KotoR 3: Ideas and Suggestions

[Jack the Ripper]

You just want different animations?

<{POST_SNAPBACK}>

 

Cosmetic changes would be a good term for part of what I want, yes. In the sense that I'd like more fluid combat animations, wherein characters are almost always making contact by attacking or counterattacking, with strategic moves like, force push or flurry, incorporated into the combat in a more convincing way. The suspension of disbelief is very important to me, and I rolled my eyes when I saw my character have a lightsaber poking through his model six or seven times as he performed force lightning because he dropped his saber behind him. A better way to do this would be to continue having the characters fight, but the casting character uses his saber to block and parry in one hand while the other reaches out to the zap the opponent, or to have the sabers lock together for a moment in one of those moments where sparks are flying, and zap the other character to break the lock. Something akin to that.

 

But I also want better AI and better camera angles, so if I want to sit back and watch the fight, waiting and hoping for a positive outcome, I can be left anxiously entertained as I spectate. I lost count how many times I had to press the attack button, lean back in my chair and sigh as I waited for yet another damn Sandpeople ambush to go down so I could move on another twenty steps and get attacked again because I forgot to put their clothes on. Fighting quickly stopped being fun or entertaining, and started being sort of a nuisance that held up the story, about two hours into KotOR.

[thepixiesrock]

Oh, ok.

 

 

 

Wait, do you just want different animations?

[Jack the Ripper]

But I also want better AI and better camera angles, so if I want to sit back and watch the fight, waiting and hoping for a positive outcome, I can be left anxiously entertained as I spectate. I lost count how many times I had to press the attack button, lean back in my chair and sigh as I waited for yet another damn Sandpeople ambush to go down so I could move on another twenty steps and get attacked again because I forgot to put their clothes on. Fighting quickly stopped being fun or entertaining, and started being sort of a nuisance that held up the story, about two hours into KotOR.

 

[thepixiesrock]

Oh, ok, sorry.

 

 

I didn't know you just wanted new animations.

[GhostofAnakin]

I guess I'm just not understanding what you actually want then. I know you've tried to describe it, but I'm not getting a mental picture in my head about how what you want can even be accomplished while still keeping the game a RPG at its core.

[Jack the Ripper]

Oh, ok, sorry.

 

 

I didn't know you just wanted new animations.

<{POST_SNAPBACK}>

 

S'okay.

 

I've never sat down and done the math for any table top RPGs, I don't know the formulas or why people are so militant about preserving them. They're hidden in KotOR and takes some menu hunting to find them, so I have no opinion on changing them, and wouldn't presume to say that they should be changed.

 

In fact, if I have anything to say about them, it's that I like how the random outcome of the formulas effects what animation is displayed. For example, the formula that tells whether or not a character is successfully hit defining whether or not that character should visually react to it, or should use a ducking away animation instead. I think that sort of thing could be applied deeper to define in what way a character attacks, and in what way a character defends themself from an attack.

 

I don't care whether the D20 system is preserved or not, I just want the animations to be more robust to reflect the system's outcomes in more diverse ways. And in ways that don't put me to sleep, or make me say to myself, "Dammit! Not another spider thing/kath hound/sand people/low level sith fight! This. Game. Is. Never. Going. To. End!"

[Judge Hades]

Sounds to me that you don't want a CRPG. Go play a Star Wars action game and leave the CRPGs to us.

[alanschu]

Actually it is more balanced without them.  Epic Rules and going above 20th level is really broken in the d20 System.  You really need to read the rules system to understand but take a look at HotU on how epic levels work in the CRPG format.  It was really way too overpowering.

<{POST_SNAPBACK}>

 

 

That's exactly what I thought too. Which is why I was surprised with your comments that Epic rules were needed to balance the game.

 

Epic rules in D&D I think are dumb. Hey, look at me, I can swim up waterfalls and run along clouds.

[Jack the Ripper]

I guess I'm just not understanding what you actually want then.  I know you've tried to describe it, but I'm not getting a mental picture in my head about how what you want can even be accomplished while still keeping the game a RPG at its core.

<{POST_SNAPBACK}>

 

It's fairly simple. I want a combat system that visually looks exciting and more dramatic. There are two things, aside from the polygons and all that, that really take me out of the story and reminds me that KotOR is just a videogame.

