i think the possibilities are a bit limited, i mean how big can the jump be,  graphics are almost real!!

add all that together and add cohesive art direction and you should have a good idea of where next gen should be, I recomend watching in HD especially the seccond to last one...

Look at some ray-tracing examples:

http://hof.povray.org/

Many of the pictures in there don't look real, but they're already far beyond what current games can show.

The problem is I've seen some people on this site, as well as on other sites, saying that actual the ps3 real-time graphics has fullfilled or even surpassed the e3 2005 expectations. I don't agree with them (even if Uncaharted is quite close), but tha fact that there is some disagreement arount this point shows that there isn't actually such a great difference, anyone can tell the difference between ps2 and X360.

Using new capabilities can be done in non-expensive ways, I'm sure (the cel-shading revolution will rise again)

Most of the features are included with the upgraded engine license.  The higher density geometry models would be easier to retopologise from the zbrush/cinematic assets as they have more polys at their disposal.  There might be extra expenses outside of the engine for facial animation mapping tools etc. This is one area that I think we'll see a big improvement, hopefully gone are the days of bad lip syncing.  

Depends on how much the new consoles will cost.

I doubt they're going for 1080p and 60 FPS just yet. That's because better graphics is valued over high resolution and smooter gameplay (people think 720p is enough, and for some reason people also think that 30 FPS is enough). Instead, it'll be something like Epic and Crytek have envisioned. On the other hand, lower-budget games will surely put more focus on resolution and framerate instead as I believe it'll be considerably cheaper than top-notch graphics.

Either way, we're going to win. Personally I'm hoping for higher framerate and more dynamic and open environments, even if it comes at the expense of graphical quality and resolution. 720p and 60 FPS is pretty good already, there's no need to rush for 1080p until we get 60 FPS. That said, I wouldn't also mind getting more seperate graphics settings for different TV sets as well: For example, game A runs at 720p@60fps, 4xAA on a 720p TV but at 1080p@60fps, 0xAA on a 1080p TV. Something like that. I can't see it being too hard in all cases.

Be consistent. Either you do the comparison just for one company, or for the industry as a whole. That way you will actually get much better data.

Earliest release - Earliest Release

Dreamcast - x360 = 7 years

360 - Wii2 = 6.5-7 years, depending on when in 2012 it launches.

Latest Release- Latest release

Xbox - PS3/Wii = 5 years

PS3 - PS4/xbox = 6.5-7 years depending on when in 2013 they launch.

Sony release - Sony release

PS2 - PS3 = 6.5 years

PS3 - PS4 = 6.5-7 years

MS release

xbox - 360 = 4 years

360 - 720 = 7.5-8 years

Ninty Release 

GC - Wii = 5 years

Wii - Wii2 = 5.5-6 years

I would hardly say that the evidence points to the gap between generations being 2.5 years long this time. For example, averaging them out we have 

(-0.25 1.75 0.25 3.75 0.75)/5

=1.25

Make it 1080p and 4-8x Antialiasing. That would make a huge difference already.