Many years ago, game level creators were both designers & artists.
Really? I have followed thirdparty level design on an off in quake and unreal since ’97 and very few mappers created textures, about the extent of “art” back then. Making levels look “nice” was fairly straight forward. Take textures and create some architecture that is modular-ish using these. Wall panels, supports, lights, trim, floors, etc. This ends up being on a power of two grid – with quake the minimum used was 16 from memory due to stair height. When mappers around here talk about “keeping it on the grid” this is what I think of – not making sure it is snapped to a single unit grid, which is quite often what they mean.
This is essentially what you are spruiking, but using meshes instead of textures and brushes. Good level design has always been like this and bad level design looked shoddy in comparison. But it was not art. By ‘99 this should have been simple and should have slapped anyone interested in level design in the face. The problem is many maps were not released this way and as a general rule they looked terrible as a result.
If level designers used a similar methodology blockouts would go a lot smoother.
At some point, game studios started to split the job into multiple jobs and now we have level designers and environment artists. Essentially these days… level designers create a blockout, test gameplay for two weeks, then artists arrive, they break down the level to create modular assets, they’ll change some areas of the map to fit the art, they’ll create unique assets to fit the blockout, etc. However, when it comes to the creation of game levels, the art is the bottleneck and there are many issues with the method that I just described. Designers waste a lot of time testing the gameplay when some areas will change anyway when artist will touch the map. Artists waste tremendous amount of time trying to break down the map. Artist will have to create more meshes than they should to fit the custom blockout, etc. The workflow that Hourences uses is a lot more efficient. He creates the modular meshes first and then build the level with those. Right from the start, that means that he doesn’t create a blockout and waste his time trying to adjust the gameplay when it will change later anyway. Because he uses meshes to build the level, he doesn’t have to break down a blockout into modular meshes, etc. He basically works at least 2-3 times faster than usual, plain and simple. And because most level designers these days aren’t also 3D artist, the quality of their map layouts often take a hit.
This sounds like web design ~8-10 years ago. You could contact a designer (level designer), get a layout slapped together in photoshop then have it converted to a page (primarily HTML/CSS) by a developer*(environment artist). Depending on the layout this could be a nightmare for the developer. The argument was that a good developer could make anything work, but unless the designer was aware of the pages restrictions it could be challenging and time consuming to do so. Of course there would be overlap, designers that knew limitations of the medium they were designing for and even designers that could put the final product together. The thing to keep in mind here is that the designers that knew how to put the final product together also generally easier to build as they were aware of these limitations.
While I do not work in the games industry it seems that you want modular assets created then levels made from that point? Essentially what Hources talks about in one of his videos (22minutes) – making 600 meshs that should cover most common uses. Then using these lego pieces to create maps. These meshes are standard sizes giving enough choice, but translate easily into some sort of blocking out methodology, be it BSP or something else?
In an ideal world where we have level designers and environment artists and it would make sense for both parties to have at least partial understanding of the both work flows. Using the quake screenshots quake you can clearly see how different assets would be created for these bits of architecture.
Level designers ignoring this and going full 13, 29, 42, 87, 429 units is dumb. Outside of a few edge cases like stairs and narrow openings (which could be a “standard” size anyway) I fail to see the reason to do it.
Your gripes with the industry sound like a mix between two groups that do not communicate, two groups that do not understand each other or deliberately go out of their way to make things difficult for each other. Or even a group that has forgotten its roots (level designers) and just make things 100% free form. How is that ever going to work?
In the past maps that did not follow some sort of style similar to the quake examples tended to be bad.
*Not sure this is really the correct term for the simplistic example but it should do to illustrate the example.
Level Design