an 'underground' videogame developer would need a relatively huge team and budget in order to make graphically competitive games; and regardless of how many good ideas they might come up with, they would be crippled by this lack to compete in the graphics department.
Yep, that is exactly the problem. The sheer amount of work that has to go into the creation of assets (models, textures, levels, etc.) nowadays has gone through the roof due to the incredibly rapid pace of technological advancements and the increasing complexity of the assets that comes with it. Seriously, it's insane. Only ten years ago almost every game was still primarily rendered in software and hardware acceleration was still in its infancy (the 3Dfx Voodoo card being the first widely available 3D accelerator). Today, just ten years later there are videocards that when looking purely at raw fillrate (which is simplifying things but I don't know how to otherwise easily compare such completely different architectures) are like 800 times faster than the very first Voodoo card ten years ago.
And that insanely rapid increase is visible in the complexity of game assets that people have come to expect. Your average player model in a next-gen PC or console game is probably built out of more more polygons than an entire Quake level is. And asset creation is still pretty much entirely a process of manual labor.
So yes, nowadays in order to make a competitive game you need a team of like 50 people with which it will take you anywhere between 3 and 5 years in order to create something that is ready to be put on the market. To put things in perspective, the original Doom game was created by a team of about 6 people in about a year and a half (with 5 of those only having worked on the project for less than a year because the rest was the time it took to write the engine which was done by one programmer before the rest of the team could start working on it).
This doesn't just affect the business end of things but also the amateur game development scene. The original Doom games and Quake 1, 2 and Half-Life (based on Q1 engine) were all very popular platforms for amateur game developers to build their own modifications on. And all of those games saw loads and loads of mods being released (most of which obviously not being too great but they all had some really good ones too). Nowadays the sheer dedication and skill it takes for a bunch of amateur developers in their spare time to actually build something that looks even vaguely on par with the game you're modifying is staggering. Which is why you see a lot less mods being released for Doom 3, Half-Life 2, Unreal 2k3, etc. It's gotten to the point where the average person simply doesn't have the technical and artistic skill, let alone the spare time and dedication, to create things of that kind of complexity anymore.
Basically the industry has raised the bar too much and too quickly, making it nearly impossible for companies or individuals without a lot of money to really compete in it anymore. Which leaves the initiative with the big publishers who do have a lot of money but are only willing to spend it on your videogame on
their terms, which means game studios often end up relinquishing a lot of creative control over their games to publishers who use criteria like "will this cost us money?" rather than "will this make a fun game?" in order to decide whether a game design decision gets the go-ahead or not.
Like I said a few posts ago, what the industry really needs is an equivalent of Miramax Studios (and other independant film studios). Companies with money that are willing to invest it in games that are not sequels of already established franchises, or rehashes of concepts that have been proven to be financially viable in the past. And I think that will probably happen at some point.
