As for cgtextures, it's a useless site for a high-polish game - the textures are inconsistent in style and quality. (And that's what causes the real cost, polish). Believe me, I've looked at procedural content for a long time, and it's still a far cry from ready.
What we might see fairly soon is human-assisted procedural content, especially for all those areas of the game that are just there for ambience, not for actual gameplay. Even then, it's a large amount.
But then again, you'll find out all that for yourself :) (Good luck with that venture!)
I'm genuinely curious how much "polish" is actually needed and required by gamers.
My theory is that publishers have push the market to the point where games are very similar (not a bad thing in and of itself, but not inaccurate either) and so the differentiating factor for driving sales and advertising is visuals. That being the case, the increasingly far end of the tail of visual fidelity is being mined heavily, and "polish" is getting more and more and more expensive.
So, I think that you can get by on pretty solid gameplay and alright polish without needing to get a publisher to fund you--after all, it was their business model that made it so expensive to start with.
Does that seem like a reasonable conclusion?
They just found a way of doing that that is less asset intensive than "visual fidelity".
If you've found another way to make that work, you're set. But you pretty much by definition need to make an entirely different kind of game.
Note also that we need to solve the art asset production problem--look at sites like cgtextures or similar for stabs at this. Look at procedural content and better tools like Unity. Those are some examples of how we can make assets cheaper and artists' lives easier.
As for trying to be an indie... job in progress (check my profile).