Customization, randomly generated maps and auto-deploy

User avatar
GFX47
Dev
Dev
Posts: 2914

Customization, randomly generated maps and auto-deploy

Post#1 » 15 Sep 2017, 13:57

Here are my plans for the future of the game, I'd like to get your feedback on it.
The different features I'll talk about are tightly linked together, that's why I'm posting it in a single topic.

Customization

I'm currently working on the bot customization system.
It will only affect the visuals of the bots, no impact on the game resolution, no pay to win.

It will be central in the business model of the game.
I'm heavily relying on it to win enough money to continue working this way.

Customization will cost in-game currency and/or real money depending on the content.
The in-game currency will be earned pretty much the same way you currently earn XP points: by winning career matches, you win more in ranked matches, nothing in private matches, etc.

I'd like to make something similar to Rocket League customization system in terms of interface:
[youtube]TcxIFSWXPr0[/youtube]

In terms of content, I'm thinking about different new 3D models, animations, colors, decals, visual effects, sounds, bullet trails, etc.
But it will depend on my capacity to produce such content.

Each bot will have its own customization options.
The goal here is to get the player attached to her bots, like you do with your squad in XCOM:
Image

Naming them, giving them a unique looks, etc.
It means that player will have to compose her team apart from the match deployment process (more on that in the Auto deploy chapter).
The team will have 4 slots and each slot will have to use a different bot class.
Right now it means you will have 1 of each existing classes.
Later, when more classes will be implemented, you will have to make a strategic choice: do you want a healer or a sniper in your team, etc.

Randomly generated maps

We already talked about it here so I won't detail it further but I really want to try it.
I don't believe in static maps, and map specialized AIs, especially on the long run if we frequently add more of them.
All maps would have to be 4 VS 4 though: 4 slots for each team would be randomly placed on the map and bots would be automatically assigned according to the team composition order.

The map generation system could also be used to generate offline missions.
The problem remains to find interesting enemy team composition and AIs.
It could be picked from online matches and anonymized (you don't know who's team it is).
Although it would need some internet connection at some point and a not negligible amount of storage space for AIs (making me reconsider the AI size limit option...).

Auto deploy

The currently existing deployment phase, where you select your bot classes and AIs, wouldn't be necessary anymore.
Player could then deploy X matches at once, automatically redeploy if she wins, etc. (pro features?)
We could also have full auto tournaments: you commit your team once and all the tournament matches can be played without further interaction.

So, what's your thoughts on it all?

ConsciouS-0nE
Autonomous Entity
Autonomous Entity
Posts: 740

Re: Future updates (need feedback)

Post#2 » 15 Sep 2017, 14:30

Give it a go, then well see about it.

Theres nothing much i could say just keep up the good work gfx, youre listining. Looks like your doing the right track.


About your business model, just like as proposed by me and by the others lets give it a hit. Its all your choice its your game.
Im not the best player, but atleast im something.I'm 8
Science is madness! 3:)

User avatar
GFX47
Dev
Dev
Posts: 2914

Re: Future updates (need feedback)

Post#3 » 15 Sep 2017, 14:53

My concern is more about the impacts on the strategic aspect of the game.
For instance, constraining the team size and bots positioning offers some interesting opportunities but limits your options.

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#4 » 15 Sep 2017, 15:36

GFX47 wrote:My concern is more about the impacts on the strategic aspect of the game.
For instance, constraining the team size and bots positioning offers some interesting opportunities but limits your options.


I don't think it limits anything. Currently people have specific map code in their AI and then switch to the main AI block. In the new system this would just be the main AI. People would focus more on improving their main AI instead of finding cheeky ways to cheese on specific maps.

Having a set 4 bot sized team would also remove the "unlucky" moments where you basically just lost because the enemy team had a better positioning.

I'm a big fan of this. I would even suggest putting the customization behind the random maps because I believe that would bring back a lot of people who (kind of) quit the game. Customization is a feature that doesn't really do anything for the current playerbase. And we've also had a lot of cosmetic changes lately instead gameplay changes. Imo it's time for a game update.

Currently when I queue 10 games I get 5 times the same map and 8 times the same player and I will most of the time already know the outcome. In random maps it would always be a new game where I could have a chance.

