Search Unity

  1. We would like to hear your feedback about Unity and our products. Click here for more information.
    Dismiss Notice

Apex Utility AI

Discussion in 'Assets and Asset Store' started by Apex Game Tools, Dec 22, 2015.

  1. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Hi Igor,

    Thank you – really good and detailed analysis of your case. Great that you are sharing this.

    I think the tweaking part is a fair point, and good that you brought that up. Now, the big question is how to find a model that can balance financial sustainability with ability to keep prices of asset low for indies. One option here is for example to make an indie version that is e.g. buy-once. But the challenge of control might be difficult to solve.

    We know many of the existing options such as free for indies / expensive for pros, revenue sharing, pay for add-ons etc. Smart price differentiation seems to be at the core here.

    If you have ideas about models that could work, we are all ears. For example, are there specific add-ons that would make you able to create and publish games faster etc?

    As I said, this is not just about Apex, this is also about how we make it attractive for other asset developers to keep making new assets – as well as supporting existing assets. The revenue model is relevant here.

    Looking forward to your feedback.

    Best Regards,

    Jakob
     
  2. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
    Hi
    I too am at the beginning stages of a complex asset and find financial stability and sales and important issue.

    Cheep for indies / expensive for pros is definitively one of the ways to go, no disagreement here, much better than subscription.

    And your biggest problem is piracy and who is who (Indie or over $100.000 a year user, with chance to upgrade).

    To tackle piracy, personally for my asset I am considering using activating the asset on first use by asking the user to enter Unity user name and order number, I think it would be applicable to your case as well, you can even control are more then one users using the same license at once. If you notice a suspicion you give a warning notice or disable.
    As far as crack versions go, just forget about it, even your current model will be cracked, you can look at it as just free marketing.

    On the second issue, who is not an indie, without unity's help with an api to check current active user, not much can be done.

    Whatever your decision, accessible one time price (or subscribe untill full price) for indies is a must, with the possibility for a sale as well (I am waiting on Apex Steer sale for ex. even though I don't need it a.t.m).
    Anything else is a Indie no go.

    Cheers,
    Igor
     
  3. Cartoon-Mania

    Cartoon-Mania

    Joined:
    Mar 23, 2015
    Posts:
    297
    I think we should wait for next update

    If this is really a good product... $ 15 per one months is not expensive

    I hope They launch a planned product at the right time

    Honestly, now I'm disappointed. In fact, now it's nothing. There is only theory

    therefore we do not evaluate anything. There's nothing we can say

    I say one more time If this is really a good product... $ 15 per one months is not expensive

    Let us keep them a little peace of mind

    Anyway, I expect their promise.
     
  4. eridani

    eridani

    Joined:
    Aug 30, 2012
    Posts:
    655
    Will you release a playable web demo of enemy AI? Thank you
     
  5. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Hi Eridani,

    Absolutely.

    Let me know what features you would like to see, and whether you have a specific game genre in mind. Then we will try and accommodate it.

    Best Regards,

    Jakob
     
  6. MitchelQ

    MitchelQ

    Joined:
    Nov 18, 2013
    Posts:
    18
    Hi Jakob,

    In the Tutorial video it is explained that the scorers and actions are project-specific and there are not available when you start a new project. So where do they come from? Do you need to program them yourself?

    Is this asset just for selecting which action when to take? If so, wouldn't be an asset like playmaker of behaviour designer be way cheaper and further developed options which have a big helping army of other users?

    If you disagree, can you explain the differences between your asset and the before mentioned assets?

    Also I would like to see a demo of what this asset can do out of the box instead of tutorial video's. What can it do without those scorers and actions which in the video are claimed to not be there for a new project?

    Best regards,
     
  7. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
    Another dilemma with the subscription license, can I use the license on my laptop as well, is it device bound?
     
  8. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Hi Mitchel,

    The Utility AI is an artificial intelligence (AI) framework for advanced reasoning. It is thus a much more advanced AI product than any of the above, and much more advanced than e.g. behavior trees.

    In modern computer games, you run into several problems using traditional AI methods such as scripting, finite state machines, and behavior trees. These technologies tend to create very predictable behavior from AIs, they have problems adjusting for unforeseen situations, and they tend to be too difficult to make complex enough to emulate complex human behavior.

    The Utility AI provides solutions to these challenges.

