What Makes a Campaign/Scenario Good?

[IskatuMesk]

As one who actually enjoyed WoL's gimmicks, I'm not sure I understand the difference.

The line in the sand is drawn between things added for the sake of being there (the fire wall, the lava, the worms that arbitrarily spawn units) and the things that actually contribute to the core gameplay (the riptide generators, the trains to an extent).

The lava and firewall are gimmicks because neither of them make sense from a lore or writing perspective, already defining them as elements forced into the gameplay for the sake of just being there. Their execution is poor and their contribution takes away from the core gameplay.

The elements provided by the firewall, which is to lift and move away from a killzone, make no sense in the context of the writing (that is most certainly not what will happen when a star is going supernova near a planet), and the execution is flimsy, with the wall looking like a cheap wc3 particle effect slowly moving across the map, leaving undamaged terrain in its wake. It's forced, ugly, and inelegant. The effect is lost on the player, who now only knows about a moving killzone. If this was executed better it would be less of a gimmick and more of an element. It's totally outside your control and was added in at the very last minute. There are better ways to do this if you want to do it. You kind of answered your own question, really. And WoL's writing is so bad that it dug a grave under the barrel.

The riptide generators function well mechanically because they give the player something to interact with.

The lava is even worse because its only function is to arbitrarily force you to move your scvs. No other purpose. It adds nothing. You don't interact with it. It's no better than the redundant "macro mechanics" that blizzard forced into the game to bloat APM because the game requires so little.

I don't think you've ever played a campaign with challenging AI yet, because quite simply there has been none. And ladder is a terrible comparison because ladder is an awful gameplay experience entwined with a dozen different things a campaign is totally free from.

The key with things like gimmicks is to avoid making them forced. The fire wall, the lava, they are both forced. They were added because the interns didn't know what to do with the maps. The riptide generators felt more natural and made more sense overall, they actually felt like a component of the map. The fact you compare things like limited buildspace to these makes me wonder if I am making my points clear enough.

And as I never explained the lore in its entirety for 0x03 I don't think you can say anything about its writing. I explained enough that the limited buildspace made sense. Maybe you should watch the video again. That map was unfinished.

/e

I will never know if 0x03 was fun or not, or if it felt too gimmicky. What I had was a proof of concept, not a playable map. I showed it because I knew the campaign would never be finished. I'm flattered that you think it's as good as an official map published by blizzard, though. Or maybe they should be insulted.

[Gradius]

but if you look closely, you could see that every single other elements, such as the AI, the new units, placement of bases, and even the tactical AI were tuned to the exact specification of the map, the Medusa for example, being a heavily stylized version of the Siege Tank, has it's strength and weaknesses heavily exaggerated to become a major set piece in the map's design, it could deal alot of damage, but since battles took place in such confined spaces, it could also ended up splashing more of your units/buildings than the enemy, and thus the placement of Medusas became important, and the placement of Siege Tanks to support the Medusas is even more....

The Medusa splash damage kind of goes along with fighting in a tight space. Sure, the gameplay will be better if you're creating a new mod and tuning all the units & AI specifically for the map. But what about the campaigns, such as Blizzard's, that have to stick with the units that were designed for multiplayer?

Then let us go back to WoL, see the lava? It's a gimmick as it serves too little purpose rather then to be a very bad multitasking challenge, the only thing you have to be doing when the lava rises is lifting up all your expos and SCVs and run to high grounds. It does not make the map more interesting, it does not add in more depth or more varied, it actually decreases variety and depth because you're forced to build only some specific gimmick units like Reapers, which also are poorly implemented.

The ability to destroy other zerg that were chasing you just before the lava rose, especially the brutalisk, was enjoyable. It turned the lava into a tool, not just something that was there for no reason. If I was too far away from the base, I enjoyed making it to a solitary island just before my units were engulfed in lava. I also enjoyed abusing the high ground mechanic as my reapers jumped over a cliff with their awesome jet packs and got away from zerglings/baneling at the last second.

The real challenge of the map itself is the practice of maintaining multiple expansion against the Zerg, and trying to do so felt even more contrived because of the second gimmick: mining X number of minerals. That gimmick limits your unit building EVEN MORE and forces you to build EVEN LESS units, further decreasing variety and depth. My friend is miraculously, able to complete the map on Normal by building no more than 2 medics, he simply abuses the living shit out of the 4 Reapers given to you at the beginning and nothing else... This is basically how abusable the map can become, all thanks to it's very flawed design which the developers didn't put much though into.

