Do you think we'll hit a Wall in Graphic Capabilities in the Next 25 years

[Alby_da_Wolf]

Mummelmann said:

Nem said:
If you only take realistic visuals into account, then yes... at one point it will hit a wall, wich is when it perfectly mirrors reality. But, the beauty of a virtual world is you can make it look however you like, so in that regard, creativity is the only barrier.

I love this answer; I personally think that the chase for photorealism, while it is both cool and an important part of the industry in some ways, is and will be only one part of the equation and that focus will lie elsewhere for the most part.
We're already seeing a resurrection of the hunt for brilliant gameplay above all in modern Indie titles and there is even a proven market for it, I think developers, especially smaller ones, will go various routes and ways towards the total gaming experience.

Input and immersion, like I said, will be the main focus overall in the next decade or so, that's my belief anyway.
Oh, and better AI would be much appreciated...

Before definitive photorealism is allowed through brute force of computing power available, we could meet a temporary wall first, when all the possible tricks and techniques to speed up calculations for the possible FX and features will be so many to be difficult to be all implemented in acceptable time in a commercial (or academic too) project with a deadline, then devs will have to rely on a few 3D engines updated to use as many new and better techniques as possible, but they'll have less possibilities than now to make their own engine, that will become more and more a specialised job.
Just to make an example, graphics quality progress made implementing functional mirrors more complicated 
http://www.giantbomb.com/functional-mirrors/3015-4618/
https://en.wikipedia.org/wiki/Transformation_matrix
And about AI, but also and even more for more realistic implementation of virtual objects, we'll need a lot more computing power, not just to handle the higher detail itself, but also to keep everything under control and prevent projects from becoming as mess, as detailed functional worlds will require strict use of object-oriented programming to keep projects manageable as game worlds grow and become more detailed through development, and with less and less possibility to take shortcuts with classic, nimbler procedural programming, this will require more power.