My original intent here was to discuss a game idea, but I've come to the conclusion that I'm probably not all that interested in pursuing it, so I'll just use it as the guiding example for a broader topic, namely the concept of using non-static levels for a racing game. Example: In the hypothetical game "Sunracers", you play a real edge-of-the-envelope daredevil whose passion is racing into the interior of stars via cooler flows embedded within the larger structure of hot fluid/gaseous stellar material. Think of the dark spots on renderings of stars' surfaces. Players would have to ride transient flows I have a couple of theories on why this hasn't come up in a game: 1) Racing game players typically concentrate on besting their own times, and an ever-changing course frustrates this aim. 1b) More specifically, when a racer becomes familiar with a course, he can find the best line through the entire course, which is more complex than simply finding the best line through individual turns. He navigates a turn specifically so that he is set up well for the next few turns (i.e., taking the turn wide or tight, hitting the apex late or early, etc.). 2) The average racer already requires pretty excellent reflexes, and changing the course ups the ante yet again. 3) This type of game doesn't really represent anything else, while racing games recreate something many people would love to, but will never get to, experience in real life. To some extent, I think a counter to objection 1) could be found in using course "characteristics", such that they develop a feel for the course without knowing exactly what's going to come up next. Example: In our hypothetical game, Sunracer, players cruise a "flare star". This star experiences frequent upwellings of superhot gas, which pushes other flows outward as well, ending up in a solar flare. The player's main skill in this course is the ability to rapidly change direction to stay within a single flow. In the "bubble star" course, on the other hand, most flows are extremely limited in scope, frequently merging and diverging and the player must be able to efficiently escape their original, sinking flow in favour of nearby rising bubble flows. The moment-to-moment of the course is never exactly the same, but the player can still learn enough about the parameters under which it operates to learn to play the course. Objection 2) simply means that this type of game appeals to a different market, probably more related to FPS games than racers. This limits appeal, although the originality factor would probably add at least some weight to the game. Objection 3, to me, indicates that the market would have to be built carefully. Overall, I don't see any dealbreakers in the above. What are your thoughts? Does this type of play have any merit, or is the standard static-course racer the only way that you'll end up with a fun game? ld Edit: Added Eric's objection #4 as 1b [Edited by - liquiddark on July 19, 2004 1:47:02 PM]

In a nutshell the idea the same as tetris. Tetris shares the three points that you listed.

For the theme, tetris is able to make it abstract and simple so that there are more audience. In your case, it sounds like a water surfing game.

Compared to a skateboard game, your game lacks the the different styles that the player can do, and those styles have references to reality.

What other interactions does the game provide? Where would you place Mario Cart on your spectrum?

I don't think you've countered objection #1 very well. If you want the player to enjoy running a course over and over again, you need to give him some kind of hard metric to improve upon.

How about this addition to your objection list:
4) When a racer becomes familiar with a course, he can find the best line through the entire course, which is more complex than simply finding the best line through individual turns. He navigates a turn specifically so that he is set up well for the next few turns (i.e., taking the turn wide or tight, hitting the apex late or early, etc.).

You could try to get this kind of depth with Sunracers:
1) Include a complex series of turns in the course.
2) Make the entire series visible to the player ahead of time, via an overhead map or whatever.
3) Put a long straight-away before the series, so that the player can relax, look at the map for a few seconds, and figure out a line through the series.

However, in most racing games that I've played (even those with overhead maps), I've only been able to find the best line via trial-and-error. (I'm reminded of a Matrix quote, "There is a difference between knowing the path and walking the path". :))

Quote:

Original post by Estok In a nutshell the idea the same as tetris. Tetris shares the three points that you listed.

This is an excellent insight, and combined with the system Eric mentions below wherein the player can see the whole area at once, provides a much stronger core focus for the game. I'm thinking along the lines of a TARGA or rally race now - piloting the vehicle is one aspect and navigation is a totally separate aspect.

So the next question is, how can those two be dealt with intelligently as a single unified concern? For a two-player coop session, the question almost becomes easy - one player drives while the other navigates, and they switch up at will. For a single-player game, I'm thinking that maybe the copilot could be an AI character, and choosing the copilot for a particular track would be one of the games within the game. Or the player could go it alone and have to make all the hard choices by themself (pardon the pronoun).

Quote:

Compared to a skateboard game, your game lacks the the different styles that the player can do, and those styles have references to reality.