    By using utility curves and scoring functions, the Utility AI makes choices based on the utility or usefulness of potential actions the AI can do, thus avoiding the typical scripted behavior of poorly designed AI. Utility-based AI responds in a more natural way, and can even choose reasonable actions in situations that the AI designer did not foresee.

    The principle of utility curves and utility scores also make it more natural to design the AI, as you can express the AI in a more natural language and you can keep adding action to an existing AI without breaking the existing structure. These are major challenges with finite state machines and behavior trees that the Utility AI doesn't have. This results in the ability to develop much more complex AI, faster, and with much less need for debugging.

    Finally, the Utility AI is better at abstracting higher-level problems. You can i.e. model high-level decision problems very easily, such as strategic problems, emotional problems, or political problems. You can also easily model different personalities in NPCs for example, by adjusting individual utility scores reusing the same decision logic.

    The Utility AI does also have many other uses. So if you want, you can use the framework to make scripted behavior, such as you do in Playmaker. Is can also be made to do finite state machine and behavior trees, if there are cases where you want this. The Utility AI comes with a full node-based editor, with all the functions you would expect suchas undo/redo, serialization, parallelism etc.

    Hence, short of machine learning - which we are working on by the way - the Utility AI technology is by far the best framework we have come across for design-based AIs. And this is the reason we have made an Apex product.

    Concerning your question about behaviors.

    The Utility AI consists of the framework and several behaviors that are needed for many game genres, such as sensing, memory, move actions, patrol, target acquisition, target prioritization, and many more. Beyond that there is a full, free implementation of an end-to-end Utility AI in the Unity Survival Shooter project, with behaviors for tactical reasoning, high-level decision making, etc.

    In addition to this, you can make your own behaviors. There is a full scripting guide and tutorials with concrete examples of how to implement your own custom behaviors, tips for best practices etc.

    Over the coming months, we will publish new tutorials with behaviors, as well as, add-ons in the shape of full projects within different games genres.

    If you have specific AI needs, be sure to let us know, so we can tailor some of our forthcoming demos, tutorials etc.

    Now we look forward to seeing all the great games that should come out with fantastic Utility AI.

    Best Regards,

    Jakob
     
  9. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Hi Donov,

    Each subscription comes with 2 activations - so yes, you can use it on both your e.g. workstation and laptop.

    And - as always - if there are problems, let us know, then we will work it out.

    Best Regards,

    Jakob
     
  10. MitchelQ

    MitchelQ

    Joined:
    Nov 18, 2013
    Posts:
    18
    Hi Jakob,

    Thanks for your thorough explanation, I can understand your asset is better in decision-making. I'm not convinced yet it will be worth the subscription fee.. I'll have to see more examples instead of tutorials.

    I hope you will give an option to buy the asset or a subscribe-to-buy model.

    One feature I would be interested in is playmaker integration (for the scorers and actions), any chance there will be in the future?

    Best Regards,
     
  11. frekiidoochmanz

    frekiidoochmanz

    Joined:
    Feb 7, 2016
    Posts:
    158
    "Each subscription comes with 2 activations"
    Hmmm....

    I think your product should offer a buy per version(s) like some other do, along side this model.

    Buy Ver 1-1.9 pay one time fee 100$, to continue to use best version upgrade price ---> 100$

    I'm not even sure how you track peoples activation status for stand alone unity products? Unless unity is loading a script inside the editor that checks for an outside key your website provides? I dunno, thats weird.
     
  12. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
    They have stubbornly decided to stick to the subscription model,
    also their 6 month subscription is $80
     
  13. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Hi Donov,

    Thanks the update - sometimes it can be good to stubborn ;)

    Joke aside - we are still considering to offer an alternative to the subscription model. However, we need to find the right model to avoid the pitfalls we have described in the previous posts - especially as our aim is a model that supports indies.

    Any input from you all on preferable models is welcomed.

    Best Regards,

    Jakob
     
  14. eridani

    eridani

    Joined:
    Aug 30, 2012
    Posts:
    655
    I would like to see a standard shooter demo (FPS, third-person, or top-down) where it shows some enemy behavior, such as attacking the player, reloading, maybe taking cover or retreating, even flanking or using teamwork, etc.

    Basically I would like to see how well your AI stacks up against these other Asset Store AI: Behavior Designer, Rain AI, Tactical Shooter AI.

