In addition to what Groover mentioned -- these consoles are a lot less powerful compared to PCs than PS3/360 were relative to flagship PCs -- there is also the concept of diminishing returns.
Let's just assume PS4 is
10X faster than PS3 in terms of graphics capability, let's put this into context:
These level of graphics in 2002 on FX5800 Ultra (Voodoo power rating of 4.6)
to this on GTX670 SLI (Voodoo power rating of ~ GTX690 = 383) in 2012
A move from 4.6 --> 383 or a GPU performance increase of 83X
The problem are not just consoles but the rate of growth in graphics capabilities is now not enough to create a big difference in next generation graphics.
Just read this quick article of how more advanced the New Dawn demo is vs. the original Dawn and you realize that to improve graphics to WOW us from the New Dawn level, we will need a graphics card
50-100X faster than a GTX980. Until then it'll be small incremental annual steps.
http://www.hardocp.com/article/2012/08/01/new_dawn_dx11_demo_compared_to_old/1#.VFAAAvmUfsd
Point is the 10X increase PS4 brought over PS3 is a drop in the bucket compared to what's necessary for a graphics revolution compared to what we have now at Crysis 3 graphics level.
Having said that I do not believe that PS4 and XB1 are fully maxed out. We should see better looking games in the next 2-3 years, at least in the form of Gran Turismo 7 and Uncharted and other gems from Sony's 1st parties who will optimize better for the console. As usual I expect PS4 to be 99% tapped out in 2.5-3 years and then we'll face 3-4 years of stagnation until PS5 in 2019-2020, unless developers like CDPR or Dice, etc. keep pushing the PC independently of consoles.
If we look at next wave of games like Far Cry 4, AC Unity, Dragon Age Inquisition, they look good, but not jaw dropping unlike how Unreal 2, Far Cry 1 or Crysis 1 or even Crysis 3 were relative to the other games in their generation around their releases. Crysis 1 really wowed me when I first tried playing it on my 8800 GTS 320mb.
Diminishing returns is going to kick in and advancements in 4K and eventually 8K gaming will put even more pressure on GPU developers to exponentially increase graphics performance in the next 10 years. With 4K, any doubling or quadrupling of GPU performance will be wiped out by more pixels, higher resolution geometry and textures alone. Then there is the problem that adding more polygons is not going to provide a huge difference anymore.
In Ryse Son of Rome the character models are very detailed at 85000 polygons and moving to 150000, while an improvement, is nowhere near 2x more detailed. Simply throwing more polygons at the game isn't going to double or triple the realism of graphics.
Old - 150000 Polygons per character
New - 85000 Polygons per character
You just barely see the difference now. Diminishing returns!
http://www.dualshockers.com/2013/09/28/ryse-old-and-new-builds-compared-polygons-vs-shaders/
For the next 5 years until GPUs catch up, developers should focus on making games more fun/unique because until TSMC or GloFo get their acts together, we will only see a performance increase of
4-5X (if we are lucky) in graphics in that time from a GTX980, which is going to be nothing compared to what's necessary for a graphics revolution.
*** But we can still extrapolate game engine and in-game optimizations by say comparing how good a game like AC Unity looks vs. the hardware requirements and say Crysis 3 or Metro Redux series or Ryse Son of Rome. I wouldn't exactly use Ubisoft's optimizations in AC Unity as the epitome of extracting maximum value out of PS4 due to 'console parity' with MS on that game and generally speaking Ubisoft's lack of proven track record to optimize games well.