Is there every such a thing as too much sampling to the point where we have so much data it messes up the resolution of our image?
lycorisradiatu
@ laurenz06 I had similar question as well. I did some research and it seems that oversampling could cause the decreasing in potential frame rates and signal-to-noise ratios.
keeratsingh2002
how does the choice of a reconstruction filter in the resampling process affects the fidelity of the texture when mapped onto a 3D model? I'm interested in understanding how the characteristics of the filter influence the prevention of aliasing artifacts in the final rendered image.
Mehvix
@keeratsingh2002 The filter's fourier representation characterizes how it'll effect the image it is convolved with (recall, conv in space is equivalent to multiply in fourier domain).
For example, consider a box filter -- the sharp edges result in ringing artifacts in the fourier domain, leading to ringing artifacts in the resulting image. Compare this with a Gaussian filter (which corresponds to a Gaus. in fourier space), which will more faithfully act as a low-pass due to the smoothness of the (Fourier domain) bell curve.
Is there every such a thing as too much sampling to the point where we have so much data it messes up the resolution of our image?
@ laurenz06 I had similar question as well. I did some research and it seems that oversampling could cause the decreasing in potential frame rates and signal-to-noise ratios.
how does the choice of a reconstruction filter in the resampling process affects the fidelity of the texture when mapped onto a 3D model? I'm interested in understanding how the characteristics of the filter influence the prevention of aliasing artifacts in the final rendered image.
@keeratsingh2002 The filter's fourier representation characterizes how it'll effect the image it is convolved with (recall, conv in space is equivalent to multiply in fourier domain).
For example, consider a box filter -- the sharp edges result in ringing artifacts in the fourier domain, leading to ringing artifacts in the resulting image. Compare this with a Gaussian filter (which corresponds to a Gaus. in fourier space), which will more faithfully act as a low-pass due to the smoothness of the (Fourier domain) bell curve.