I don't really think of the game in the same headspace as a "tricks" game like water surfing or skateboarding. Something like Driver or, as you mentioned, Mario Cart, might be a better starting point of reference. Driver has the beating heart of a dynamic-course race game, although not to the extent I'm talking about with SunRacers specifically, while Mario Cart has the FPS element in spades. In all three cases, the core focus is on control, speed, and turn-on-a-dime decision-making.

Quote:

What other interactions does the game provide? Where would you place Mario Cart on your spectrum?

My intuition tells me that the game's focus should be primarily on the classic racing experience - the player can unlock or buy chassis and tweak them for specific characteristics, plus maybe the above-mentioned quest for a wingman to run the map if you're running a car by yourself. It could be worthwhile to consider a few powerup items - a coolant gun to extend the flow just that little extra bit, heat shielding, an emergency rescue device, those sorts of things.

ld

Quote:

Original post by Eric I don't think you've countered objection #1 very well. If you want the player to enjoy running a course over and over again, you need to give him some kind of hard metric to improve upon.

This is a hard assumption to get around, isn't it? I don't necessarily agree with it, however. I think with a dynamic course the player has rewards besides their time. Maybe they could choose from a set of "configurations" which would govern the difficulty of the race, taking from the idea that stars go through activity cycles. So if they wanted to just sit back and enjoy the ride, they could choose a low-intensity configuration, whereas if they wanted the ride of their lives they could choose the high-activity point. I'd compare it to another game - I think it was Grand Turismo 3 - that had a car that you just couldn't drive sensibly on anything except the simplest courses. Once you unlocked everything, the whole challenge of the game became being able to complete the hardest courses with that car, which was so high-performance that you couldn't keep a set of tires on it. In this case, the courses would be the high-performance items, but the idea seems similar to me.

Quote:

2) Make the entire series visible to the player ahead of time, via an overhead map or whatever.

This is the choice that strikes me as strongest. What is your opinion of the use of a navigator separate from the driver? Of course, the driver could still choose to look at the map (which would be in 3 dimensions for the case in question) themself and decide on a through line, but as courses got tougher they'd want to start utilizing a navigator so they could focus specifically on the driving aspect of the game.

Hmm. Another thought occurs to me: What if they could choose to be a navigator, and hire driver AIs to pick specific through-lines? This seems to put the game's focus on two separate fields of play, which don't necessarily have much to do with each other. But it might be useful to be able to run the course from the navigator's seat a few times just to get a feel for it. Meanwhile, the player could always choose to switch into the driver's seat if they wanted to run a particular section.

ld

Quote:

Original post by liquiddark My original intent here was to discuss a game idea, but I've come to the conclusion that I'm probably not all that interested in pursuing it, so I'll just use it as the guiding example for a broader topic, namely the concept of using non-static levels for a racing game.

Well, nevertheless, congrats on a VERY cool and original idea, though I don't even want to think about what sort of technology is involved in such an endeavor on a SPORT basis. [grin]

Quote:

I have a couple of theories on why this hasn't come up in a game: 1) Racing game players typically concentrate on besting their own times, and an ever-changing course frustrates this aim.

I love racing games but generally hate fixed tracks, in part because I know that there is a right and wrong solution which the AI has already been given (I don't mind them in PvP games unless I know the other guy has practiced on the track for weeks).

Midnight Club-- which very much approaches dynamic racing because you do so through waypoints in an open city-- I really like because although there are optimal paths, luck, risk and skill can even out many of them. Smuggler's Run is another, ironically in its mission rather than its pure race modes.

Having said that, I still end up using the cruise mode in the game to learn the city. If I don't, then my success is more likely to be based on luck.

Quote:

1b) More specifically, when a racer becomes familiar with a course, he can find the best line through the entire course, which is more complex than simply finding the best line through individual turns. He navigates a turn specifically so that he is set up well for the next few turns (i.e., taking the turn wide or tight, hitting the apex late or early, etc.).

I want to note that this doesn't necessarily apply if the track is very linear. Hydrothunder is an example of this: The tracks do turn, but because you're in boats you rarely get sharp turns. There is a certain amount of drift to the speedboats, and so the game's focus is on collision avoidance with terrain and objects in the water and manuevering into position with (sometimes changing) speed boosts or jumps. I never really had to learn the course all that much, and so the game focus for me and my friends was on navigation. The tracks were wide enough that it was very hard to crash into the walls, and you could often see well ahead, which made course planning very secondary and the trip enjoyable.