So your friend focused on micro instead of macro - that's not a design flaw. The point of the mission was to use the new reaper unit that ended up being "tuned" to the specific map. Clearly maintaining the multiple expansions was not so hard if your friend just used Reapers (the point of the mission). The missing minerals gimmick simply forces you to make the most out of your units as opposed to going through the traditional route of massing up marines & medics & steamrolling the map.

The key with things like gimmicks is to avoid making them forced. The fire wall, the lava, they are both forced. They were added because the interns didn't know what to do with the maps. The riptide generators felt more natural and made more sense overall, they actually felt like a component of the map. The fact you compare things like limited buildspace to these makes me wonder if I am making my points clear enough.

And as I never explained the lore in its entirety for 0x03 I don't think you can say anything about its writing. I explained enough that the limited buildspace made sense. Maybe you should watch the video again. That map was unfinished.

/e

I will never know if 0x03 was fun or not, or if it felt too gimmicky. What I had was a proof of concept, not a playable map. I showed it because I knew the campaign would never be finished. I'm flattered that you think it's as good as an official map published by blizzard, though. Or maybe they should be insulted.

Again, let's forget about the writing for a second for the sake of discussion. Your maps were great, but my question is more of how to introduce what you were going for without radically changing the core RTS gameplay & creating a new mod. I agree that the firewall was one of the worst gimmicks in WoL, so what would you do to achieve the basic gameplay of the supernova mission without creating a new mod or introducing gimmicks?

[IskatuMesk]

Writing is the heart of a campaign. The elements all must make some degree of sense from a writing perspective and fit with the story and universe. They are there, after all, to give the world life and give the story life. We can make compromises for the sake of gameplay, but preferably nothing out of the ordinary.

You talk about killing the zerg with lava, which I knew you were going to mention, but the problem with that map was that the zerg were wholly non-threatening after the first quarter other than their annoying harass early on. If the zerg were much more dangerous, instead making you rely rather heavily on abusing the lava (or having ways to control where the lava goes), this opens up some more options.

so what would you do to achieve the basic gameplay of the supernova mission without creating a new mod or introducing gimmicks?

Well, for one, building stuff on a planet getting vaporized by a sun probably would result in failure all around. That and the supernova mission was based on artifacts, and... well, yeah.

Putting that aside, I would melt the terrain gradually. So instead of a firewall arbitrarily moving across the map, portions of the map would heat up and melt as the intensity of the atmosphere rose. First, doodads and stuff would set on fire, then the terrain would glow and stuff would start taking damage. Pools of lava would form, stuff would start setting on fire, and gradually the map would be filled with these "void zones" that continue to grow. How to choose where these void zones appeared would be the most difficult part of balancing the map. If you had them random it would turn into a mess pretty easily. You'd also have to make sure the AI doesn't throw themselves into the fire and die pointlessly.

This would result in different, albeit still predictable, gameplay from supernova. The void zones would destroy resources and close off portions of the map after a while. If we say that the protoss base is on the dark side of the planet then we have a bit more lenience on placing the zones, where we would end up with something a bit like the firewall's horizontal progress but in a more realistic and plausible fashion. Instead of having this killzone we have a bunch of pockets opening up and growing across the map gradually becoming more and more inhospitable until entire regions are a slagged ruin getting vaporized.

Also, the lighting settings for the map. It's night time, wuh? The sun produces light, it's exploding, and the map is dark? No? You can do way more than that with sc2. Gradually change the lighting and bloom settings over time. Have pools of water that get vaporized, set all the destructible doodads on fire. Make the world alive.

Of course, this would take more effort to do than a cheap particle effect with a kill region attached to it, so I wouldn't ever expect this kind of thing out of Blizzard.

If we moved away from the whole sun exploding deal and just wanted the gameplay, this is easily achieved with advancing zerg bases that take over the map as you try to escape past some other obstacle.

[GnaReffotsirk]

I don't know much about science, but I strongly feel that supernova event was unrealistic. I mean, the effects would be planet wide and simultaneous.

[Lavarinth]

I don't know much about science, but I strongly feel that supernova event was unrealistic. I mean, the effects would be planet wide and simultaneous.

Slowest supernova I've ever seen.

[Pr0nogo]

So, we've talked a bit about what makes a campaign bad. What about what makes a campaign good?

