation that minimizes the Lipschitz constant of the function.
The text was updated successfully, but these errors were encountered:
Hello, there.
At least in my experience, the output of a curve function is very sensitive to the absolute value of its derivative, not by the continuity of its derivative.
If a curve is too slanted, thus separating appart two near RGB values, artifacts will start to form, since compressed subtle noise will become more evident due to this separation.
So if you want to, say, white-balance an extremely tinted input, you want to achieve the right blacks, greys and whites. But when you force them to be balanced, the curve might be too slanted if you have different strong tints at darks and brights, for example. A linear interpolation in this case is the best option, since it's the interpolation that minimizes the Lipschitz constant of the function.
The text was updated successfully, but these errors were encountered: