Game-Design: Tilebased, Full-Analogue, and an alternative

Chat about anything
Post Reply
Posts: 101
Joined: Sun Apr 18, 2010 4:28 pm

Game-Design: Tilebased, Full-Analogue, and an alternative

Post by Lyx » Sat Sep 11, 2010 1:56 am

To avoid misunderstandings, something important first...

This is NOT a proposal how to change frozen synapse for multiple reasons:
- It is way too late for frozen synapse to make such a fundamental change
- The alternative model presented by me, would drastically change FS gameplay - it would play very differently.

What this is mainly is education about what effects certain design decisions have, and how to pick one which matches the best what you want the game to be like. It's an article (and of course discussion) about game-design - especially tactical ones - in general.


With that out of the way, lets look at what we currently have in the videogame world - which mostly is full-analogue (well, not really, but close) XOR maptile-based. Instead of full-analogue, pixel-based also is used frequently... the practical differences between both however aren't that big.

The advantage of "full-analogue" of course is that it is free of all the restrictions of usual tilebased. It also is more realistic - but since i dont give crap about realism, i'll wont go into further detail in that regard. The disadvantages about full-analogue are the same: that there are no restrictions.

There being no restrictions not only gives you more freedom - it also gives your opponent (the one whom you try to outsmart) more freedom. Furthermore, both players of course must deal with all that freedom... regarding setting their own commands, as well as expecting the opponents commands.

As you can guess, thats quite difficult if the amount of possibilities is practically unlimited. If you can do one million things, and the opponent can do one million things, then thats a lot of possibilities to consider. Fortunatelly, in practice it isn't all that bad, because the gamelogic strongly reduces the possibilities. For example, while in frozen synapse the FOV can in theory be set in countless different ways, the only thing that matters is if a unit is in the FOV or not... as long as you have an idea where to look, you will succeeed in looking in the right direction - regardless of if you set your FOV 5 degrees too far to the right, or 5 degrees too far to the left. What mainly stays very unstable and unpredictable are just the border-cases. This way, even though the players still have to deal with that fine detail, they can set commands without succeess being like a lottery win. At this point, its important to recognize the irony in this: Full-analogue can work, because there will be aspects that will NOT be full analogue.

The downside is that the full-analogue thing may still strike back in all kinds of ways - like for example, on the temporal dimension. Our FOVs may not make a difference in "how much something is in the FOV", but it will still matter if you move around a corner 0,1sec earlier or later, because that corner will in a tiny timespan drastically change what a unit can see. So, even though full-analogue can be tamed.... it may strike back whenever you're not paying attention. And of course, players need to also apply that fine precision when setting commands. Finally, placing a unit right near a corner by necessity must turn into precise pixel-tuning - else it wouldn't be full-analogue.

So, what you get with full-analogue is a model in which you indeed can do whatever you want, anytime - but you also have to deal with it, no matter if you need it in a given situation, or not.

The traditional alternative, is "Maptile-based". In that model, every "object" tends to be exactly one (or multiple) tiles large. The positions at which you can place objects also are exactly the same grid. This makes stuff much easier to do and to predict, but it also can be very limiting: Take a window - in a tilebased approach, you can stand immediatelly in front of the window to look through, or next to the window (no look through) - nothing in-between. So, now you don't need to deal with countless possibilities and a scale that is rather "microscopic" for a tactics game... same for your opponent - stuff becomes easy... you can finally think "highlevel". But: The level now is so high, that there actually isn't enough accuracy, resulting in cases where you need to decide between two placements that both are suboptimally.

Another drawback is movement and orientation: Movement can only happen from one tile to another - which in the case of squares leads to 8-directional movement. Okay, may still be acceptable, but it is a bit weird (and we will soon see that it can become even weirder). Whats even worse is if orientation also is limited to 8 directions - i guess you can imagine the frustration.

Well, here's an idea - why should the tiles that determine gamelogic (position, visibility, etc) have the same resolution as the map-tiles? Let's first try just making the gamelogic-tiles have double the resolution and shifted by half a maptile: You can stand on the center of a maptile, or at one of its 4 edges and 4 corners. Great, now we gained a lot of flexibility - it may actually be the optimal precision needed for gamelogic - but there is a little problem: If you stand on the "edge" between two tiles, then on which tile are you standing? In case you don't notice the problem yet: Lets say to the left is a water-tile and to the right a land-tile - if we stand on the edge between both, are we now in water or on land? And what about collisions?

So, even though this model is tempting, it isn't that logically sane. The next thing we could do, is to divide each maptile into 3x3 gamelogic tiles:
grid-styles.png (2.24 KiB) Viewed 8734 times
This may actually be more precision that is really needed, but at least now we know on which maptile we're standing. AFAIK, Jagged Alliance actually used a similiar system internally to calculate cover more accurately - but sadly didn't use it for anything else - we shall soon see why. Okay, so we now have terrain at just the right granularity, and have gamelogic in a granularity that may actually be even higher than needed. Sure there will still be cases where we cannot do something super-ultra-precise, but for most situations, the gamelogic-tile resolution would be more than enough. There is just one little problem....

Movement and Orientation.