When I was working on a WarCraft III project, I distinctly remember wanting to take advantage of every opportunity to surprise the player - pleasantly, or otherwise. For example, when I was implementing a new race of demons with completely different mining mechanics and gather-able resources, the player could naturally only build maybe one or two military units in the beginning of the campaign, as they had to become accustomed to the massive amount of changes to the base formula. One of the things I did to add to the 'surprise' factor was edit the tooltips for general upgrades whenever the player was able to train a new unit, adding in the names of each new unit as it was introduced. If you didn't know what the new unit was, why should an upgrade's tooltip give such a thing away? It shouldn't. Sure, this is a lot of blab about a small touch, but stuff like this adds to the immersion - the player doesn't know anything that they don't need to know, gameplay-wise, so that they can focus on the story. Otherwise, I feared that they'd sit there, wondering what the hell a Hell Wyvern was.

I suppose the main point of that blather was all about confining unnecessary information to a minimum. The goal of a campaign generally isn't to overwhelm the player with dozens of new gameplay options and a crazy amount of textual direction flashing across his or her screen. The goal of a campaign is to deliver a story through a series of playable stages. That's why it's never good to be overly pithy when you're writing tooltips for abilities, units, and the like. You have to present what you're going to present in a way that does that feature justice, but also lets the player explore some uses for themselves. If I were to introduce a new Protoss siege vehicle in a Brood War mod, I might have some dialogue about how such engines of war might be most useful against masses of enemy ground units, but then have another function for them that the player might only discover by accident (or if they're playing intelligently). This allows the player to feel truly successful and removing the possibility of them attributing their victory to a textual tip or some too-helpful dialogue.

So, don't detract from the immersion with too much information, and let the player have their own victories once in a while. I probably should have just written that, instead.

[IskatuMesk]

So, we've talked a bit about what makes a campaign bad. What about what makes a campaign good?

A sense of accomplishment alongside a good story. No gimmicks, no bullshit, nothing that derives from the core experience of our objective. To attain immersion, to build a sense of meaning and progression, and to establish that progression tangibly in decision-making and milestones.

To overcome a challenge genuine, to improve oneself, the player will feel accomplished and feel as though his efforts were meaningful. This is why an easy campaign does not work.

To have beaten the odds in a manner that felt like the player was in control, that he improved as a player to overcome the obstacles, to feel refreshed every mission. This is why difficulty based on weighting (e.g. stat scaling) does not work.

[Falchion]

The key with things like gimmicks is to avoid making them forced.

Well, there's always the difference between making gimmicks, like Blizzard does, and making gimmicks that are well-executed and well-conceived.

So far, the only missions in WoL that I've liked were the ones that involved little to no gimmicks at all, like the first Tychus mission (Smash and Grab), the pirate-world (Cutthroat), the Tosh tug of war (Breakout), and Haven's Fall. Normally I'd add the Maar mission of Zeratul's and the Steal the Odin thing, but there are things quite screwed with it.

The first Tychus mission, which is blazing through the Protoss lines in the bottom while they're busy with the Zerg is decent at best, the idea that having the Zerg attack your base time to time contributing to the flavour, but the problem is none can react until you do something. The same with the Odin thing.

The Tosh tug of war inspired me in making the first mission of my former campaign because of the crude fundamental in it: Your ally's army winning or losing depending on how you helped out. Of course, my Tug of War mission was a bit more complex than Blizzard's making because it involved a limit vs limitless army, upgrading AI, heavy defenses and such.

The Odin mission is screwed along because the Dominion, so much like the Taldies in WoL, acts like there isn't any AI, not reacting at all nor thinking about it. When I poke the most guarded bases with my cloaked Wraiths, killing Siege Tanks, Bunkers and heavier stuff, it doesn't build an Orbital Command, it doesn't build any Ravens or Missile Turrets, it doesn't do nothing! If I was to assault the base, it doesn't even react building more units and fortifying further. The AI doesn't make any sense at all.

The Maar mission has two problems: The boss design and the weak AI of the dark toss. The toss throws that little bag of runts at you, which you can kill easily, and you can easily neutralize Maar's power with the High Templar's Feedback. Sure that Maar's capacity to lift stuff, one-shot buildings to death and stuff are plausible, given he's a godlike hybrid, but sticking them to the energy bar is an unforgivable fuck-up, because he's abusable. In order to make him more menacing, you'd have to give all those abilities for free.

So far the only mission somewhat free of gimmicks were Haven's Fall, Media Blitz, Tyrador and the rift generator thing. The rest is about involving senseless active action upon senseless active action without a moment's rest.

[IskatuMesk]

That's because their AI is just a series of auto attack triggers and that's it. They don't even make use of melee defense functions or anything. Like I said, WoL is barebones and gimmicks. Sc2 is capable of way more then what Blizzard did with it, which is sad all in its own.

