Under dx12, you can call texture.Gather() to get the four texels used for bilinear filtering.
Or you can set texture sampler to linear to automatically do bilinear sampling.
Nowadays GPU supports hardware bilinear sampling, which increases the efficiency a lot compared to old days.
However, bilinear filtering could be too blurry because it doesn't take the tilt of the texture into account.
somaniarushi
I find fascinating the big picture idea of linear interpolation. The idea of using the limited data we have and making inferences for a query of a point using them is very compelling. Behind the layer of mathematics, it really is just estimating a color value for a given coordinate using the available data around it.
Under dx12, you can call texture.Gather() to get the four texels used for bilinear filtering.
Or you can set texture sampler to linear to automatically do bilinear sampling.
Nowadays GPU supports hardware bilinear sampling, which increases the efficiency a lot compared to old days.
However, bilinear filtering could be too blurry because it doesn't take the tilt of the texture into account.
I find fascinating the big picture idea of linear interpolation. The idea of using the limited data we have and making inferences for a query of a point using them is very compelling. Behind the layer of mathematics, it really is just estimating a color value for a given coordinate using the available data around it.