 

One, the poor animation in general being peppered throughout the game. Like the super conservative dance animation or the horror animation, which is what I do when I'm trying not to throw up after too much Irish coffee. And while the lip synch is pretty good, the conversation animation is pretty bad. Flailing arms around too slowly and deliberately, greeting people like Sultans, rocking shoulders back and forth. They don't move like real actors. They don't stand up and sit down in chairs or lean against walls, or use their environment at all for that matter. There's maybe twelve animations in the game for what was it? Ten thousand lines of dialogue?

 

Two, the combat animation, which looks more like waltzing than a choreographed fight. I want to see constant, exciting, suspenseful action. I don't want to take control of the fight away from the player so he can no longer command the action, but I do want doing that to be optional. If you never touch the controls in KotOR or TSL, your characters very likely will die when they fight higher level characters. If the AI was improved to the point where your party members all fight as if they were controlled by humans, then the player having to command the action is no longer necessary. That way, players that are more interested in the story than manipulating the d20 system don't have to deal with monotonous combat, and will have something interesting to watch on screen.

 

A dramatic camera system as I have described previously would also help to keep things interesting, so players don't have to simply observe from the limited angles offered by the Matrix cam, but can watch fights from interesting angles picked by professional cinematographers, or at least picked by the people who made the animations. Watch a fight in a movie. There are very few fights that only use one camera. It's constantly moving from image to image, cutting away from one angle and going to another. The Odessa steps scene in Battleship Potempkin would be a good example of camera techniques I'd like to see used.

[Jack the Ripper]

Sounds to me that you don't want a CRPG.  Go play a Star Wars action game and leave the CRPGs to us.

<{POST_SNAPBACK}>

 

Why does it sound that way?

[GhostofAnakin]

I sort of got that part. I meant I don't understand how that would work in the concept of a cRPG.

 

As best I can tell, cRPG combat is based on rolls and stats and numbers. I don't see how your suggestion can be implemented to fit smoothly for a system which, at its core, uses "dice rolls" for each action to determine hit or miss.

[alanschu]

What you propose will absolutely explode the development costs.

[Surreptishus]

What you propose will absolutely explode the development costs.

<{POST_SNAPBACK}>

That thought occured to me but to give hime a chance to explain i imagined this to be in development fantasy land - infinite money and time.

[Jack the Ripper]

I sort of got that part.  I meant I don't understand how that would work in the concept of a cRPG.

 

As best I can tell, cRPG combat is based on rolls and stats and numbers.  I don't see how your suggestion can be implemented to fit smoothly for a system which, at its core, uses "dice rolls" for each action to determine hit or miss.

<{POST_SNAPBACK}>

 

Well, I described the KotOR combat as this: Attack. Return to battle ready animation, wait. Attack. Return to battle ready animation, wait. Critical hit. Return to battle ready animation, wait. Flurry. Return to battle animation, wait. Etc. Each time the action pauses, it's only for maybe a fraction of the second. But that disruption of the pace of the fight is really distracting and takes the sense of urgency and intrigue away from the combat. Yuen Wu Ping would gnaw his own arm off before he choreographed a fight like that.

 

Now this is what I would like to see: Attack. Attack. Critical hit. Flurry. Etc.

 

How? Simply fill in the space between moves with interesting and very varied filler animation. Sabers continuing to smash into eachother, one opponent taking steps back while the other repositions, a saber lock, a couple of pithy lines of dialogue being quickly traded. Or aiming or reloading or taking cover animations for blaster users. Things like that. And if two opponents are both being controlled by AI, then there's no reason to compensate for our human slowness, and the filler animation wouldn't be necessary. All that they would need would be a lot of transitional animations, like how you can go from one martial art style to any other style smoothly in Jade Empire, only now the KotOR characters would be going from one type of an attack to a type of defense to a totally different style of attack, constantly keeping the action going.

[alanschu]

What you propose will absolutely explode the development costs.

<{POST_SNAPBACK}>

That thought occured to me but to give hime a chance to explain i imagined this to be in development fantasy land - infinite money and time.

<{POST_SNAPBACK}>

 

 

 

Either that or they'd have to do some sort of algorithmic process ala Spore. Though I'm not too sure how that game does it either, nor how easy it is to program or anything like that.

