Example, look at aliasing and sampling artifacts on the utility poles for example. They appear softer and not "pixel perfect".
Sure, I specifically turned off AA to take those shots to emphasizes that effect. The fence sections provide even better examples, as sections that look like random pixels floating in space in the native image look a whole lot more like actual fence in the downsampled one.
Anyway, had I used AA, the native image wouldn't have those "solid black vertical lines with a distinct break between object and sky" either, and rather both would show what one might describe as "fuzzy unclear intermediate lines and blotches that are the averages between the pole and sky", though "fuzzy unclear" and "blotches" aren't the terms I'd choose in either case. So, am I to take it you are of the opinion that rendered images look better without AA?
This is just a small isolated example; the effect is present throughout the image and contributes to an overall softer fuzzier picture when full screen and moving as the averages change.
The texture detail is obviously softer/fuzzier on parts of on the native image, the cracks in the ground to the left and right of the gun being the most notable examples. I specifically turned off AF to emphasis this difference, though it will still show to a lesser extent with AF on.
And you used software to resize the image offline with advanced CPU intensive sampling and filtering. in non real time anyway. That operation isn't as good in real time by display processors and scalers.
I used standard bicubic resampling in Photoshop, which provides comparable results to the scaling I've seen on plenty of displays, including my own plasma.
A static image isn't a big deal, but when its in motion on a full screen you can tell that there is a resolution mismatch.
Yeah, I often play games at higher than native resolution on my plasma, and I can tell there is "a resolution mismatch" because it looks much better when running at higher than native resolution. This hold particularly true in motion, as the downsampling helps smooth out the aliasing which looks especially nasty as it crawls across the screen when the view moves.
For the second one, did you just take a 1080p screenshot and then resize/resample it in Photoshop? Or did you actually have it render in 1080p and make the game downsample it? Is there a difference.
I used Photoshop since CSS doesn't have the option to set the rendering resolution separate from the display resolution, and because it really doesn't make a notable difference. If you'd like to see for yourself, I recommend ARMA II, as it does allow changing the rendering resolution separate from the display resolution, and there are
demos available.