[GnaReffotsirk]

Regarding AI, what should it do besides the basic to make it interesting?

In terms of campaign/scenario of course and not those TD and melee maps, etc.

[IskatuMesk]

The basics; some degree of micro, different attack routes, defends, rebuilds its base, expands if applicable, different unit compositions (more interesting unit compositions at that, as I touched on in the GEC, it's easy to make the AI spam a-move units it's good at using, but effort should be made to make stuff like casters viable), stuff like that.

What I would aim for; at least in wc3/starcraft 2, the AI needs some degree of intelligence in how it handled units. With the Raven tactical AI, it reacts accordingly to the strength of enemy units nearby, plants drones, and very aggressively uses missiles. Unlike the vanilla raven AI, which if you gave it 40 ravens a single missile turret would probably blow them all up.

Through custom units, especially things like bosses or major units, tactical AI would be written to take advantage of their abilities. Alternatively, data can be constructed in sc2 to make a sort of requirement tree, which would simulate AI. These are both preferable over triggers. In BW the AI is entirely hardcoded, so if you're making custom units, you shouldn't change the spellcasters from their base id's unless you want the AI to never cast spells.

The idea is to minimalize the dependance on triggers. A B&D mission should only have triggers for cinematics, dialogue, objectives, the AI attack waves, and stuff like that. With sc2, everything else like tactical AI can be done through libraries and data. This will make systems the most flexible in unpredictable play environments.

Sc2 has the advantage over wc3 in that you do not have that idiotic captain system making things a mess. So, you can have much more complicated terrain than wc3. This, plus bullies, lets you "fake" the AI taking high ground or putting defenses in "smart" locations.

Let's say I employ the mission where you are trying to move your base away from an advancing zerg hive, while attacking an enemy terran. I would do the zerg base construction through bullies (only reliable way outside of galaxy). Regions would be segmented by time and position. So if the Zerg activates trigger D which says, "push this platform", it will add some waypoints to attack waves or change the attack waves entirely to become way more aggressive. Meanwhile, it will activate bullies in that region. As I discovered, it's most effective activate and then deactivate and then reactivate bullies, because lol blizzard and programming.

The end result is that the Zerg will keep aggressively trying to build your preplaced structures and, if applicable, any units you designated as bullies, and try to stick units in those spots. That's how I did the tank pushes, bunker rushes, and "lockdown" ordinance for 0x03. All of those huge sprawling terran bases largely start off as 1 command center and a engineering bay or so. All of the ledge abuse and stuff is done through bullies. The AI isn't really any smarter, and with the way the triggers are designed they could be abused, but they "seem" smart. Mechanically, it's very easy to set up.

[Falchion]

Now that it made me curious, is the pathing for AI in sc2 as stupid as it is in wc3 or it's, thank god, a bit more smarter than that?

[GnaReffotsirk]

Now that it made me curious, is the pathing for AI in sc2 as stupid as it is in wc3 or it's, thank god, a bit more smarter than that?

I think it's better. Much butter, AFAIK.

Through custom units, especially things like bosses or major units, tactical AI would be written to take advantage of their abilities. Alternatively, data can be constructed in sc2 to make a sort of requirement tree, which would simulate AI. These are both preferable over triggers. In BW the AI is entirely hardcoded, so if you're making custom units, you shouldn't change the spellcasters from their base id's unless you want the AI to never cast spells.

The idea is to minimalize the dependance on triggers. A B&D mission should only have triggers for cinematics, dialogue, objectives, the AI attack waves, and stuff like that. With sc2, everything else like tactical AI can be done through libraries and data. This will make systems the most flexible in unpredictable play environments.

Sc2 has the advantage over wc3 in that you do not have that idiotic captain system making things a mess. So, you can have much more complicated terrain than wc3. This, plus bullies, lets you "fake" the AI taking high ground or putting defenses in "smart" locations.

I have some minuscule idea about Tactical AI data, but how far it can affect AI, zero. I thought this was just for things like using abilities when certain conditions are met. Is it possible to have this so that each unit, for example, would look for a fellow unit of his type and when they reach a number of 10, look for targets in the entire map with a certain range of conditions? say, look for mutalisks 0 to 10 units.

Let's say I employ the mission where you are trying to move your base away from an advancing zerg hive, while attacking an enemy terran. I would do the zerg base construction through bullies (only reliable way outside of galaxy). Regions would be segmented by time and position. So if the Zerg activates trigger D which says, "push this platform", it will add some waypoints to attack waves or change the attack waves entirely to become way more aggressive. Meanwhile, it will activate bullies in that region. As I discovered, it's most effective activate and then deactivate and then reactivate bullies, because lol blizzard and programming.