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#5 » 15 Sep 2017, 15:49

My fear with the 4 bot system though is, that this could become a rock paper scissors situation with the AI becoming irrelevant (mg, sniper, healer, stealth just flat out beats sg, sniper, mg, commander) OR one combination would be superior to every other.

This sounds like a balancing problem and can be fixed by adjusting numbers, but that seems pretty difficult. I would pay attention to not making so many counter classes, or at least make classes that counter 2 classes and are weak against 2.

For example currently the MG is very strong. It only loses against sniper on long range and beats every other bot at every range. This should not be the case. The MG would be in every team combination because it pretty much beats everything else.

User avatar
Kanishka
Skynet
Skynet
Posts: 1421
Contact:

Re: Future updates (need feedback)

Post#6 » 15 Sep 2017, 18:34

Sounds good!
Fixes break an AI more than bugs do. :ugeek:

Gladiabots Off-Topic Chat


My Stats: Kanishka_RN3;Kanishka_MiPad

darktheatre
Script
Script
Posts: 34

Re: Future updates (need feedback)

Post#7 » 15 Sep 2017, 18:37

several skin theme if it could be

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: Future updates (need feedback)

Post#8 » 15 Sep 2017, 18:41

GFX47 wrote:Randomly generated maps

We already talked about it here so I won't detail it further but I really want to try it.
I don't believe in static maps, and map specialized AIs, especially on the long run if we frequently add more of them.
All maps would have to be 4 VS 4 though: 4 slots for each team would be randomly placed on the map and bots would be automatically assigned according to the team composition order.

The map generation system could also be used to generate offline missions.
The problem remains to find interesting enemy team composition and AIs.
It could be picked from online matches and anonymized (you don't know who's team it is).
Although it would need some internet connection at some point and a not negligible amount of storage space for AIs (making me reconsider the AI size limit option...).

As I've said many times by now (and am in the minority for), I don't think the whole "all AIs should have been generic only" thing will be the best for the game. However since I ideas of that for y'all will change, I guess I will ask a more important question: will there be an option to play on the current maps offline? If not I surely think that we will run into issues where either I will be extremely hard to consistently test your AIs or the maps would be forced to be so similar that it would be more stale than current maps.
GFX47 wrote:Auto deploy

The currently existing deployment phase, where you select your bot classes and AIs, wouldn't be necessary anymore.
Player could then deploy X matches at once, automatically redeploy if she wins, etc. (pro features?)
We could also have full auto tournaments: you commit your team once and all the tournament matches can be played without further interaction.

And yes I do think bots being placed beforehand without any sort of set up from the player will be a pretty heavy loss in strategic depth and control over their side of the battlefield

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#9 » 15 Sep 2017, 19:13

mcompany wrote:And yes I do think bots being placed beforehand without any sort of set up from the player will be a pretty heavy loss in strategic depth and control over their side of the battlefield


This game is about making the best AIs, not winning with an inferior AI because you randomly selected a better positioning than the enemy.

Since you don't know the enemy positioning beforehand this is pure guessing and has nothing to do with strategic depth. It's a gamble.

Or are you trying to say rock paper scissors has any strategic depth? lol

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: Future updates (need feedback)

Post#10 » 15 Sep 2017, 19:38

sollniss wrote:This game is about making the best AIs, not winning with an inferior AI because you randomly selected a better positioning than the enemy.

Since you don't know the enemy positioning beforehand this is pure guessing and has nothing to do with strategic depth. It's a gamble.

Or are you trying to say rock paper scissors has any strategic depth? lol

1. Let's face it: besides really dumb setups (a sniper as the close bot on Mind Game), there is no way in the world that an inferior AI will beat another AI from positioning alone.
2. Even with than notion, most of the time enemy's starting position is never truly random because any other position would be heavily disadvantageous (part of the reason why 5.2's rock paper scissors happened and why it hasn't been in play to no where near the effect in the current setup system despite still existing). Even then, you don't even have to make much of a gamble because it isn't often on the current maps that what bot your opponent places where has any relevance to your AI or strategy
3. On the flip side, under a completely random system gives you zero chance to have any clue how you are setup, with no indication beforehand what to expect, and the AIs current know nearly absolutely nothing of how to figure out any sort of positioning of where it is in relation to anything, so under this new system, you are forced to decide to gamble on pretty much everything with zero tips on what might be a good idea, or to play it safe and make an AI guaranteed to be 100% inferior

