XSpire AI Mega Demo

OK, here it is, a large Demo using the XSpire AI
to show 3 types of gameplay.

Deathmatch
Domination
Capture the Flag

It contains 3 Maps
Frank_G´s CTF Level
The Backyard Level from the Downloadpage
and a Level from Jostie

http://www.wasteland.at/XSpire_AI_Demo.exe
32MB

It shows the current Status of the AI, that is still in development, and
will undergo some major changes.

hey damocles very nice work. although this demo shows the limits of a6 block levels. it looks very good but it lags like crazy in some areas. there you also have different mouse movement wich is a very bad thing for a shooter. it should feel the same in every place.
another thing i noticed is that the weapons stick through the walls, you should fix that too

great work, keep it up!

The Level from Jostie is more suitable for older machines,
But with Franks level we wanted to test how a "large" Level works out.
I know that the performance is not good on smaller machines,
although it has got full visibility calculations.
But the other levels should work fine on older machines.

There are lots of things that can be improved, but my main aim
was to get the AI work smoothly and get a good team-gamplay against the AI.

Amazing

That was pretty good. I've only played a CTF match but yeah it was good.

Do enemies prioritize the bot/player with the flag? I didn't really notice a change.

They do take the flagholder with a higher priority when searching for enemies,
but they dont switch to an exclusive "follow and attack the flagholder" mode yet.
I will work on that.

Well maybe it's best they don't...in a way. I mean if they are simulating human players, human players dont usually figure out who has the flag unless they see them.

Hah kinda contradicting myself.

Sure, that's true, nor does every player go chase the enemy flag carrier, but at least one or two should.
Anyways, let's see what Damocles further improves, it's already really good,

Cheers

I really like this demo! The AI is great and hard to beat on the high levels, but what is missing is teamplay! If they would work together and storm towards the flag, the AI would be like any professional game.

That depends, I don't consider a rushing team as being really intelligent. You'd need some good players to guard, others to go for the flag. With good 'ol UT1 the main tactic of the bots was kill a lot and then occasionally play more offensive, but they never all ran away at the same time. Sometimes they did go with 2 bots quite near to eachother to the player's flag. Wether or not their behavior is realistic is also a matter of taste and choice I guess,

Cheers

What is planned, are team positions, such as
Offender (attacking the base, and walking together or with the player)

Midfield-Defense, mostly for capturing enemy flagholders on their way back

And defenders, that take strategical positions, to guard the flag.


These positions are already implemented, but they are not tuned yet.
So the midfielddefenders attack the flagholder, but might run in the
middle of the action to a healthpickup.
They need to somehow project the way the flagjolder will take, and then try to
outrun him.

But the main part in CTF that is missing is the ability to pick up flags, whenever
the teammate was killed. Its just not implemented yet.

But if the bots where too smart, it would not make much fun for the player, and
get frustrating. It heavily defends on the outlay of the map too.
In Franks map you can try to use the side-coridor to outrun the main battle.
I had a more open map, in wich it was really hard for the bots, to score at all.

Anyway, if you want to make a good CTF Map, and want to see it in action,
just PM me.

i wouldnt say i have a slow or old system. i can play actual shader hungry games like ghost recon advanced warfighter fairly good.

ati radeon 9800xt pro, 1gig ram, athlon 2ghz

in this demo there comes no real fps feeling because of the lag from the mouse. sorry to say that but iam a fps player i used to play those games extremley often in past times. you just need to have a smooth mouse that feels the same in every area otherwise you mess up aiming wich is bad. you have to compensate the fps lag with your mouse wich just plain sucks. its heavy fps depending like i said in some areas and you have to move your mouse in another way.

this is no insult ! you are doing a great job with your project and are probably one of the best wdl coders i know. but you need to improve that either by optimizing your levels or code somehow.

good luck, please take this as constructive critisism

This is more of a 3DGS related bug then, since I have no mouselag.
I did have it about a year ago, but then it disappeard, probably due to some system or driver changes.

