I see nothing wrong. You will always have antialiasing one way or the other, especially if you scale stuff. It's inherent in the process. You simply have a wrong understanding of how this works and what the differences between AI and AE are. Sorry to be so blunt, but it really comes down to that - you have designed your artwork without giving any consideration to these matters beforehand. You could try to set the layer quality switch to draft, but if your lines are not perfectly aligned with full pixel increments, they will still look "pixelated", just in a different way - more like game artwork. Again, you are expecting AE something to do that goes against how it works. Even if you get it right for one frame, as soon as things move you will be back to interpolation in the intermediate steps. That aside I don't know why anyone would design video content this way - it's just gonna look awful because the lines will look flimsy and flickery when they move because they are so thin.
With images this large and strokes this small you're going to end up with some very thin lines when you turn them to pixels. Especially when using vector images it's important that you work at something close to your composition size so that you can visualize how these lines will look in the completed artwork.
Simply put, if you have a stroke that is 20 points in an vector drawing the line will be 20 pixels wide when it is converted to pixels. Scale down that line to 10% and the line is now 2 pixels wide. Scale it to 5% and it is one pixel wide. Now shift the line 1/2 pixel to the left or right so the color values no longer line up with the pixel grid and your single pixel black line now turns into a 2 pixel wide gray line because of sub sampling.
Your first example shows shows that your lines are 12 pixels wide and not precisely lined up with the pixel grid:
You can count 13 pixels in the above screenshot but the left and right edge are the same color so if the image was shifted 1/2 pixel to the left or right every pixel would be black. That's why I believe that the stroke on this line is 12 points or 12 pixels.
If you created the first example in illustrator and exported the png file from Illustrator then you don't have your lines set up to snap to the pixel grid in AI either.
When this line is scaled down in your third example it is only 2 pixels wide and also not perfectly aligned with the pixel grid:
I'm guessing that the lines are 2 pixels wide because you actually have 3 pixels shown but one is light gray and the other is dark gray. If you precisely lined up the lines with the pixel grid you might be able to get a line that is two pixels wide until it starts going around a corner or isn't perfectly vertical or horizontal.
Unless your vector artwork is perfectly aligned with the pixel grid and every line is perfectly is exactly a whole number of pixels wide and perfectly horizontal or vertical the edges are going to be interpreted and not perfectly represent the designed colors of your artwork.
Even your "perfect example is not perfect. It only has vertical lines but the edges are not black and it's showing 14 pixels when I believe that the stroke is actually 12 pixels:
If you make any of these lines move across the screen, especially the very thin ones, you're going to have constantly shifting edge colors as the artwork lines up with the pixel grid.
The other thing you are missing is that movies move and illustrations don't. You cannot achieve the same level of detail in a moving image that you have in an illustration and you shouldn't try. The final judgement of how your project looks can only be made when you are watching the movie at it's designed frame rate at full size on the screen you designed the project for. IOW, if the movie is destined for YouTube HD at 1920 X 1080 24fps then you have to judge the final product by watching it on a computer monitor that's 1920 X 1080 that has a normal screen refresh rate. Looking at your final product at any other size or with any other settings will give you interpolated pixels.
Just for my own understanding, I believe PP will rasterize any vector graphics upon import, or at least upon addition to a sequence. Does the same hold true for AE? Are vectors rasterized when added to a comp, or can they remain vectors up until export?
Thanks, Todd. Good info.