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#11 » 15 Sep 2017, 19:52

mcompany wrote:On the flip side, under a completely random system gives you zero chance to have any clue how you are setup, with no indication beforehand what to expect, and the AIs current know nearly absolutely nothing of how to figure out any sort of positioning of where it is in relation to anything


Yes, but I don't see how this is bad. In the future we will get things like this.

I think of the game like this: the match is a blackbox, the map and enemy AI is inside the blackbox, I don't know them. What I choose to put in my blackbox is my choice and strategy (AI and bot classes). The AI as to handle everything else on their own.

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: Future updates (need feedback)

Post#12 » 15 Sep 2017, 20:12

sollniss wrote:
mcompany wrote:On the flip side, under a completely random system gives you zero chance to have any clue how you are setup, with no indication beforehand what to expect, and the AIs current know nearly absolutely nothing of how to figure out any sort of positioning of where it is in relation to anything


Yes, but I don't see how this is bad. In the future we will get things like this.

I think of the game like this: the match is a blackbox, the map and enemy AI is inside the blackbox, I don't know them. What I choose to put in my blackbox is my choice and strategy (AI and bot classes). The AI as to handle everything else on their own.

What I'm saying is that you are complaining that under the current system, everything is a gamble because of a rock papper scissors. Under the new system, everything would be ten times more of a gamble because everyone has to fight blindfolded because in a match the AIs almost never has the ability to gather the information you want (and this would be true even with an overuse of tags). So what, the AI has to handle everything? It wouldn't even know what it was handling in the first place. So what that everyone's AI is a blackbox? Even in the current system most of knowing what 1 or 2 opponents do either are stuff that should have been considered since day 1 or are stuff not worth going crazy over because there are so many other potential opponents you could have.

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#13 » 15 Sep 2017, 20:22

I mean, how often do you even change the AI setup?

I set it once for every map when I started the game and never changed it.

In the new system everything would be decided by the randomness and it would be mirrored. So not only would you have more variation in the game, you would always have an equal and fair setup for both players.

Your AI might be the strongest in grandmaster in a assault/sniper vs assault/sniper setup on mindgame, but because nobody plays this we will never know. In the current system there are like 2 combinations that are viable and everyone improves in super small steps. In the new system every possible combination would be played and you might find huge flaws in your AI. Which system do you think is better for a game that is about making AIs?

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: Future updates (need feedback)

Post#14 » 15 Sep 2017, 20:40

Well considering such a theoretical system where you have to play against assault/sniper, I would think that a generic AI would have had to deal with that situation at least a few times (and making anything beyond openings and extended knowledge map specific usually makes no sense), so I'd say that regardless, it wouldn't make which system would be used, the amount of bugs on any decent AI found going through every setup would be minimal, and a good number of the setups would either be unrealistic or irrelevant anyways. Plus on either version you are making an AI and refining it over time is super small steps. From this standpoint, I see no difference either way for the AI's development, except of course that on one side the AI actually knows what game it is playing and on the other side it does not

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#15 » 15 Sep 2017, 21:17

OK so let's take my SYP on the current system. My AI is super specialized I would argue it plays the map optionally. I can't even remember the last time I lost on this map. This would be the prime example of your pregame setup system. If GFX would change one little thing about this map my AI would break. If I change the setup around, my AI would break. My AI is super dumb because it plays according to one single set of rules. It can't react to enemy bots because it has to play optimally and there is only one optimal thing for each bot to do.

