[AaronWizardstar]

This is for a turn-based tactical RPG like Shining Force and Fire Emblem.

What kind of movement systems do you like in these kinds of games? Do you like systems where a character first moves then attacks, like in the two above examples? Or do you prefer characters having action points that can be spent on moving and attacking in any order, like in Fallout?

I've been leaning towards action points because of their flexibility compared to the strict move-then-attack system, but when I think about action points they seem hard to balance. Suppose I dictate that a character can move 4 tiles and attack once in a single turn. The character can move and attack in any order, as well forgoing his attack for a greater movement range or some of his movement for an extra attack. How should the cost to attack and move be balanced? One could make attacks cost 4 points and movement cost 1 point, where the character above would have 8 points total. He can attack once and move four tiles, or stand still and attack twice, which seems fine. However, if the character doesn't attack he can move up to 8 tiles. Doubling his movement range by not attacking feels a little extreme.

What I'm thinking right now is to have attacks cost a percentage of a character's maximum action points instead of a fixed point value. That way a character's movement range can be kind of separate from the number of attacks the character gets per turn. A character's movement range is increased by increasing his total action points, while the number of attacks he gets is increased by decreasing the attack cost percentage.

[RampantCoyote]

My favorite tactics game wasn't actually an RPG... X-Com (AKA UFO: Enemy Unknown). It used action points for everything. Tactically, it was more realistic and flexible.

But that being said... it's always a little annoying as a player trying to make that trade-off between movement and attacks. You always feel like you are "sacrificing" an attack to move - and in a game like X-Com, where a single hit can be very significant, it's even more frustrating.

So just to make it easier on the players - especially less experienced tactics players - just having a move action followed by an attack action is a lot simpler and probably less frustrating (which may mean "more fun.").
_________________
Tales of the Rampant Coyote - Old-School Game Developer talks Indie Games, RPGs, and the Games Biz

[AaronWizardstar]

[RampantCoyote]

I think earlier editions of D&D assumed a certain level of movement in conjunction with a normal attack... then third edition changed it into move actions, attack actions, or a "full action" that allowed multiple attacks if you didn't move more than 5 feet.

One thing I really liked about X-Com and similar systems (though again, it may only be appreciated by experienced gamers) was the idea of holding back some of your action points for "attacks of opportunity" that could take place during your opponent's turn.

But really, I'd suggest not breaking things down too far. Spending only two or three points per turn would probably be more appealing to a player than having to figure out how many points it would take to move 5 tiles diagonally. That way, you add a little bit of complexity & flexibility, but it's still easily digestible.

But if I were designing it for an audience that wasn't already quite familiar with tactics-style games, I'm probably just break it up into being able to make a BIG move, a SPECIAL attack, or an attack + move in either order. That's really what the whole thing boils down to, isn't it?
_________________
Tales of the Rampant Coyote - Old-School Game Developer talks Indie Games, RPGs, and the Games Biz

[AaronWizardstar]

[Rainer Deyke]

[RampantCoyote]

[AaronWizardstar]

[Malignus]

The action points approach is cool; so is move-then-attack. It's just a matter of how you want the game to play. I find move-then-attack more accessible and conducive to planning out moves in advance, particularly when dealing with large numbers of units. However, if your game features small numbers of units (say, 6 or less under player control), AP-based may help lend the battles some much-needed complexity.

From a development standpoint, the thing that scares me most about going with an AP-based approach is the thought of coding an intelligent enemy AI that can cobble together its turn by performing any of a variety of actions in varying combinations, in a varying order. Coding that sort of AI just sounds like hell on wheels to me. But then again, I'm sure there are programmers out there who think that sounds like fun.

[RampantCoyote]

I'm one of those guys who thinks that sort of thing is fun. Yeah, they keep me in a cage and let me out for just that kind of thing... ;)

I'm fairly happy with what I did with the spellcasting AI for Frayed Knights, though it's not exactly brilliant. It's just a weighted heuristic system based on a number of random spell / target choices based upon the AI's intelligence ("brains") rating. A smarter AI will consider more candidates. It includes a random factor in the scoring, and it can (but currently doesn't) include a personality bias as well. Now that I have fixed a couple of bugs in it, the AI can make some pretty clever moves now with spells. Though they do like to pick on a target that's nearly dead.
_________________
Tales of the Rampant Coyote - Old-School Game Developer talks Indie Games, RPGs, and the Games Biz

[Malignus]

Jay, that sounds pretty much exactly like what I'm doing in Telepath RPG: Servants of God. In varying degrees, and depending on how high the game's difficulty is set, enemies favor attacking healers, the hero, characters that are more wounded, characters with lower remaining hit points, closer characters, characters they can backstab, characters that don't resist the element of their chosen attack, and characters they can take out in a single hit.

But I'm sure you realize that with a strategy RPG, it's much more complicated than that. The battles occur in 2D space. So enemies have to be able to navigate terrain, determine who they can backstab, weigh opportunities for group damage with AOE attacks against focused attacks against high-priority targets, protect/capture objectives on the battlefield, and so on.

Coding all of this in TSoG was a huge, huge pain in the rear. But TSoG's AI is pretty simple as strategy RPG AI goes, in no small part because enemies move in a fixed turn order and don't have AP. In the new engine I'm developing, enemies can take their turns in any order, which exponentially increases the number of factors a good AI will have to account for. Not only does the AI need to know what actions are optimal for each character and how it can perform those in 2D space, it also needs to consider the optimal order for characters to go in, and--in cases where one character performing its optimal action forecloses another character's optimal action--whose preference to give priority to.

Personally, I'm still at the "holy hell, I'm not even sure how I am going to code this" stage. The mere thought of adding in the complication of AP on top of it all just makes me want to dive screaming through a window.*


*Not that this should dissuade you, Aaron. There is an excellent chance that you're a better programmer than I am, if for no reason other than there's excellent chance that anyone who programs video games is a better programmer than I am.

[RampantCoyote]

Heh - WAY back in the day, I experimented with Genetic algorithms for racing around a track in the Playstation game Jet Moto. I figured I'd let the AI figure out an "optimum path" for the track (in the first game, the physics were way too expensive for the AI to use the player physics - the AI had to run ten times faster, including all collision detection, etc.).

I left it running overnight, thinking it was a stupid idea, as the bikes were all over the track, not getting anywhere close to finishing the race.

The next morning, I came in to work, and discovered that the AI had found a couple of holes in the track's collision modeling, and were universally exploiting that bug as a shortcut.

I never ended up using anything from that experiment, but a few people got really excited about the idea of using my technique as some kind of automated testing procedure. I tried to explain to them that it's not like a brute-force search or anything, but it was an extremely amusing experiment.

As far as pathfinding - Void War was a bit trickier (as simple as it was) as it was a free-form, 3D navigation environment where ships had to contend with navigating with newtonian-style physics (which required turning not just in the desired direction of acceleration, but combined in such a way that it also canceled their previous movement vector) , and to point in the direction of their target. And avoid those trying to shoot at them. And to try to "miss" believably, as the Newtonian physics made target prediction and leading incredibly easy.
_________________
Tales of the Rampant Coyote - Old-School Game Developer talks Indie Games, RPGs, and the Games Biz

[AaronWizardstar]

I'm pretty sure the AI will be hard no matter what I do, so I'm probably going to wing it when I get to it. At the very least, my AI might be some combination of "attack the closest enemy you can do the most damage to" and casting spells at random.

I also decided to go with action points in the end. Lets hope it works out.