One interesting fact is that users do not need to use the camera's built-in firmware to demosaic, and many modern digital cameras can save images in a raw format allowing the user to demosaic them using software.
SainanChen
How do we linearly interpolate the color of the pixel where the pixels above and below it has value? Shall we only average the 2 pixels near it or 4 pixels including the two on its left and right that we calculated by interpolation?
Leo-godel
Nowadays, many modern cameras can use software to demosaic raw pictures, which is much more user-friendly to common people
hku8
how does the result of the demosaicking algorithm change based on the number of nearest neighbors we choose to interpolate? could it cause colors to blend together more, resulting in a less accurate representation?
briana-jin-zhang
@SainanChen https://en.wikipedia.org/wiki/Demosaicing#Simple_interpolation
According to wikipedia, your intuition is correct, it just averages the one above and below it.
mhallac
What do the images look like if we don't demosaic? From a distance with enough pixels and a high resolution, or if the pixels are small enough, I would expect an image to look the same as if it were demosaicked, especially since computer screens work right now with separate RGB pixels.
daniswords
I think the outcome of not demosaicking depends on the camera. I was reading this paper https://www.ece.lsu.edu/ipl/papers/IEEE_SPM2005.pdf , and in the end they talked about how some modern cameras that are able to capture RGB information at every pixel don't need to demosaic in the digital camera pipeline
One interesting fact is that users do not need to use the camera's built-in firmware to demosaic, and many modern digital cameras can save images in a raw format allowing the user to demosaic them using software.
How do we linearly interpolate the color of the pixel where the pixels above and below it has value? Shall we only average the 2 pixels near it or 4 pixels including the two on its left and right that we calculated by interpolation?
Nowadays, many modern cameras can use software to demosaic raw pictures, which is much more user-friendly to common people
how does the result of the demosaicking algorithm change based on the number of nearest neighbors we choose to interpolate? could it cause colors to blend together more, resulting in a less accurate representation?
@SainanChen https://en.wikipedia.org/wiki/Demosaicing#Simple_interpolation According to wikipedia, your intuition is correct, it just averages the one above and below it.
What do the images look like if we don't demosaic? From a distance with enough pixels and a high resolution, or if the pixels are small enough, I would expect an image to look the same as if it were demosaicked, especially since computer screens work right now with separate RGB pixels.
I think the outcome of not demosaicking depends on the camera. I was reading this paper https://www.ece.lsu.edu/ipl/papers/IEEE_SPM2005.pdf , and in the end they talked about how some modern cameras that are able to capture RGB information at every pixel don't need to demosaic in the digital camera pipeline