Another image: As far as smoothing errors: 1, then 2, then 3 would be the best, but if your system isnt properly synced, all will cause smoothing errors. The easiest and most effectient solution is to simply omit these details, that will rarely take up more than a pixel or two of space ingame.
The gradations in the normal map are the problem. You will see pixel artifacts and some warping in the specular. An additional fix to split UV and hard edges, is to add support edges much like you would with support lopps in SubD modeling to the problem areas. The the extra edges will reduce the length of the gradations.
The answer is that it depends. How far away will the object be? How many pixels of screen space is the edge going to take up? It is true that the stereo image can break the illusion of depth from a normal map but small edges don't produce a different enough image in each eye for it to matter.
At most, polygons should not be smaller than what a single pixel will be on your baked textures. It really depends on how dense your model is, sometimes 2 is enough, sometimes 5. And sometimes different parts need subdivided more, like a tiny screw versus a large smooth curve.
thanks guys for the crit. Add3r: Yea i know what you are talking about the blurring. I think its the compression to be honest because I submitted a 5k pixel image and then blogger re sized it. Do you have any tips on uploading a full rez image?
If you can repeat parts of the model, without being too obvious, then that's to your benefit since you can make the UVs larger and thus increase the number of pixels per meter. We have some info on our wiki about UVs. Have you seen it yet?
Your links are borked. I get the small pixels to. But, when I did see your 3d stuff, I think you should just sell your self as a concept artist. 3d intigration of 2d is cool, but your 3d props are not very stunning say a few.
Thanks guys! Damm crazy and bronto! Those are lookin real nice!!! Also I setup the posed render inside a realtime viewer. Its cool cause you can add the other pixel shader 2.0+ maps to it! woo Mirana Nightsilver Sentinel (click to view in 3D)
Not really related to the OP but Eric you might have some success feeding the Shader a fixed vector rather than camera position etc. And then rendering to texture. Failing that you can do the same sort of thing in substance designer if you're willing to dig into the pixel processor a bit
Yes, the height map will create the parallax effect and displace the pixels but the lighting is still calculated using the normal map. You need both. When you set Normal to true in the base material, it expects you to plug your own normal in the filter instead of generating it from the heightmap.