Will using a min or average potentially introduce more aliasing in areas that are further away? I imagine in cases where the two L candidates are very different, min() and avg() would lead to higher mipmap levels that might cause more aliasing.
GarciaEricS
At first I was very confused with this slide. I didn't understand each expression in the max below, and I didn't understand why the max was there at all. Know I understand that each individual term, the square root, and everything under it, corresponds to a distance. It is just the Pythagorean theorem telling you the distance from the bottom left point to the bottom right point on the left expression, and the distance from the bottom left point to the top left point in the right expression.
These two distances could be very different, so what mipmap level should we go to? This algorithm is conservative in that it will select mipmap level based on the direction with the largest distance, ensuring there is little aliasing in that region. If it were to select the min for example, we could have a lot of aliasing because our sampling rate in the texture frequency could be much slower than the signal frequency (our texture resolution) in the direction that wasn't the min. Using max ensures this does not happen, and our sample rate will be always at least the signal frequency in all directions, meaning aliasing is greatly reduced.
andrewn3672
I originally didn't understand the purpose of mip-mapping, but after looking more into it and completing the assignments, I find it to be very smart and intuitive. We essentially just to make sure we are mapping the texture with the correct ratio.
Will using a min or average potentially introduce more aliasing in areas that are further away? I imagine in cases where the two L candidates are very different, min() and avg() would lead to higher mipmap levels that might cause more aliasing.
At first I was very confused with this slide. I didn't understand each expression in the max below, and I didn't understand why the max was there at all. Know I understand that each individual term, the square root, and everything under it, corresponds to a distance. It is just the Pythagorean theorem telling you the distance from the bottom left point to the bottom right point on the left expression, and the distance from the bottom left point to the top left point in the right expression.
These two distances could be very different, so what mipmap level should we go to? This algorithm is conservative in that it will select mipmap level based on the direction with the largest distance, ensuring there is little aliasing in that region. If it were to select the min for example, we could have a lot of aliasing because our sampling rate in the texture frequency could be much slower than the signal frequency (our texture resolution) in the direction that wasn't the min. Using max ensures this does not happen, and our sample rate will be always at least the signal frequency in all directions, meaning aliasing is greatly reduced.
I originally didn't understand the purpose of mip-mapping, but after looking more into it and completing the assignments, I find it to be very smart and intuitive. We essentially just to make sure we are mapping the texture with the correct ratio.