Exactly! It's down to the devs to pick their targets based on their vision. I don't think it's quantifiable to determine a best practice for any particular game.
I don't think it's necessarely their
vision, even if that's what they'd say in public. (Complex) games are always at some tradeoff. When a racing game developer decides to go 30fps, it isn't because it's their vision of how the game plays best - but probably how their game will sell the most, have the best tradeoff with that goal in mind. Because sales are what matter. And sales are in some form dependant on PR, how you can market the game, which is largely dependant on how the game looks... The more shiny-shine you have, the more attention you'll get (for the most part - or unless you've already created a big name for yourself in the industry).
Shifty said:
I'd say it argues for the opposite. The stuff that's moving really fast you don't care for. The stuff you care for (road and other cars) is moving proportionately very slow, like a few feet per second, and at distance they're moving a few pixels frame to frame. The motions being tracked are slight so the temporal resolution can be lower and more spent on making those cars shiney-shine plus time-of-day sunsets to look pretty when you round a corner, which lots of drivers like.
Really? Think about it again. You yourself said it:
the stuff that's moving really fast, you don't really care for. I agree. So that means that on those objects, we can effectively spend less resources. When you continue to the cars, which are moving slower and stick around in your peripheral vision, we do have objects that would benefit from more resources. What you are describing here as more or less the optimal trade-off is precisely what Polyphony Digital has been doing for years: Spending foremost their resources on making the cars ridiculously good looking with all the detail while to a degree neclecting background, landscape to the point it's "good enough". Also, you might not be aware of this, but GT5 (and 6 I believe) do in fact have changing daylight conditions - and even different weather settings as well (less dynamic though). A title, on PS3, at 60fps. Maybe not on all tracks (AFAIR), but that's probably not a limitation of their engine, but a limitation of time and resources.
Extremely detailed cars, real-time dynamic time of day or weather effects are not the holy grail that are only doable at 30fps. If we go back a few years, you will even find that the PS2 game Jak & Daxter had it all back then - yes, dynamic time of day ranging from sunny to night, without a loading screen inbetween. And that all while the engine produces arguable still one of the most precise and accurate 60fps platformers we've seen since.
A racer, like DriveClub is, can effectively concentrate its resources on the things we focus on while playing the game; the interior of the car, the other cars you are racing, complex lighting, perhaps even real-time weather and daylight and take some short comings (tradeoffs) on landscape because it's usually fast moving and with a clever LOD system, you could get away with a result that I think would be quite kind on its resources. That already sounds like less of a resource heavy game where 60fps should be easier, than perhaps a FPS shooter where you don't have the benefit of high velocity moving pixels all the time. In a shooter, you might be standing still, observing the terrain, so every detail needs to be sharp as the player will not be focused on a small part of the screen - meaning that any trade-off you made to get there will be more obvious. A racer is way more predictable in that sense and poses different challenges to a developer.
Hypothetically:
If you give me a racer at 1080p and 30fps with 12 (or 18) cars and extremely detailed terrain or a 720p and 60fps with 6 (or 9) cars at slightly less detailed terrain - I'd take the latter any day. I also doubt that the 6 less cars would make or break the game - because the A.I of the cars are way more a factor than the quantity of them. Point in case: Gran Turismo. You can have what, 20 cars, but they are pretty worthless if they don't offer any challenge at all. Most of the time, you're only effectively racing with a small number of them, so you could just as well save some resources there.
For me, fast shooters should be 60fps, as you can turn in an instant an need to be able to interpret what you see ASAP. It's all about reactions which is where HFR is vital. That and resolution when peering into the distance. But some games are pretty slow and can get away with 30 fps for sure. Warhawk is a great example. The turning on that is sooooo sloooooowwwww I can't believe I ever played the game after playing more recent games. So the delta between events is small enough that 30 fps covers it. Factor in 4xMSAA for really crips visuals and distance detail, IMO it's the right choice for that game.