The problem is, FSX is over a decade old. For instance, would you compare the performance of a decade old computer (or even graphics card) to the performance of the today's latest tech? Most people would say of course not, yet, after years of FSX dominating the market, many users still seem to work with an assumption that sim is still near the pinnacle of what's possible.
I dont think thats the case at all, but only time will tell.
FSX is finally over IMO, as well as FSX SE (and FSW in the row).
P3Dv4 opened a huge breach by dramatically increasing the sim performances and bringing the long awaited 64 bits architecture, finally allowing editors to take full advantage of their raw material to propose real dense sceneries without collapsing framerates.
As an example, with my current PC I used to hardly get a stable 30 FPS from either FSX or P3Dv3 with a 1m/pixel photorealistic Switzerland covered with full density autogen (OSM base for buildings and Corine for vegetation), shadows casted by aircraft, ground and clouds, received by everything, real weather with broken cumulus ceiling and cirrus layer on top, ground and air traffic.
With the same scenery (not recompiled with P3Dv4's SDK) and equivalent settings I easily get >80 FPS from P3Dv4.
As we found out, recompiling sceneries with LM's P3Dv4 SDK tools brings even further improvements (as well as new challenges, since some old solutions have changed in the process) and heavy sceneries with millions of 3D objects behave wonderfully.
BTW, 40 FPS in FSX/P3D feels smoother than 120 FPS in Aerofly FS due to some micro-stutters and image hickups/tear-off in the latter when not limiting framerate.
As a result all third party editors are now busy updating their products for P3Dv4, widely considered the long awaited but really well-born successor of FSX.
Talking to several scenery editors, I really struggle to convince them AFS2 is worth a closer look.
Many of them unfortunately consider AFS2 is too far behind with an unsuited "all-in-one-project" scenery architecture, while LM provides an open platform with a complete, professional and documented SDK, without the need for external 1'5 kEUR/year licence 3D tools.
For many reasons XPlane 11 isn't an alternative to many editors, or only to feature some aerodromes...
I'm convinced Aerofly FS2 has the potential to become a great simulator, offering something slightly different, easier to access and maintain than XPlane or P3D. I love the fact that 20 seconds after the double click I'm ready to fly on cristal-clear textures with high ranging FPS.
I believe the AFS2 graphical engine will be able to smoothly display large, detailed and consistent sceneries (not just large desert areas or micro-patches of detailed ground in the middle of nothing) provided that they're correctly optimized. The real challenge for autogen optimization is the roof of houses. In that sense I tend to consider LOWI is quite sub-optimal, dropping my FPS by 60%. The graphical engine doesn't collapse thanks to the fact it's only a few sq km, but I don't believe in making a country-wide scenery like that.
My 2 cents...
Cheers
Antoine