Imagine 8-directional movement on such a fine grid - units will look like epileptics. Also, the question is still open how to handle unit-orientation. More importantly: A grid strongly limits gamedesign of unit-speeds: the speed of a unit must either be a multiple of tiles per tick, or we start doing those weird "action-point" calculations. Both kinda sucks.

So, with grid-based we're in the dilemma that on the one hand, we may want higher resolution for unit positioning.... but at the same time the flaws of the grid-model will be come worse, the lower we make the resolution. But if we choose full-analogue, we immediatelly are confronted with all its sideeffects, which may bite us in the ass later.

Well, here's a possible way out.... a way to cash in the best of all worlds, minus their downsides....

We let units move full analogue - we also allow full analogue orientation - but everything will be quantisized to those 3x3 gamelogic-tiles. So, you'd set commands and waypoints on those gamelogic-tiles.... but units can move in a straight line from A to B... while they move, their gamelogic-position will be the gamelogic-tile on which it is standing... it's FOV also will be quantisized to those game-logic tiles.... same for possible visibility rules. A good way to imagine this, is a person walking in a room over tiles on the ground... when his feets touch the ground, the relevant tile lights up and he is registered as being on that tile.

I someway suspect some readers will at this point not understand what would be the advantage of this. Well, the advantage is that precision is dynamically at runtime scaled down to a level, that is needed for gameplay. Lets say you want to set the FOV of a unit - you'd still be able to rotate the cone arbitrarily... but you only need to apply as much precision for that order, for the desired tiles to light up - not more. Same for movement: You can make a unit walk straight whereever you want... but you dont need to pixeltune the target-location... just precise enough to hit the desired gamelogic-tile. When expecting enemy commands and movements, you too only need to think in as much precision as the gamelogic-tiles have - not more - from a gamemechanics-pov, a unit can only be on one of those tiles. And finally, we dont need to do strange stuff when designing unit-movement speeds - we can pick any speed, since "screen-position" and "gamelogic-position" are decoupled (in theory, what actually will happen in principle is a bit like a unit may at the end of a turn already be "0,2 tiles" towards the next tile - which will count as bonus in the next turn when it continues to move).

- Unconventional and therefore possibly not as easy to implement (lack of existing docs and code). That it AFAIK hasnt been done before in a high-profile game also may result in a lack of "memes" - players know tilebased and full-analogue already... but they dont know this model yet and thus may initially be surprised.
- If you actually want "full analogue" freedom/precision, then obviously nothing can be more fitting than full-analogue.
- If you dont really need higher precision than "maptile-based", then the alternative scheme just adds unnecessary bloat and complexity.

Things that may falsely assumed disadvantages:
- FOV border-encounter issues during movement - won't happen, because FOV is calculated from the actual screen-position and THEN quantisized to tiles - not the other way around.
Lyx: "Wot? Why do i lose that one?"
Chem: " if(playerIsLyx()) chanceOfWinning *= 0.3f;"
Lyx: "Damnit om!"
Posts: 12
Joined: Wed May 11, 2011 9:16 am

Re: Game-Design: Tilebased, Full-Analogue, and an alternativ

Post by Srekel » Mon May 16, 2011 10:13 pm

Interesting read. I had a similar idea for an RPG I was designing a few years back. However, seeing how FS works, I have to say that a sub-tile grid would offer no immediate benefit IMHO. (not just for FS, but for strategy games like this in general)

Correct me if I'm wrong, but the main problem you identify and want to fix in full-analogue mode is that there is some amount of luck to it, because you can lay out your plan and test a few cases, but you can never test all of them and there are a lot of edge cases - like you say, the enemy might be 0.1 seconds faster than you anticipated and your whole plan goes wrong. I've personally felt this pain in FS.

However, that said, I think that it's still the best solution.

Consider a case where you have instead a bunch of tiles that you can move on (even if it's analogue on top of them). First of all, you still haven't solved the above problem, because you still can't try out all combinations of your team's movements against your enemy's - there are too many combinations, even when there's just 1v1 soldiers left. Instead of 0.1 seconds off, you'll be 1 sub-tile off.

Second, you've introduced paralysis analysis to the game, or at least made it significantly worse. In analogue mode, you kind of have to build a plan that's "safe enough" - you know you can't test every case so you'll try a few different ones out and go for it. When you have tiles, you start to experiment much longer: "What if I move one tile to the left first before I step in front of the window tile", etc.

I can't offer any proof that that is what would happen, except from having played a lot of games and seen a lot of players get stuck in A.P. :)
Posts: 15
Joined: Mon Sep 12, 2011 1:42 pm

Re: Game-Design: Tilebased, Full-Analogue, and an alternativ

Post by dinhdai88 » Thu Oct 06, 2011 2:09 am

Great post, i like this, thanks alot
Posts: 3
Joined: Sun Nov 16, 2014 9:03 pm

Re: Game-Design: Tilebased, Full-Analogue, and an alternativ

Post by Kasztanek » Sun Nov 16, 2014 9:06 pm

Wow man ! what a great post :D
Posts: 3
Joined: Fri Nov 21, 2014 11:06 am

Re: Game-Design: Tilebased, Full-Analogue, and an alternativ

Post by Kasztan » Fri Nov 21, 2014 11:20 am

Nice to read this !
Post Reply