Do you have a mouselag in the menu too? Meaning, is the mouse not as quick as
on the desktop in responding?
Or is it just in the CTF Level?

I know that the CTF Level really takes some performance,
but thats the reason, why I have also included Josties Level, for higher FPS.

I know this thread isnt about level design so much, But just from the screenshots I see a ton of blocks that could have been models.... and then used blocks with the 'none' flag set to make the shadows. This could increase performance a bunch

I just downloaded, i'm gonna check it out!

Jason

Thanks Jason. I did one of the levels and you are right. I could have done a lot of lamps as models and not from blocks. That will be done in the future. I learned my lesson while waiting 3 days for compiling visibility. In the middle of the process there was a time where the compiler took 7 hours just to calculate one single portal.

Next level will be much better with more models for details and less windows.

The XSpire demo would be much cooler if it had per-pixel lighting.. looks very dated. Of course that's not the point is it...

The AI seems to work very well. Although I think it gets too hard too quick when you jack up the difficulty. I thk the bts have too good an aim, especially with machine gun and such. This is "artificial" difficulty, because no human can aim like that, even a good human player. I'd say you should always have the bot's aim be a bit off when you are moving.

I am very conservative with shaders, to keep compatability with older Systems,
if a shaders makes a level significantly slower, I will not implement it.

That makes the bot difficulty:

Turnspeed, with which speed can the bot refix the target
Movespeed, (in hardest mode, bots move as fast as the player)
Attack-angle adaption, bots shoot with an offset, to compensate
for the bulletspeed in moving targets (on the harder modes)

Stand to attack, in weaker modes, bots have to stand to shoot.

Normal weapon recoil, as the player experiences too.

I know, that the aiming should have another offset, to make it
harder for the bots to fix a point.
But there are also players, that shoot as good as the bots,
so I made the hardes bots for them,

The weapons do a bit too much damage currently.

If you want, look into definitions_2, and tune the weapondamage a bit.

The Menu starts, but when I select a level and when I start it then, I get an Error: "Nexus too small(26457)" (or any number like this ^^) and then the game crushes.

You have to start the .bat file, to set a larger nexus

Yeah, that was it. Will play it later, have to shut down now ^^

By the way: as this demo is a submission to the coding contest
(open to read sourcecode)
I hope on your vodes

(in the Acknex User Magazine Part of this Forum)

I downloaded it, it was really good! That demo gets me very addict

And the level was nice

The demo is impressive, somewhat entertaining and perhaps it demonstrates that something of value can still be produced even without the immediate availability of an abundance of top-notch visual and audio production resources or assets (textures, models, music, sound effects, expensive software + skilled artists to produce such, etc.) when a developer/programmer is resourceful, obviously skilled, well organized, clever, knowledgable, etc.
Although there probably isn't much emphasis on deriving value directly from the demo itself, as it is probably intended primarily as a aid for promotional purposes in order to better sell products at a later date, it is of personal interest to me that so much value was accidently derived from the demo. Despite the impressiveness and moderate enterainment value of the demo, normally I would have likely only played it through a few times each on, at the most, two or three separate occasions.

First I will state that my son differs some what from other children of the same age, but, no doubt, many parents will make or have made similar statements.
I have a tendency to casually introduce certain games to my son, which he might find entertaining, while exercising a degree of caution when doing so.
I ran the XSpire demo while my son was in the room, and after playing it once, he was thoroughly addicted.
(This post shouldn't necessarily be an open invitation to those that might lecture about violence, video games and children. I am well aware that an act which introduces a younger child to violence in gaming may represent a critical lapse in responsibility and better judgement.)
He is able to move the player about himself, but prefers that I move the player while he turns, aims and shoots.
Although with my assistance he seems to do quite well on the easiest setting, he still has the patience to tolerate being killed over and over by the blue enemy AI players without being frustrated.
Rarely does he fire on the red allied AI players except when making some sort of point.
His favorite weapon is the rocket launcher and his favorite level is probably "Jostie's level" (I may be mistaken) which seems to have the easiest access to rocket launchers and rocket launcher ammo.
Quite a bit of enjoyment and excitement has been derived from the demo on numerous occasions, by myself and my son, partly due to the teamwork involved. (Neither of us would have thoroughly enjoyed the demo to such an extent without each other.) So for that, although such was probably not intended, Damocles, I thank you. If I vote, I will likely vote for XSpire.

