This topic is about something that has been nagging me for a long time. I tried to discuss it once on the Openarena forums, but got the usual kind replies from leilol

In a nutshell, I believe that too much detail in an FPS game actually hinders your play. So the current goal of game engine and video card developers, i.e. cramming more and more detail (higher texture resolution, intricate shadows, etc.) on the screen is pointless as far as FPS games are concerned. (Unless display technology is radically changed in the near future and we can have a nice holographic 3D display for next Christmas...

I base the above conclusion on my experience with modern games (starting with Q4 and UT2003). If I set the screen to high detail I can hardly distinguish my opponents from the background. The more detail I have, the more difficult it becomes. In Q3A I already noticed this problem, and UT2003 was for me unplayable exactly because of this phenomenon. OTOH the older UT99 with standard default settings had a good balance of detail so I hardly had this kind of perception problem there. It is not a matter of low FPS caused by extra CPU/GPU load. I have at present an Athlon XP64 dual core machine with an Nvidia 7300 GT video card. I use it at 1280x1024 resolution. With other settings at default my FPS in Q3A almost never goes below 166. This should be more than sufficient for playing.
I think that the root of the problem is the lack of 3D info in the rendered image. Consider how in real life we perceive different objects in 3D. First there is the shift of the object (relative to the background) in the left and right images that gives a sense of 'depth'. Also, when you look at an object standing at some distance in front of a background, the latter will be somewhat blurred because the point of focus is at the object. Both these 3D cues are missing from the images rendered on a flat 2D screen. For this reason player models kind of 'blend' into the background unless they are very different in color or brightness. And the more the detail, the more difficult it is to see the outlines.
This explains why some of the advanced players use extremely low detail settings. Setting a lower level of background detail provides half of the 3D cues (i.e. the amount of detail in the object and the background behind is now different and approximates what you see in real life). The problem is that only slightly lowering the detail makes you dizzy because your eyes continuously try to focus on the background -- but of course they cannot because the background image is not sharp to begin with. (Try to play for a few minutes using low texture detail setting in a map that has lots of brick textures. You will see what I mean.) So the solution (for Q3A and derivatives) is setting 'r_picmip' to 4 or higher and forcing a nicely visible model (no bright skin needed). The textures look mostly homogeneous (and ugly, but who cares

IMO it would be possible to (partially) eliminate the problem by careful selection of simpler textures and/or simplifying geometry when creating maps. However, this goes against the current trend of making maps with as much detail as the engine can handle.
I would like to have your opinion. Do you also have this problem? What other solutions are possible (besides drastically lowering texture details)?