Contrasting this to something like Star Wars pod racing, whose course I had to memorize to play well: You have various ships with drift and navigation challenges AND you have to learn to course. If Hydrothunder's course were dynamically different, it would be fine so long as the difference wasn't so overwhelming that it was impossible to respond to individual changes (no light-cycle racing with 90' turns). But then you'd have to question the impetus for dynamic courses at all: To provide challenge, or to provide changing scenery.

Quote:

3) This type of game doesn't really represent anything else, while racing games recreate something many people would love to, but will never get to, experience in real life.

Given that there have been space racing games in exotic tracks, through rings and on weird landscapes of all types, I don't think this is much of an issue. If someone builds it, sci-fi fans will play it. But as you've noted, this is a subset of ALL racing fans.

This sounds like a really cool idea, and I hope you go somewhere with it. I would say the biggest draw would be the challange involved in navigating the course. All you really need to do is have a way to create dynamic tracks of increasing difficulty and you'll get the player hooked.

A big challenge would be to get the AI to handle the dynamic course. I remember an Arcade racer (forget its name, but it was a sequel to HydroThunder with a space theme). If I played the easier courses, the AI would whomp my butt. But, on the harder courses, the AI had so much trouble that I could beat them almost every time. I ended up playing for over an hour on one credit.

Imho, a compromise approach is the best. First, make the gameplay focus on player interaction, not on the course itself. The course is of minimal importance - what matters is sucking yourself into other player's jetstreams, ramming them into walls, tricks like that. Thus, the featureless ubiquity of the course is irrelevant. Second, let the players learn the track - keep the map very short and roughly constant. Only change it per race, not constantly. But still randomized, so players who've memorized tracks won't have upper hand.

Personally, I love racing games, but I never play them just for that reason: the winner is not the best driver or pilot or fighter, but the one who knows every turn, every angle. If I wanted that kind of gameplay, I'd just play trivial pursuit with a stopwatch. This is why one of my fave racing games is an obscure half-life mod called Turbo. The tracks are dead simple-the focus is on obliterating the other racers. Same goes for Tread Marks (too bad the game is simply too clunky to enjoy).

Hey, that's a great idea Pxtl :)
So the game could look a bit more like Wipeout. This could get really cool, like the best parts of the pod race in Star Wars (I mean, bumping someone so he rams against a wall has got to be cool :) The weapons and powerups can be simply holographic, and the judges detect when you pick up something and you automatically have permission to use it (or it gets teleported or something).

Also, don't just make it random. You need to make it sensible somehow, because if for example the width of the track is constantly changing between 2 extremes it will get really frustrating. Perlin Noise would be perfect for this: sometimes the area would be wide for a large area, sometimes it would tighten for just a few meters, and it would be more interesting and "organic" than a plain distribution of random numbers. The human brain loves that :P

An example implementation of this:
Right after the visible area of the track, you're basically laying down bricks, constantly creating the path for the player to race on (likewise, the track behind the player gets destroyed). No matter what method you're using to represent the track, you'll need to create one "horizontal" line at a time. Using Perlin Noise, the intensity (if it's on fire or on shadow) for each patch of this line can be figured out, with a 2-dimensional grid where one axis is the line, and the other is time - as time passes, you feed the Perlin Noise function each one of the positions in this line as one coordinate, and the elapsed time as the second coordinate. It might be kinda hard to visualize but once you do it it's really simple. Another perlin noise function could be used to figure out the turning rate of the track, to create different turns as the player progresses.

There's another problem... how do you make sure there's a track that can be travelled on?

Sorry if I kinda wandered off into the topic of finding the algorythms to implement this, but I tend to do that :)

Quote:

Original post by Korvan I would say the biggest draw would be the challange involved in navigating the course. All you really need to do is have a way to create dynamic tracks of increasing difficulty and you'll get the player hooked.

Maybe the courses would start out with cool, stable stars or even brown dwarf types, and progress through a set of increasingly hot star types, then loop back through with the activity parameters increased. The player would have the same basic experience, but every time they'd have to take their game up a level, so to speak.

Quote:

But, on the harder courses, the AI had so much trouble that I could beat them almost every time. I ended up playing for over an hour on one credit.

I hadn't really considered the implications for an AI, but of course you're right - an AI faced with a track that they didn't have the keys to would be hard-pressed to work through the bumps. I can't think of an answer for this right now, but I'll keep pondering it.

ld