    Thank you!
     
  15. MitchelQ

    MitchelQ

    Joined:
    Nov 18, 2013
    Posts:
    18
    I would like to see this as well and especially the teamwork (characters helping other wounded teammembers, creating cover fire and flanking)

    Also enemies using multiple weapons and choosing the best one at the right time (gun / knife / grenade) and enemies with a knife who do not stupidly rush at a machinegun equipped player but try to get closer under cover.

    And from my earlier post:

    One feature I would be interested in is playmaker integration (for the scorers and actions), any chance there will be in the future?
     
  16. Dawar

    Dawar

    Joined:
    Dec 7, 2014
    Posts:
    116
    please remove monthly subscription
    sell with fixed price with full source code .not only dlls.
    thank you so much.
    you did very great job. I like it so much
     
  17. Crusare

    Crusare

    Joined:
    Feb 14, 2014
    Posts:
    21
    And that right there is how you kill a promising project... you know, "jokes aside". And that's coming from someone who's purchased most of your other assets.

    You should revert to the model accepted by everyone, and charge an upfront price, even if it's fairly high. All the "pitfalls" you described above sound like an excuse to be honest, and not actual pitfalls. The subscription model is a ripoff, no matter how you try to justify it.
     
  18. grimmgames

    grimmgames

    Joined:
    Jan 11, 2011
    Posts:
    44
    I love Apex Path and Apex Steer, and I'm sure I would love Apex Utility AI, but a subscription? Not going to happen. The closest you can get to this without outright revolt is to do something like Allegorithmic and have regular paid upgrades, and even that was a tough pill to swallow for many.
     
  19. MitchelQ

    MitchelQ

    Joined:
    Nov 18, 2013
    Posts:
    18
    I tried to try your trial... the license key I got didn't even work..
     
  20. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Hi Everyone,

    We just gave a lecture on AI at the Nordic Game Jam here in Copenhagen, and we have been giving classes on AI all week to students and the game jamming community here in the region. We will upload the video recording of the lecture and slides once we have it.

    It also means it has been a quite busy week.

    One thing we learned was that many developers know what a Behavior Tree is, but few know what a Utility AI is. Is that a picture you can recognize?

    In that case, we will post some conceptual papers on Utility AI to everyone can get to understand what a great technology this is.

    Best Regards,

    Jakob
     
  21. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @eridani

    Very interesting idea about stacking the AIs against each other. The question is how we make a fair comparison.

    Behavior Designer and RAIN are behavior trees, and the Apex Utility AI is a - well - utility ai. That means the Apex Utility AI can do behavior trees, but the others cannot do utility based reasoning.

    Thus, by design, the Utility AI will be superior in most all aspects.

    Another problem we would run into could be whether it is the technology or the design of the AI that's the problem. I.e. a great technology with a poor implementation is not going to stack it against a poor technology with a great implementation.

    We could, however, look at parameters such as usability, performance, how quick it is to design certain behaviors etc, But again that might be biased towards certain technologies and not give justice to whatever technology is biased against.

    But the idea is great, and learning from each other is what this community is about, so we can make greater games. Hence, if you have some concrete ideas, please share them.

    There is an AI battle thread somewhere on the forum, but it doesn't seem to be very popular. Perhaps we should consider invoking this again?

    Best Regards,

    Jakob
     
  22. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @MitchelQ

    We have done a lot of work during the propotyping of the Utility AI in exactly this genre of tactics, collaboration etc. So this is something we can definitly make both behaviors and tutorials on how to do.

    I will put this on the list of example projects. The Apex Survival Shooter has a light implementation of this, with e.g. reasoning about tactical positioning, and selecting when to use a bomb versus shooting with the rifle. Hence, you can already take some principles from here. But we have much more advanced examples.

    Regarding playmaker integration, that is a good point. We will put this on the list as well.

    I think it would interesting if you can elaborate a little on how you use Playmaker for game design. I.e. is this something you use all the time, or do you mix it with other tools? Are there certain tasks that Playmakers lends itself very well to?

    As you know, the Utility AI currently features a node-based editor, since this was the tool that seemed most convenient to use from the beta-tests we did. However, if there are other ways that you prefer to work with AI, it could be great if you would share.

    Best Regards,

    Jakob
     
  23. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @MitchelQ

    Can you describe how you have retrieved and entered the license?

