Key takeaway: filter out high frequencies (sharp edges), then sample (get pixel values). This results in a smoother/less jagged image.
wangjim19
This is really interesting. It seems like there is some kind of tradeoff between blurriness and jaggies. The pre-filtered version seems to be a bit more blurry. I'm wondering if there's a specific way to balance this tradeoff, and if it's possible to get the best of both (i.e. sharp edges of jaggies and non-jaggedness of pre-filtered).
BohanYu
Wangjim posted a really interesting question. Assuming we are sampling through human vision, I feel like an easy way to evaluate "how blurry is good" (beyond the antialiasing nyquist frequency) is by actually looking at it and see if we can visually distinguish the blurred area without much attention. As in this example, we can easily see the jaggies from the image on the left, but we can not easily see the blurriness of the second image, which appears to be much more "smooth" than the first image at a first glance.
Key takeaway: filter out high frequencies (sharp edges), then sample (get pixel values). This results in a smoother/less jagged image.
This is really interesting. It seems like there is some kind of tradeoff between blurriness and jaggies. The pre-filtered version seems to be a bit more blurry. I'm wondering if there's a specific way to balance this tradeoff, and if it's possible to get the best of both (i.e. sharp edges of jaggies and non-jaggedness of pre-filtered).
Wangjim posted a really interesting question. Assuming we are sampling through human vision, I feel like an easy way to evaluate "how blurry is good" (beyond the antialiasing nyquist frequency) is by actually looking at it and see if we can visually distinguish the blurred area without much attention. As in this example, we can easily see the jaggies from the image on the left, but we can not easily see the blurriness of the second image, which appears to be much more "smooth" than the first image at a first glance.