 

Even then, I doubt you could get the algorithms to properly apply rules to people so that their animations would look proper. I bet you'd get wierd results.

 

 

 

Though I'm not so sure because he is tossing around words like "simply" and "just do this" a bit liberally.

Edited January 22, 2006 by alanschu

[Jack the Ripper]

What you propose will absolutely explode the development costs.

<{POST_SNAPBACK}>

 

Mmm, I was going to reply, but I'd like you to expand on your thoughts before I do, please.

[Judge Hades]

Sounds to me that you want Jade Empire style combat for KotOR 3. That might look good on a console but remember that KotOR series is a multiplatform game. What works on the console may not work on the PC. They need a system that works on both.

[alanschu]

What you propose will absolutely explode the development costs.

<{POST_SNAPBACK}>

 

Mmm, I was going to reply, but I'd like you to expand on your thoughts before I do, please.

<{POST_SNAPBACK}>

 

 

What you are asking for would require many, many, many, many, MANY more animations in order for it to work. Art content has already resulted in an explosion of development costs, which is part of the reason why a game like Spore uses algorithmic patterns for its animations. But it's being used to animate non-human creatures based on their size, skeletal structure, etc. Stuff like this is also more expensive processor/RAM wise.

 

Furthermore, the AI required would also be very expensive. Not just dollar wise, but in system requirements. You'd be requiring the AI to do some sort of search algorithm to determine the best course of action to do at any instance. And given you want these actions to be fast, it's going to be performing this very frequently. But the thing is, for the AI to actually be good, it has to anticipate what it thinks YOU will try to do, as well as predicting how results we look into the future, based on a whole bunch of conditional probabilities that would become murderous for the CPU to try to calculate in a quick manner. Even if you store the probability distribution table in memory.

 

 

 

Finally, this is only concerning a 1-on-1 fight. What happens if there is more actors? Even if it's all 1-on-1 fights, it still causes more AI searches to be performed to determine the best action in every situation (I couldn't even imagine the heuristic involved to determine the cost of each action).

 

What about N vs 1 fights, where N > 1. Now you have to have your AI determine what to do when fighting MULTIPLE targets (which is sick), and create even more animations to take care of that. Trying to mimic the Darth Maul vs Obi-Wan and Qui-Gon fight would not be easy to do in a video game.

 

And how does the AI react in a multiple enemy situation? There's been no success on finding an optimal solution for a battle with two sides, both sides having more than one guy. So if you're leaving things up to the AI for determining who it fights, it's not unlikely that it might do something "stupid" based on its target selection algorithm.

 

 

 

If you'd like me to go into more detail, I can.

[Jack the Ripper]

Sounds to me that you want Jade Empire style combat for KotOR 3...

<{POST_SNAPBACK}>

 

I don't think you're reading my posts very well.

 

Alanschu, please continue.

[alanschu]

In which aspect in particular? I'm also about to step out and get something to eat, so it might be a bit.

[Jack the Ripper]

In which aspect in particular?  I'm also about to step out and get something to eat, so it might be a bit.

<{POST_SNAPBACK}>

 

You said that you could continue, so please do. I've been reading these forums for a few days and don't plan to stop checking them, so whenever you get around to it, I'll reply in full.

 

Just to start off with one of my major points, this is a sequel to a Star Wars game, and a game of the year, and will likely be available on a next generation console like the 360. Production values will of course be expected to increase, and in all areas of the game. For example, I certainly don't think that the public at large would be satisfied with the quality of the current models, having seen what games like Oblivion are capable of rendering, so I don't think pleading for a smoother combat system and better animations is out of the question.

[moreKOTORplz]

Sounds to me that you want Jade Empire style combat for KotOR 3.  That might look good on a console but remember that KotOR series is a multiplatform game.  What works on the console may not work on the PC.  They need a system that works on both.

<{POST_SNAPBACK}>

 

i hated jade empire, most overrated combat! if K3 goes that route i'll be very pissed

[Judge Hades]

Overrated or not it got the job done and did it well enough.

[alanschu]

You said that you could continue, so please do.  I've been reading these forums for a few days and don't plan to stop checking them, so whenever you get around to it, I'll reply in full. 