    Normally, the process is that you sign-up, you get an email with a download link. You need to use this version of the Utility AI with the license to have it working, as the license is tied to the DLL.

    Hence, if you could please go through the steps, then we can help you get it up an running.

    Best Regards,

    Jakob
     
  24. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @Dawar

    We are considering making a non-subscription option available for indies, since this option seems very popular. Hence, thank you for sharing.

    I don't think we plan on getting rid of the DLL's and licensing. However, the AI behaviors themselves are in source code.

    If you need access to the source of the DLL this is also an option, but then we need to understand a little more about your usecase. The DLLs are mainly editor and core loops etc. Hence, for building AIs this is not something you would need to access.

    Best Regards,

    Jakob
     
    Last edited: Apr 9, 2016
  25. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @Crusare

    We are considering making a fixed-price option available. The more opinions we get on this the better. I described some of the challenges in the previous posts, as this also affects indies, directly and indirectly.

    That is why it is so critical that we get feedback, since there might be aspects we have overlooked or not understood. So please see this as the start of a conversation, rather than an attempt of coming up with excuses.

    In that sense, we have a joint mission: Make great assets available at low prices to indies. But that requires a sustainable business model for the asset developers, not just low prices for the indies.

    I would be very interested if you had additional feedback, especially on how we can get the best of both worlds.

    Looking forward to hear from you.

    Did you try the Utility AI, btw?

    Best Regards,

    Jakob
     
    Last edited: Apr 9, 2016
    Tiny-Tree likes this.
  26. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @grimmgames

    Thanks for the feedback.

    You have a good point that Allegoritmic's model is quite unique, and by many seen as a breakthrough.

    I think first point here is that products such as Allegoritmic's substance designer and Apex Utility AI are professional products that require research, and hence are not cheap to make. To fund that development process, they need to retail at a certain price.

    Now, the problem for many indie developers can be cash flow, and thus breaking down the payments into chunks, means that you can get access to the software right away.

    That you are given the option of getting the software at one point - thus not having to pay anymore - is quite cool. The full price for Allegoritmic's products seems to be massive, so selling this as a fixed-price asset wouldn't seem to work since it would be out of reach for most indies I guess. Hence, could you perhaps elaborate on what you would see as a better model?

    The other option to keep upfront prices low could be to do something like revenue sharing, or to charge a high price if a game using the product is successful. I don't know whether that's an interesting option, and it is also very difficult to control as you are not automatically notified of how a game with your asset it doing. Hence, we haven't considered these options, but they also don't seem as viable roads to travel.

    A third challenge is how we compare cost versus benefits. The discussion currently is very much about the price of the asset itself, and not so much about how the asset benefits the game development project. I think we need good measures for also measures benefits of different assets. I.e. much time do you save? Are you shipping better games with some assets? Are you chances of actually shipping better with some assets etc?

    That companies like Allegoritmic is making an indie version is great. They seem to have quite a professional market already, and the last thing we would want is to scare away professional companies from the indie market with too high demands, I guess. There are many solutions that already don't reach the indie market, thus putting indie games at a disadvantage from the onset. This was the reason that we started Apex Game Tools in the first place: to make professionals tools available for the indie market, so indie developers weren't outgunned by the professional development studios.

    Hence, any feedback appreciated.

    Best Regards,

    Jakob
     
  27. eridani

    eridani

    Joined:
    Aug 30, 2012
    Posts:
    655
    Thank you for your reply.

    I don't really care about the underlying implementation details... As an indie developer with a staff size of one (myself), I simply want to know how *easy* it is to set up a *good* shooter AI using your system.

    I'm an asset buyer and I own all three assets I mentioned. The other three assets actually all have fairly good shooter AI right out of the box. However, the AI for each system has quirks and weaknesses, which is understandable as even AAA titles have AI with quirks and weaknesses.

    I simply want to see if your product can quickly (or better yet, out of the box, if you include a demo) deliver a comparable or perhaps superior AI to the three assets I mentioned, since you say you are using a superior underlying technology.

    Otherwise, I will have to assume that the inferior (as you claim) behavior trees employed by these other three assets can deliver a better shooter AI more quickly and easily than your Apex AI asset, even though Apex AI utilizes your highly touted utility AI that you claim is so superior.

