Price for a good AI?

Hi, I am an AI scripter, and I would like to know how much you'd pay for this: an AI which follows waypoints, interacts with other AIs (currently talks to them, maybe eventually even exchange information), if he sees the player on a clear day, from your choice of meters away he'll start shooting at him (I could make it so he charges at you), and alerts all the other close AIs to be on alert, if you sneak up behind him, he currently runs to a place that he finds safe, but I could make it so that he charges you, or even calls in other AIs, and you would have the choice of how far he can see, and if he's affected by enviroment (on a foggy day or at night, he will only be able to see half as far), if he has low health and he has a potion (food, what ever heals him), he will regain health, and if he has low health as he talks to one of his buddies, and has no potions left, he will ask for a potion, and you can easily assign all of the animations. The code would be well documented, so that even someone who does not know much about scripting would be able to change something (like the tag of what would separate one of his buddies from the player). I am only partially through making this, so it would take a while to finish, but I would like your honest opinion, how much would you pay for it? How many sales do you think I would make? And from people who have sold on the asset store before, how much did you make on your first item?

 

50$

 

I think that depends on what would make your product different from other AI solutions. Is this for a specific scenario where you expect plug and play? Is it flexible and extensible? etc...

I would suggest you buy and test every AI solution out there to understand where the state of the technology is. Note that I am NOT encouraging you to copy anything, just to know the feature sets, etc. You might be better off picking your favorite and contributing to the developer, or creating an extension, rather than spending your time reinventing the wheel. if you DO find a big gap that you think you can fill, then you'll know you have something that might be worthwhile.

You also need to consider all the time and energy it takes to create docs, support issues and release new versions. This all takes away from the time you will spend on your own apps.

Best of luck in whatever you decide to do!

 

Sadly, I am no app maker. I prefer to make large scale xbox games, but first I will need some money, and this is what I love to do. If anyone has any suggestions, please feel free to add them.

 

I don't mean to discourage you, but even if you make the best AI tool ever invented for Unity, it will not find an XBox game. If it is what you love to do though, then this won't matter, which is the right reason to do it IMO!

Still, my advice would be to know what is already out there. That will help you know what features are missing and may be valuable.

 

if i was you i would make a AI pack with like 3 ai scripts for strategy games,3 for fps games etc...

 

If I could add in seeing the player depending on how bright it is (day, night, near a lightsource), would people want that? And if so, would you want it done via variables (current hour, maybe the transform of all the lights?), or via collision? And also, how would I go about this?

 

Frankly I'd recommend the opposite - what's out there is already done don't reinvent the wheel here. What would be really nice is to see a kit that would give you a script that could be attached to an arbitrary npc (character controller or rigidbody based) and perform some behaviour. Instead of assuming light source, day/night etc.... just have it event driven so you pass in some parameters and call the method, allowing us to return a true/false to identify that the script should take that behaviour.

Maybe you can develop a small framework to support these behaviours using callbacks/an event driven model then just release behaviours as you finish them - make the framework free and a couple basic scripts then sell the new scripts or do them as custom development services.

 

Do you have any AI coding that you could show us?

 

so far, all I have now is interactions with other AIs, pathfinding, and currently working on his reaction to finding the player

 

Nice.

 

WORD. What I would love to see some AI primatives that could be just tossed on a object.

 

Do you know how to build Ai for strategy games? Like turn based or RTS?
Something along those lines could be worthwhile.
B

 

I may get into some new projects as I get more experienced, but for now I have to finish this project. So far, he now accounts for chance of seeing the player (1-100), depending on user input, distance, fog, and working on time.

 

Any webplayers of your AI?

 

Not yet, hopefully you'll understand my caution towards those, with them being easily decompiled.

 

Then what about a video?

 

Quick question, how would I do this?

 

use fraps or similar to record a video

 

camstudio

 

Right now just trying to make "friendly" variables (fogIsVariable and so on...). Almost every variable (except for 15%) has a sentence of commentation, is organized into groups of relevence to other nearby viariables, is alphabetically listed within these groups and the private and static variables of each group are placed at the end. Each function is separated by a long set of lines and a title. I am using gizmos to show the AI's long, medium and close vision. He can now communicate with other AIs, follows a path (and waits before going to next pathpoint), uses animations (models by Arteria3D of course), is able to see player while taking into account fog, distance, and user-defined chance of seeing player and runs to place that he finds safe if is too close to player or player spooked him (ran up behind him, soon will also be if he killed his friend). Now, to take daytime into account, would you be willing to make the fog denser at night? He is now 500 lines long, and still running smoothly. I avoided using character motors to make him run faster, but now he can't use collisions unless one of him or his buddies has a rigidbody. Also, are you okay with him only talking to another AI if he's in 35m range? I can't go bigger, or he slows down the game sometimes. I can use raycast if that would be better. I value your imput. Video comming after I perfect this.

 

Sounds good.

Some ideas I have:
Seperate scripts that can be attached would be easiest for the user.

1) Hearing: Distance from the character is actually hearing - perhaps divide the sound the character is emitting by the distance to the enemy. If it's above a specific value, the enemy is alerted.

2) Visual: Ray casting will give you line of sight and should only apply infront of the enemy - so that you can sneak up behind him.
I find this to be the most tricky and usually a big issue of realism. If I'm hiding in the grass wearing camo, visibility sould decrease.
perhaps if you give your character a visibility variable - so we can set it to e.g.) if he's standing 1, crouching 0.5, lying down 0.2, minus an additional .1 if in grass. Then divide the distance by visibility and fog. If it's above a specific level alert the enemy.

3) Talking: have the range variable so that it can be tweaked. I would be fine with 10m. perhaps the character can have a variable "how much I feel like talking" - goes up when he's not talking and down when he is - so that he walks away after he's said what he needs to.

4) Wandering: This is not a trivial task. random wandering looks unreal. Good wandering would add much realism. Perhaps influence maps would be a good idea here. Something like a hotspot on the map where people chat, or a grazing hotspot that will attract animals (if they are hungry).

Woa, that almost turned into a lecture. I huess you can tell I've been working on my own AI

 

It would be GREAT another input like the real player visibility based on the area of the player which is exposed or the contrast to the background.
I know it's not easy at all to achieve, probably very heavy on resources, but still for a hide and seek game it would be so damn nice.

 

Well, I'm using raycast and chance of seeing player (in percent) in a way so that if the player is halfway behind a wall, the AI has less of a chance of noticing him. I still have to work on my own player's combat system before I can make my AI's combat system. But for anyone who leaves a suggestion, please say what you think this is worth.