The whole point of the multi-pass renderer is to do AA through multi-pass. Every new pass will jitter the camera position a little bit inside each pixel thus producing AA.
The AA you get with multi-pass is the best you can get. It will AA everything including motion blur, soft shadows, DOF, etc. This is non-adaptive AA though which means that every single pixel gets AA the same as every other pixels even when there is no need for it. The number of passes you select with multi-pass determines the number of samples in the AA. Simple as that.
The non-multi-pass renderer does an adaptive AA which means that AA is computed only where it is needed. The non-multi-pass renderer uses some heuristics to determine where AA is needed. In computer science, heuristics are heuristics which means that they are not always right on spot so there are particular instances where it will not AA properly. This is the main reason why multi-pass was implemented. That is to take care of those situations.
So right now, you have the choice between which type of AA you want to use depending on your particular project. You can try them both and decide.
Using multi-pass with AA does not make any sense. It would be a waste of time and CPU power. At 9 passes, the AA is pretty good. Adding AA over that would mean that some pixels would get 144 samples for AA and the worst case would get 576 samples for AA which is totally pointless.
As for comparing different outputs. To me, it is only relevent for someone who is looking for a very specific type of look in the render. If I where looking at some other artist render and whishing to get that render in A:M then maybe I would start comparing. This is not my case. I’m not trying to reproduce the LW look or the C4D look or such artist look who happens to use LW or C4D or whatever like that. I try to develop my own look and this look is definitely not photorealism. And A:M have everything I might need for that. I remember, a few years ago, a A:M user had posted some stills and animations with a very specific, stylised and beautifull look. It was all black and white with extremely harsh light and shadows. I really liked that. Another example would be the work of Sinj or Jimmy Maidens with his “Boring 3D” work. But I’m not trying to replicate their look. I am trying to develop my own stylized look which might explain why, to me, this question is kind of moot.
Yves Poissant