    A fair comparison is playing a demo against each asset's AI. You can go ahead and do that right now by playing the demos provided with all three of the assets I mentioned. The demos quickly let you determine which AI is most suitable for whichever game you are making.

    No offense intended, and I am just speaking as an interested customer. Basically, I don't want you guys to discuss how great your AI is. I want you to come up with a demo quickly that is on par with these other three assets, to show off what it can do compared to the other guys.

    Basically, I want a test drive! I don't want more discussions about how great Apex AI is, I want to see it in action in a concrete way (a demo).

    I feel that, since you guys are the experts of this system you have developed yourselves, you shouldn't have any problem coming up with a shooter AI demo quickly. Any other reply about wanting to discuss the problem further just seems like you are evading a simple request: we want to see your AI in action in a playable demo =)

    I hope you consider this a friendly challenge and not a criticism. Can you deliver a playable demo to show us quickly? I am curious to see how long it will take you to make such a demo, as I can only assume it will take anyone who buys your asset much longer (as we are not experts at your system).

    Thank you!
     
    Meltdown likes this.
  28. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
    These discussions about licensing and price are really pointless, and as someone already said it's just an excuse for the rip off subscription is.

    If you stand by your claim that accessibility to indies is your reasoning behind it, please just go the Allegoritmic' way.
    Make Indie and Pro bundles of all your products, set your price tags, and allow buying it outright, or subscription until full payment. For any current owners of your products allow upgrades to the bundle package.Set much higher prices for the individual assets to make the bundle a better deal. make some good marketing to get some hype going. thats it

    I'd buy/upgrade/subscribe today for whatever price you think is reasonable for the bundle.

    Anything else or more is just excuses. I even hate Unity for being asses about Unity's subscription. It should subscribe until buy, plain and simple.

    Regards,
    Donov
     
    Last edited: Apr 9, 2016
  29. christoph_r

    christoph_r

    Joined:
    May 20, 2013
    Posts:
    467
    I actually feel bad for you guys. You release an apparently great and innovative product that was probably quite work-intensive to create and you get criticism and lack of interest because of your licensing model. I can understand both sides. A large part of unity users are hobbyists or indies with almost non-existent budgets, for whom shelling out somewhere around 100$ to keep on using their assets is very unattractive. A big reason for that of course is the current situation where that amount of money can get you up to 3 high quality assets that won't require subscriptions. I think another big factor is that it's not immediately clear what happens when you cancel your subscription. It's not like the SpeedTree modeller, where it's obvious that you can work on and export the trees you need for you game for a month, cancel the subscription and you're good. For a start, it might seem like you HAVE to subscribe for the production length of the game. Additionally, in most cases it's necessary to iterate gameplay and rework the AI accordingly. The third issue I see is that the underlying concept is relatively new. So you have both an unorthodox pricing method and product, in a market where many participants are rather risk-averse due to their limited financial resources (at least as far as unity goes). The unpopularity of highly-priced assets (100€ +) supports this hypothesis.
    That the pay-once model is very much to the advantage of the buyers is evident. You've listed plenty of reasons, most of which I agree with. However, as was commonly suggested (and it's being increasingly used in the asset store) is the option for paid upgrades and expansions. Paired with the option to pick which feature package is needed/wanted (with the possibility to upgrade later) it would most likely work for many people. This does not completely mitigate the "unlimited" free support aspect, but there are several pieces of software with this issue (e.g. operating system.) In fact, most technology (speaking of hardware here) is pay-once only, which probably established the idea for most people, making subscriptions in turn less popular. On top of that, there's some basic "I might need this later" mentality that makes temporary purchases less attractive, even though it might be irrational.
    Irrational in the sense that a subscription based model might be cheaper for most people. However, most don't get to the point where they compare what a product would cost as a one-time purchase compared to the subscription costs depending on how long they use it. They will most likely have made up their mind before they get to that point. Additionally, it's difficult to judge for how long you'd be using the product as well, especially because indie-projects usually don't have a clear-cut timeline.

    Summing it up, I think the biggest reason for this resistance is the fact that people are not used to this model as well as the "not really owning it" effect. My suggestion would be to emphasise the advantages of a subscription model to the user, if you decide to stick with it. It should probably not center around "more future updates" as people already get that with pay-once assets, rather show by example how it's more flexible and might even save the customer money. The alternative, of course, would be a pay-once solution with paid upgrades.

