Interesting, thanks. Curious, what Lumasharpen value did you settle on?
Well, that's a bit tricky to answer with any AA that has a temporal element applied in motion, I think.
It's also going to vary on the native resolution you're running the game in, if you're applying any sort of downsampling (which will obviously need quite a bit more sharpening, in most cases), how large your screen is, how well the scaling is, what your refresh rate is, and how far or close you sit from it.
From now on, my sharpening numbers will only apply to a native 1080p, 60Hz display with no downsampling applied.
For FXAA, I've found the default value of 0.65 matches the sharpness of the original image without AA. TXAA is much blurrier, and in static images, I've found that a value of 1.00 matches the sharpness of the original image without AA. But again, this is where it becomes more subjective. Seeing as any game with temporal AA is its blurriest in motion, static images are literally only part of the picture. Then again, temporal blurring could be considered another form of motion blur, and thus be irrelevant to resharpening the image anyway; you're mostly going to notice the blur during static or low motion scenes, at least at 60Hz.
For TAA, I found that, like FXAA, at least in static scenes, the default amount of 0.65 is enough to match the original image without AA. Comparison below (note the ground textures, particularly):
http://screenshotcomparison.com/comparison/151834
In my opinion, you really don't need much. And when people go on and on about how much TAA or TXAA or even FXAA blur the image, I don't think they realize how little they do blur, when compared to the original image without AA applied.
That said, what's great about PC gaming is choice. If you want more sharpness, go for it, if you want less, bring it down. Up to you really