I don't expect you to cater to my wishes or the wishes of a 4 year old boy.
Still, despite the fact that such will probably be ignored, I might like to see the following in future demos:
*a slightly easier difficulty setting or configurations for specifics of the AI like turning, aiming, intentions, aggressivity towards player, etc.
*additional levels with a pluggable interface which allows encapsulated downloads of just levels and accompanying information
*breakable level components (glass windows, exploding barrels, damagable structures (bases? + maybe base protection tactics) etc.)
*doors, elevators, rotating, spinning platforms ("lets get dizzy") which might fling actors
*danger zones with (possibly team unaffiliated) flying drones
*bigger explosions + greater area of effect for rocket launcher (no point hull), grenades, mines, + maybe trip-wire, traps, drop-a-dummy-auto fire, etc.
*a "help I'm stuck in the corner in Backyard Level" self-destruct + area of effect option
*keep the floating players, it's amusing
*while I'm making a list of everything + the impossible, mounted turrets, vehicles and drones
*model choices from the menu including Dubya (you can have mine with marine animations)

Hi,

good to hear, that you and your son enjoy the gamplay.
(you can turn off the gore/blood, if it seems not appropriate for him)

I can shurely make the bot even easyer on the lowest mode,
there is really a wide range of skill. The shooter cracks find it still easy in
the hardest mode, and as you sayd, it is still very challanging for others in the
easyest mode.

The Artwork and effects are still more of a Template, and not totally disigned though,
but that is actually not the most important point. Gameplay is.
Whenever I have more characters, I can include a variety to choose from, by the player.
Old Quake Characters are very easy to implement, thanks to 3dGS beeing a Quake-like engine.

I wanted to implement elevators, but they are actually quite hard to make for
the AI to use it correctly.

The weapons are not balanced yet, if you want to give it a try yourself, you can
change the damage rates in the definitions_2 script. They will be more balanced
in the next demo. (if you change the script, you cant run the game normally, you have to run it with SED !)
The plugin/download of external levels is a good idea,
whenever the AI itself has all features, I can do that.


I can even implement a splitscreen mode, for 2 players. Others also enjoy to play
it with their son. And seeing that people enjoy the gameplay is the point that gives me the most satisfaction.

The floating bots is more of a bug, is fun though

Turrets are a fun thing to implement too, so that a base is protected by
the teams turrets.
I have to see, how to implement destructable things, like windows and stuff,
or some bottles on a shelf, to get extra frag-points

In the Antaris Demo (Showcase 1) you can see how it plays in a Singleplayer Missions,
dont scare your son though..

Greets,

Damocles

Hi Damocles,

I downloaded the demo, but when I booted the file it looked like starting but
just ended and vanished ?

Never mind, any idea when a simple version of your AI will be available to purchase ?.

Please when you Password the programme dont use the method of the software
producing a different number each time it is installed and then we have to send you the number, and you then send us the registration key.

Mythic AI did this and now they don't answer their e-mails and that means that the programme cannot be installed without the registration key supplied by them.

What I am getting at is the software produces a different code number each time it is installed so the key you used last time is no good for this Re-install.

This is the third time I have been saddled with this problem. Sometimes the software developer looses interest in the project and does a bunk.

Did it install?

whenever you start the programm manually, start the .bat file,
else you will have a too small nexus for 2 of the maps included.

Normally thare should be a desktop icon to start it after installing it.

There is a non-installer version here:

http://coniroot.de/~pirvu/au/demo/zips/xspire.zip

About the registration:
I have not yet worked on that issue. But the person who purchase it, should
be able to install it of course.

Some minutes ago I played a demo of your AI. It's very very impressive. You can have a lot of fun with these bots. They "seems" to be not stupid, and they fight fair. Good work at all.