    I hope this works out for you financially, because this looks very promising, support, presentation as well as the product itself.
     
    donov likes this.
  30. nasos_333

    nasos_333

    Joined:
    Feb 13, 2013
    Posts:
    7,368
    Hi,

    If one owns the other AI packages (Apex path etc), will these work together with Unility AI or this new system covers also the other packages ? Can they be combined with it if it is the former ?

    Thanks
     
  31. MitchelQ

    MitchelQ

    Joined:
    Nov 18, 2013
    Posts:
    18
    Where can I find these examples?

    Nice!

    I use Playmaker all the time as I cant program a single line. I will need it to create scorers and actions. I do mix it with other assets to make it work in my game. I feel things like your "highest score wins" are quite easy to create in playmaker as well, but I think it can be more easy to manage in your asset as you need to create less connections.
     
  32. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
  33. eridani

    eridani

    Joined:
    Aug 30, 2012
    Posts:
    655
    They don't have a demo either...
     
  34. sicga123

    sicga123

    Joined:
    Jan 26, 2011
    Posts:
    780
    What's the situation once a game is published that uses this AI Utility? Does one have to keep paying during the period the game is being sold? In the case of mobile it's quite possible the monthly subscription for this would cost more than was made by the game on a monthly basis if the subscription also had to cover the sales period.
     
    Last edited: Apr 11, 2016
  35. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @donov

    There is a previous post regarding alternatives to the Utility AI.

    Regarding DecisionFlex, it is great to see that other have picked up on the technology.

    Thanks for posting.

    Best Regards,

    Jakob
     
  36. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @sicga123

    You pay for editor access, not the AI. Hence, the AI keeps working, so you only pay for the months you need to tweak the AI.

    Best Regards,

    Jakob
     
  37. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    We have received a lot of requests regarding demos of the Utility AI, including e.g. simulating personalities, tactical decision-making, FPS, RTS, RPG, x4 dialogue etc. So far we have tutorials in the making, but having actual, playable demos is of course a great way to share concrete implementation knowledge.

    Making the AI for these scenarios are not the problem, but making a the game around the demo - i.e. assets, game code etc. to make it playable is - of course - time consuming.

    Hence, there are a couple of options, and I would like to understand what is the preferable:

    1/ We make demos on existing Unity projects.
    2/ We make demos on existing games from the community that has already shipped.
    3/ We make demos based on game in production - i.e. we basically make the AI for a number of games you guys are working on.
    4/ We make completely new Apex demos.
    5/ We make complete Apex game templates for selected genres
    6/ Another way to do this?

    There are pros and cons to each scenario, I have listed some - feel free to add more.

    Ad 1/

    Pros:

    There is a quite a number of Unity projects available on the Asset Store to choose from.
    Assets can be freely distributed.

    Cons:

    The projects don't cover all genres.
    The projects might need refactoring - which will take time.
    There will be a lot of junk code and assets, as they are generic demo projects.
    The projects have all been seen before.

    Ad 2/

    (Similar to e.g. the Angry Bots demo)

    Pros:

    Known game and game play.
    Delta between existing AI and Utility AI can clearly be seen.

    Cons:

    The owner needs to grant permission for sharing the assets in the demo.
    The might be need for refactoring, depending on the state of the game code.

    Ad 3/

    Pros:

    Apex helps building AI for one or more indie developers.
    Close to real production cycle, and hence gives a realistic image of the AI implementation process

    Cons:

    The owner needs to grant permission for sharing the assets in the demo.
    There might be need for programming of game code, depending on the state of the game.

    Ad 4/

    Pros:

    Project can be custom tailored to AI demo purposes (this might also give some bias, though)
    No permissions for assets etc. needed

    Cons:

    Will take longer to materialize


    Ad 5/

    Pros:

    You get a complete game framework with AI

    Cons:

    This option is expensive and time consuming to make

    Also, I would like to understand priorities - i.e. what demos / genres / AIs should we do first?

    Input welcomed.

    Best Regards,

    Jakob
     
  38. Cartoon-Mania

    Cartoon-Mania

    Joined:
    Mar 23, 2015
    Posts:
    297
    I Think

    1) You have a lot of time and You have a lot of money

    You can make a lot of game kit

    FPS game kit, RTS game kit, 3rd person game kit, sims like game kit, RPG KIT

    2) consider the current situation.

