Why is the Antialiasing Image sampler not a Post-Process?

[secondplace]

I've been exploring ways to Antialiasing my images better. I also have access to Nuke and played around with my renders.
It made me wonder why the 'Image Sampler' is not a Post-Process. Is the 'Image Sampler' getting information directly from the 3D information and not only filtering the resulting pixels?

[Muhammed_Hamed]

It cannot be a Post-process.. it doesn't act on the resulting pixels from what I understand
For me personally I turn off Filtering for all my maps in Vray/Maya, and I override texture filter with "Nearest" for bump and normal maps to preserve the high frequency details in my maps
This makes it harder on AA, and rendering is a bit slower .. but it is worth in my view for the visual difference

In Vray 5 there is a Sharp Isotropic filter, that does a better job than the old filters. So I will not need to turn off filtering/blurring on my images. I've used it in Max, and it works nicely

[Joelaff]

AA at a basic level works by oversampling the image. It is like a taking a higher resolution photo of the scene where every pixel represents only ONE point in 3d space (like jagged edges of objects where the pixel is either FG or it's BG) and scaling it down to smooth things out.

There is a technique where you can blur edges of existing images to attempt to get a poor man's antialiasing. It doesn't work so well be today's standards. There was even a commercial product in the early 90s that would anti-alias already rendered footage. It was faster than rendering at higher quality settings, but it did not look as good, even then (anti-aliasing back then was poor, often due to lack of memory of CPU power). This was called JAG (Jaggies Are Gone), and it was made by Ray Dream, the same people who made the eponymous 3d app.