First, we know something to be true. However, we are uncertain about how it relates to something else. But we know that something to be true.
We know that, in real-time, the graphics engine does not come anywhere close to what it is able to do in non-real-time. We know this.
Now what we are uncertain about is our ability to assess real-time graphics performance based on non-real-time results. Can we look at a stack of Forza 3 Photo Mode images and conclude that, in real-time, its engine must be WAY better than its predecessor? The two seem scarcely related. All we can really say is that non-real-time performance has greatly improved.
Lastly, we know that non-real-time graphics performance trumps real-time. But it is a bit disheartening (to me anyway) to see that in-game performance did not get closer to the bullshots.
If the Photo Mode in Forza 2 was unchanged in Forza 3, we would be looking at gameplay and 'Photoshop' images and saying, "Holy %$#@!" But its evolution has raised expectations.
Now you guys know what to expect given the scope of the development cycle and hardware limitations and such. But it can be a problematic for a mainstream audience -- Forza 3 targets six to 65 year olds -- who do not have a technical outlook. Given the exclusionary nature of simulations, they are probably more familiar with the arcade games, where graphics is king. The most popular arcade racers are over a year old so it seems plausible that a yet-to-be-released title would push the visual envelope even farther -- this is, after all, what arcade games do. So it may be confusing for them who, after gawking at pretty pictures on mainstream websites, are expecting one thing (arcade sensibilities) but will be getting something else (a game that puts physics first).