    You should integrate with popular products

    UFPS, 2D+2.5D Platformer Corgi Engine, Adventure Creator, Third Person Controller - Melee Combat ...

    Integration with other products is very important, You must integrate with other assets

    3) My hope is..

    Dungeon Architect + Third Person Controller - Melee Combat + APEX Utility AI


    The concept is just a concept.

    Playing is believing

    You should provide playable web demo as soon as possible
     
    Last edited: Apr 11, 2016
  39. sicga123

    sicga123

    Joined:
    Jan 26, 2011
    Posts:
    780
    Ok, that's good to know, seems like a more reasonable proposition. I have your other assets so knowing that definitely makes it worth considering again.
     
  40. NightAvail

    NightAvail

    Joined:
    Feb 23, 2015
    Posts:
    13
    I own one of these products as well (Apex Path) and I think the devil is hidden in the details in the statement made by Jakob. Here's another way of saying the same statement, "You cannot tweak the AI unless you are subscribed"; for me at least, this is not reasonable as throughout the course of the game's life you will always need to tweak the AI.

    I will congratulate you on the release of your product, and I hope that in the near future you reconsider other options besides the subscribed version, and provide a lifetime version (even if that's alongside the subscription version).

    - Furthermore, 1 license per seat would make this very pricey for small indie teams that are using the Unity editor.
     
  41. eridani

    eridani

    Joined:
    Aug 30, 2012
    Posts:
    655
    UFPS demo would be nice. I think the vast majority of people looking for a tactical AI own UFPS already.

    Bootcamp demo would be second choice I suppose.
     
  42. eridani

    eridani

    Joined:
    Aug 30, 2012
    Posts:
    655
    This
     
    EvilGremlin likes this.
  43. RecursiveAnomaly

    RecursiveAnomaly

    Joined:
    Oct 29, 2014
    Posts:
    1
    You have talked a lot about how your subscription model more evenly distributes the cost between large and small developers. I think this is deeply flawed, but I don't feel any of the previous comments have contested this point.

    Personally, I develop games in my off time. This means that the amount of hours per week that I can work on any project is significantly less than a studio working full time. (This is one of the reasons I love the asset store, as I can find things here that let me use what time I have a lot more efficiently.) I commonly have projects that don't get any work done for several weeks at a time. If I want to work at all with a subscription based service, than I am paying much more per development hour each month than a full time developer.

    This asset would cost me (and many idie devs like me) more than full time studios per person for the same project.

    I purchased an asset over two years ago for around 100$. If I add up the number of months I have used the editors in that asset, a monthly subscription would have been about equal price wise. But I would never have used it that many times because I would have avoided reactivating a subscription every time. Every time I would want to change the tiniest thing, I would have had to decided "Is this worth a month's subscription." Many times I would have not done it, and at that point I am not getting the benefit the library.

    The comfort that I can use something whenever and however I want, without any strings attached, means a lot.

    Larger studios have more stable budgets, hire more expensive people, and have more regular cash flow. Indie studios have irregular cash flow, so taking on a subscription is a larger risk. I feel like this is the root of many of the "I might not be able to afford it later" reactions. A large studio knows it's burn rate, and it can decide if another subscription is within the budget. Indie developers usually only spend money when they have it (they get paid, contract is finished, ect) For them, they know they can afford a one time payment. They don't know if they will get any income at all next month.

    You claim that some projects may only need to change AI behaviors for a short time period of use. No software project I have been on, in games or otherwise, was ever comfortable saying "this part is done and will never be changed." Even if there is only three months of work done on AI, there will never be a time where you wouldn't want to change/fix/tweek/add it.

    I appreciate the obvious care you have for this product, and I want to support it. I do not blame you for trying to find ways to make the asset store more stable to develop for.

    Unfortunately, I will not buy subscriptions from the asset store.
     
    christoph_r likes this.
  44. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Apex Utility AI Without Subscription


    Hi Everyone,


    We are very happy to tell you that we have made an edition of the Apex Utility AI available on the Unity Asset Store that does not require a subscription.


    The price is USD 65 for the next month. So go get your copy!


    The edition on the Asset Store is the Personal Edition. It has the exact same - and fantastic - functionality as the professional edition, but with the condition that you have to make less than USD 100,000 a year in revenue. If you make more than that you are wealthy enough to support the development of great tools for indie developers by paying more.