Do you really want this kind of AI? It knows everything about the game beforehand, it doesn't care about the enemy. This sounds more like an IKEA manual on how to build the best AI for map X instead of a game where you want to make a better AI than your opponent.

pier4r
Skynet
Skynet
Posts: 3389

Re: Future updates (need feedback)

Post#16 » 15 Sep 2017, 22:09

small nitpick against "it plays optimally":

syp
24.8.2017, 00:19:55 759558 Miojo (2254) 1880 -2 Set your priorities

moreover in about a month you played syp like 15 times. When something is happening so seldom, it is not worth it to optimize against it.
http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

Jeffrey Harpen
Automaton
Automaton
Posts: 122

Re: Future updates (need feedback)

Post#17 » 15 Sep 2017, 22:51

sollniss wrote:My fear with the 4 bot system though is, that this could become a rock paper scissors situation with the AI becoming irrelevant (mg, sniper, healer, stealth just flat out beats sg, sniper, mg, commander) OR one combination would be superior to every other.
Similar here, but my concerns are mostly about assault. Assaults work very well in teams with other assaults. One assault in 4 bots team is not very useful. Sometimes even team of 4 assaults is not enough to use their potential effectively. I don't want them to extinct, so GFX47 please consider a bit bigger teams and stay with the current system (one of each special bots, many assaults).

sollniss wrote:For example currently the MG is very strong.
Funny: I find it the weakest because of its' slooownessss. ;)

sollniss
Automaton
Automaton
Posts: 178

Re: Future updates (need feedback)

Post#18 » 15 Sep 2017, 23:02

We could also just rework Assaults or buff them.

MG does massive damage and wins every 1v1 except against long range snipers.

TheKidPunisher
Autonomous Entity
Autonomous Entity
Posts: 564

Re: Future updates (need feedback)

Post#19 » 15 Sep 2017, 23:05

Customization

Do as you want (i prefer game that focus on game plays more than esthetic but for future income this is one of the multiple possibilites)

Randomly generated maps

We will have to define how random are the different variables (and which ones),
i think it could be a new ranked mode (in parallel with what we have actually)


Auto deploy

Can you explain what you mean by "auto deploy", i see 2 potential meaning:

-You record your AI for each map,
and then when you press play a match, it will directly found an opponent instead of a "deployement phase"
-You record your AI for each map,
and then the game will automatically deploy match (every X hours/minutes...)


GFX47 wrote:The team will have 4 slots and each slot will have to use a different bot class.
Right now it means you will have 1 of each existing classes.
Later, when more classes will be implemented, you will have to make a strategic choice: do you want a healer or a sniper in your team, etc.


it is classified in "customization",so you will have a slot by bot class for customization or you are talking about the 4 slot for radom maps?
If it is the second option, so you reduce the team composition possibilities (atleast for the moment as there is only 4 bot class we will all have to use a compo of Assault+Sniper+ShotGun+MachineGun)
No more choice to replace one (or more) specific class by an assault.

I now it is the same in the majority of online game but balance between class will add a lot of analyse/work to make the game work properly (without a strategy dominating others)

And for auto deploy if i understand it seems to be that with your new proposition:
-You record your AI for each bot slot,
and then when you press play a match, it will directly found an opponent instead of a "deployement phase"


GFX47 wrote:I don't believe in static maps, and map specialized AIs


Sorry but well constructed static map, can give multiple viable strategic choice.

And if we keep them, it will be the possibility to continue to have maps with less or more than 4 bots at the begining

User avatar
0x1de
Automaton
Automaton
Posts: 156

Re: Future updates (need feedback)

Post#20 » 15 Sep 2017, 23:08

GFX47 wrote:Customization
It will be central in the business model of the game.
I'm heavily relying on it to win enough money to continue working this way.

But it will depend on my capacity to produce such content.

Good luck with that. A new skin is fun but not worth money, especially with the current match rendering. You just wouldn't see it.
Also I imagine it would take a lot of time and imagination to come up with new stuff.

People pay for an advantage or to save time.
GFX47 wrote:Randomly generated maps

Random maps (and worse, deployment) means you need a better generic AI. That makes the game harder and more difficult. Not going to pull in new players.
GFX47 wrote:Auto deploy

A formality. I almost always use the same deployment because it defaults to the test set ups.
Saves a few clicks.

Remember that most of the feedback here is from hardcore players. We love the game as it is.

If you want to make money, I suggest something easier to get into (slimmed down ai options), more dramatic bigger robots, and a richer environment (trees to burn, buildings to blow up, etc).

Return to “Devlog”

Who is online

Users browsing this forum: No registered users and 1 guest