<{POST_SNAPBACK}>

 

Okey doke then.

 

 

Basically, performing a meaningful AI search is computationally expensive.

 

Lets start at the highest level, which is the current situation in the conflict, with a 1-on-1 fight. At this point, the AI can choose any action it wants, be it flurry, attack, heal, etc. There are a variety of ways it can choose on option. The easiest way is to just randomize it. Naturally, this would not result in a smart AI, as it just chooses from a list. Next you could give a weighting to each action, and have it pick from that. This would mean you could have it choose to attack more, but it's still static and not very smart. The most obvious way (to me) is to set up conditional probabilities based on the current situation that both you and your opponent are in. But this is not easy.

 

First you have to determine how you can determine which action is the best course of action. And we have to properly define things. Do you want the AI to think that attacking high with a chop is better, or are you content with just "attack." If you're content with just attack, then you aren't going to get much better than what you have now.

 

As an aside: Even if the "attack high" has no actual difference from a gameplay mechanic, you'll still need a mechanism to select the correct animation to provide a fluid looking motion, as I'm assuming poor animation would be unacceptable. If we simplify it and say we have just 4 areas to attack, which would be high, low, left, and right then you have to take into consideration that each attack location would require at least 4 different animations. This is because attacking high will look different depending on where you last attacked. This is also too simplistic, as it ignores the possibility that your opponent might attack you at any of the 4 locations. Therefore, you'd actually need 8 separate animations, depending on whether you had to defend at any of the 4 previous locations, or attack, prior to your attack. Even then, this still gets repetitive as you only have one animation to choose from for each attack. So how many animations are actually required for it to provide that cinematic nature you want? The situation becomes exponential when you have to consider the fact that a character could get attacked by multiple targetsEnd Aside

 

Chosing whether or not to perform a type of attack depends on the situation, and could be better if you follow it up with something different. This means you have to start performing predictions, all of which have a conditional probability on them (with an increasing variance), and next thing you know your computer is now chugging along at fractions of a frame per second. The situation is worse for each additional character you have fighting, and becomes even worse if you start getting characters fighting multiple characters at once. You also have to make concessions, as you could literally go on forever, and eventually cut off the search telling it to stop. Trying to determine the heuristic for something like this would be a nightmare for a programmer as well.

 

To do something like this, I would think that at the very least, you'd need to have a multicore processor, with one of those cores used just for determining how the AI should proceed with battle. For example, it takes modern computers a moment to determine where it should put queens on a chess-board to make sure that the queens aren't attacking each other. If you have 6 guys parsing through trees of information, calculating the probabilities of the Bayesian network, it's not going to be very fast. But the thing is, you want fast. I am quite confident you aren't happy with the AI taking a second (or more) just to determine what type of action it should do at this moment. This is especially unacceptable if you are looking for sequences as fast as the Obi-Wan/Anakin fight.

 

To do what you want to do, in real time, would probably take computer hardware that hasn't even been designed yet. Maybe a little faster if you restrict everything to 1-on-1 encounters (and not even 5 different 1-on-1 encounters...I'm talkin ONE 1-on-1 encounter at a time).

 

 

 

With respect to the animations, you can either try to acheive them procedurally (might as well add a few extra years on to both development to make sure it mimcs human movement perfectly, as well as even more demand on computer resources), or you can hire even more animators to work on them.

 

The thing is, creating art in computer games is VERY expensive. Hiring even more animators to make sure that they have plenty of smooth, animations that flow nicely into each other, is going to skyrocket that cost up even more. Either that or they can take even more time making the game...which is probably even less acceptable as I doubt they want to go 6 years without selling a product.

 

 

And this is just concerning fights. You also mentioned dialogue animations and stuff like that. Animations don't become cheap. And the more you add, the more testing time goes up exponentially. Especially when you commented on finding a clipping error with the lightsaber as being unacceptable.

 

 

 

Doing what you would like is not cheap. With respect to both money for development, as well as computer resources.

 

 

Overrated or not it got the job done and did it well enough.

 

It's not what he's looking for though. You still selected the attacks you wanted to do. He wants AI to take care of it.

Edited January 22, 2006 by alanschu

[Judge Hades]

What? He wants the game to play itself?