    It is a per-seat license, but comes with two activations. We kept the license so the decent developers don’t get ripped off by pirates copying the software. If you don't have the money to buy a license, send us an email instead and we will figure something out. This includes students, as we are planning a free edition for education.

    Hence, also do not waste reviews on the licensing issue - if it doesn’t work please send us an email. We have some of the best support on the Asset Store, and we will help you.

    Please instead write reviews about how the AI tools actually work, so other developers can see whether this is something for them or not. Also note that the current Utility AI is a framework / foundation with example behaviors and some already decent tutorials - including a full refactored project implementation. It is already a fantastic tool - and more will come.

    We want to thank you all for your feedback and support for the discussions here in this thread, and for your continued commitment to making great tools for Unity together with us.

    We - at Apex - want to underline that we are committed to our mission about making AAA tools for indie game developers. So we hope that you keep backing us, and let’s make some great games!

    Now we will get back to work, and make some magic demos for the Utility AI.


    Best Regards,

    Jakob
     
  45. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    Procedural Animations Tools

    Hi Everyone,

    While working on the demos, we got tired of rigging meshes and creating animations. So, instead we made a small procedural animation prototype, so characters are animated automatically in the forthcoming demos.

    It looks like ****, but we wanted to share an early video with you, to get feedback.

    Please check out the video:



    Basically, what we do is that instead of going the old fashioned way of mesh->rig->animation, we animate the whole body via AI.

    The big question is should we make this into an asset?

    Procedural animation basically means that you don't need animations and IK (Inverse Kinematics) anymore. Instead, the characters animate themselves via real-time modifiable scripts.

    It also gives you a lot of other advantages:

    a/ You can precisely control all limbs
    b/ No foot sliding
    c/ No need for animation controllers
    d/ No need for expensive animations
    e/ Hands can grab with exact precision
    f/ You can send characters across dynamically changing environments, such as moving ground, ships etc
    g/ You can dynamically adjust character behavior, such as modifying / varying the run cycle of a characters while running so they don't look static
    h/ You can seemingly run multiple behaviors, such as aiming while running
    i/ You can use physics on characters while they move, to simulate e.g. impact
    j/ You can blow off limbs run time, or modify the character in any other way
    k/ You can modify characters run-time, e.g. growing a new leg or arm or wings
    l/ You can have characters with different movement patterns, such as limping, bumping walk, just by modifying parameters
    m/ You can animate non-bipedals - e.g. insects etc with little effort

    We could imagine this would allow indie game developers a completely new tool for animation in their games - potentially going beyond the options that the AAA's have.

    Hence, let us know if you want this, and please help us drum up support.

    We can make a separate thread for this as well.

    And if you have feature requests, let us now as well.

    Best Regards,

    Jakob
     
  46. Tiny-Tree

    Tiny-Tree

    Joined:
    Dec 26, 2012
    Posts:
    1,270
    now it does not require subscription i become interested. does it require to use other assets like apex path or we are not tied?
     
  47. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
    This is a good move. Just a few more questions.

    Where is the license stored, is it in the project or activated for the machine? Can I safely delete my project without worrying about it?
    Do I need to deactivate it when updating/deleting Unity/Windows and how?
    You really should add some FAQ for this.

    Other than that, great job and product.
    I am almost at a stage to implement some AI in the current project. Can't wait to try this out.
     
  48. donov

    donov

    Joined:
    Apr 15, 2013
    Posts:
    45
    It does not, this is purely for AI decision making, like tell the entity to attack etc.
    To actually get the path to attack you need Apex Path or some other pathfinding solution.
     
  49. nasos_333

    nasos_333

    Joined:
    Feb 13, 2013
    Posts:
    7,368
    Great decision, will definitly be looking into it.

    Does it play together with Apex Path and other Apex products ?
     
  50. Jakob-Rasmussen

    Jakob-Rasmussen

    Joined:
    Jul 26, 2014
    Posts:
    92
    @Damien Delmarle

    As @donov wrote - the Utility AI does not require any other products to work. It is high level decision-making (and low-level as well).

    We are creating out-of-the-box implementations for other Apex products, of course. If you need e.g dynamic pathfinding from Apex Path or e.g. formations from Apex Steer.

    Best Regards,

    Jakob