The end result is that the Zerg will keep aggressively trying to build your preplaced structures and, if applicable, any units you designated as bullies, and try to stick units in those spots. That's how I did the tank pushes, bunker rushes, and "lockdown" ordinance for 0x03. All of those huge sprawling terran bases largely start off as 1 command center and a engineering bay or so. All of the ledge abuse and stuff is done through bullies. The AI isn't really any smarter, and with the way the triggers are designed they could be abused, but they "seem" smart. Mechanically, it's very easy to set up.

No idea about bullies. I thought they were just a flag for the AI to rebuild or retrain?

I've only gone so far with touching the AI during the beta. This was the time when GreenTea was so young, non-existent even (not yet). We were using galaxy back then, and I have no clue what's the true maximum potential of SC2's AI engine.

[IskatuMesk]

I've never touched galaxy, so I don't know the absolute extents of Tactical AI. But I think it could do what you wanted, you'd just have to have the AI range very high and have some way for the units to communicate to each other (at the very least, you could have two units, both identical mutalisks, but one in which is a leader and the rest followers). I would just use attack groups for a group like that.

Tactical AI determines stuff like how the unit acts. In Jade's provided Raven AI, it assesses targets on a provided weight for when to use defense drones. For some very powerful long-range anti-air units, like my planetary fortresses or Rhino tanks, I added a strong weighting to encourage DD usage. But for a single mutalisk, a defense drone is a bit of waste. The tactical AI also determines the placing of hunter-seeker missile and turrets. I believe it also tells the raven to micro away from a target when a missile is launched.

I believe it is conceivable to build things like dancing and strafing vikings with tactical AI (I planned to give my vikings and independent auto attack turret for their air weapon, so their AI would always try to circle or be at max range), dancing marines, stuff that avoids AoE attacks, things like that.

Bullies >

I use bullies to control all base construction. Blizzard does this a single time in their campaign, for the Tosh AoS map, to control the raynor base construction. My tests have yielded that the AI build commands in the triggers are broken and unreliable. So what I do is I pre-build all of their bases and tech on the map, in the editor, but set them to be non-spawning. Then I activate bullies in the specified regions (usually tech and construction first, then defenses), segmenting their construction based on trigger timings and stuff.

Bullies need to be activated, and then deactivated, and then reactivated in an infinite loop to keep them building. It's not as reliable as I'd like to have, but it's as good as you can get. They'll also build any units you've set up this way, and place them in those locations.

As an example, let's say we have a tankable ridge above an expansion. For a basic trigger application, it's easy to tell the AI the player has a CC here. Then we can activate bullies in the region on this ledge that contains non-existent tanks. The AI will build these tanks and bring them to that ledge, and siege them. We've now successfully faked an intelligent ledge camp. We can string attack waypoints into this trigger as well based on switch or boolean conditions in the attack setup, so that the AI will also bring in transports full of units to this base. I believe it's also possible to tie all of these triggers to a condition that they've physically seen this CC (using a LoS or range event that sets up a switch to make sure it fires appropriately).

I'm very chobo with triggers, so I probably do things a lot more complicated than they need to be. But there's a few ideas.

Oh, and the pathing is like 69x better than wc3.

[GnaReffotsirk]

Tried looking into AI, specifically Bullies.

I've noticed it's possible to just create a trigger to "Create bully" at a predefined point or Function defined point. The AI will then produce and send this unit over. SCVs were the once I tested and sure enough the AI made 10 scvs for a trigger that says something like,

Add New Bully
AI - Adds a new bully of type UnitType at location Point for player 2 with rebuild count RebuildCount.

I set the unit type as scv, then the point, and rebuild count of 10.

What the ai does is actually create all these 10 scvs, pull them back to a CC and start mining.

Is this supposed to be how Bullies work?

Anothing thing, if the AI cannot complete a bully, sometimes the AI wont pursue creating other bullies. This must be happening due to one trigger issuing an order, which puts the unit into Script Controlled. Being script controlled (enabled) the AI will relinquish control of that unit. Meaning it will not include it in AI stuff and will have to be purely commanded through script/triggers. Disabling script control though gives it back for the AI to do as it wishes.
edit: Sorry, had it on a for loop so...

Anyway, I remember the AI building prerequisite stuctures like depots ovies etc when a wave is defined. Is this still true?

Also, Set Defense Radii makes the AI respond more to player attacking. Presence nope.