I wouldn't call it a trick, but I think (hope?) next gen should net very nice increases in visuals without a ton of extra work on the backend of content creation *in all cases*. Take Gears of War where Epic is already using a lot of very high resolution source art--millions, even tens of millions, of poly source art. Their current ingame content models are in the thousands of poly range with normal maps.
Next gen, hopefully, will allow for a nice bump in the game models geometry + higher quality normals, diffuse, specular, etc as well as POM and maybe even some intelligent displacement mapping where it makes sense. In addition to much better lighting and shadowing techniques. These things will require more engine work, especially displacement maps (but the same thing was said of PRT on animated objects this gen). Anyhow, once the render tech is worked out the same killer source art used this gen can be utilized to much better results in game.
Laa-Yosh and others can correct me if I am wrong, but it seems this generation had a lot of growing pains as many developers were going from hundreds to thousands poly models and then moving on up this generation to models in the 10k-50k range with an emphasis on faked geometry via normals. This requires new tools and new skill sets and a lot more time. The tools are a lot better now, the industry is more experiences, and while cutting edge games may need even more detailed source art I am not sure every games art budget needs to take the kind of hits from the Xbox/PS2 to the 360/PS3 did.
