Jump to content

Jock Tamson

Members
  • Posts

    443
  • Joined

  • Last visited

  • Days Won

    2

Posts posted by Jock Tamson

  1. Why is it so important that I mentioned OpenGL in that post?

     

    My point still stands, the games run on completely different engines, therefore coding the AI like ARMA in Combat Mission, isn't a "two week" affair.

    You are the only person who has said two weeks.  Nor do you know anything about coding.  Is it really such a leap to understand that the units in CM could have their waypoints set according to conditional logic rather than a time based plan?  Because the last time I looked, they were already doing it with triggers.

  2. If you can't understand how coding an simulation level strategy games AI compared to a milsim shooter are two completely different things than I give up.

     

    Just to add on to this, don't you think BFC would have added this apparently obvious AI logic to there games if it only took two weeks?

    Um, right.  That's Arma with the industry leading editor with triggers and scripting of a complexity that blow CM out the water.

     

     

    Arma - has units

    CM - has units

     

    Arma - has groups of units with a command chain

    CM - has groups of units with a command chain

     

    Arma - has waypoints

    Cm - has waypoints

     

    Arma - has fog of war

    CM - has fog of war

     

    Arma- has a map

    CM - has a map

     

    Arma - has a TacAI

    CM - has a TacAI

  3. Modding Arma and adding complex AI to combat mission are completely different things.

     

    ARMA. Seriously. It's a shooter.

    That mod provides an opponent that dynamically sets waypoints for hundreds of AI units based on its understanding of the battlefield.  Its understanding of the battlefield is simply cyclical sweeps of the battlefield and applying its logic against changing values.  Map creators can pre populate a map with values - for example strategic points - but the basic set up is to give the commander objectives.  It will then recon them, choose a group of units to achieve them, and set the waypoints accordingly.  If later sweeps of the battlefield present a changed situation - for example newly spotted units - it will set new waypoints.

     

    If you can't understand how this is the basis for a more dynamic artificial opponent in a game like CM, I give up.

  4. Like a human would react? You are asking for something no AI can do in any game at the moment. And probably wont be able to do for decades

    I'll answer this once you've bothered to read any of my preceding posts eg the one where I describe creating the artifice of a reactive opponent.  Note the word artifice.

  5. Hehe  this game takes a fair amount of experience to get good at.  you need to really get down into the nitty gritty of the terrain and be careful and slow while trying to have a numbers advantage wherever you might be engaged.  I was just playing First Clash with my friend, he was telling me what he wanted his guys to do and I would give em some orders.  15 or so minutes in he decided I "wasn't doing it how he wanted" or whatever, and relieved me of direct command.  15 minutes later many question marks have been revealed but not too many enemy smoke stacks on the map but almost no Abrams left mission capable.   He controlled everything from too high up.  I think he's hooked though.  Ive gotten 4 friends to play CMBS so far and none of em wanted to at first but after their first time they are all hooked! :D

    What kind of games do your friends normally play?  Because this forum is full of people that think no-one except dyed-in-the-wool wargamers will play it.

  6. It seems odd that players involved in wargamming don't understand that simple fact. There are numerous articles on the internet specific to our hobby explaining how proper AI is a trick and how we are probably still decades away from any true AI where a computer actually thinks its way though a problem or situation like a human. All any game has now is pre-programmed routines. Players wishing BF would have a proper functioning 'real' AI want something that others have poured untold millions into and are still no closer to success.

     

    I can understand the call for much better graphics - I'd love that myself in an ideal world and it is possible if their are no monetary or time constraints. But to call for a thinking AI is absurd to anyone who has spent five minutes reading about AI in games - it's history, where it's at now and where it is going in the future. 'Good' AI now in certain games is a mirage - a trick played by programmers who have found a semi-plausible way to use set-routines to mimic 'real' AI.

    Nobody is asking for a thinking AI in that sense, just one that can react to the player.

  7. But in chess it sees one move and responds to one move with no LOS concerns building an AI to do what you want would be far beyond BFs capabilities and if they could do it, they would likely be working on something far more lucrative. Is it bad to want better overall AI, no but it is also not bad to want world peace. It is simply a question of likelihoods.

     

    I don't think anyone is asking for anything more than the type of AI opponent that is standard in games like the Total War series.  At the end of the day it is just using numbers and rules to create the artifice of a reactive opponent.  I'd even settle for one that could modify its pathing in response to casualties.  And before we go down the "it can't be done" line, I've got years of scripting in IT behind me, business logic and game logic are not so very different.  Line of sight doesn't matter, the AI doesn't take it into account when it plots waypoints now.  I don't think anyone is asking for an AI that plots moves with perfect LOS like a human, just one that can seem a bit more reactive to what is happening on the battlefield.

  8. Aye, but what I'm saying is that an overall AI for CM is going to be immensely complex if it can be made to work at all.

     

    It would have to make decisions based on objectives and terrain ( LoS or lack of it ), troop type availability and so forth. Then it must be able to adjust on the fly based on enemy encounters ( reinforce or retreat, hold position ? ) and so on. It would need some form of "memory" so that it wouldn't ignore units it had seen earlier. And human players would probably still be able to spoof it fairly easily.

     

    The more you think about it, the more complex it becomes. Even if the CPU cycles weren't being mostly used by all sorts of other stuff that CM is doing, you'd probably need a Cray in your basement just to come close to what you're asking for.

     

    AI triggers have helped, but in the end, if the AI isn't good enough, you just have to play another human.

    All of that has been done in Arma by one modder - 4 years ago.  Check out Hetman AI Commander.  Scans the battlefield, sets or amends waypoints, attempts to recon and flank, orders fire missions, transports troops by helicopter  That is on a battlefield many, many times the size of a CM battlefield.

  9. CM is a niche game and it'll always have a limited audience. That's not saying our audience couldn't get bigger with a bit more exposure but to try to suggest as some have on this thread that CM could take a lot of the Total War or Company of heroes players if CM just gets better graphics is off the scale ridiculous as far as I am concerned.

    So when we went from CMx1 to CMx2 there wasn't an increase in the player base?

  10. Does the AI in Arma know what a building is yet?  Don't get me wrong, I have got hundreds of hours in Arma2, but the amount of modding you need to do to get an immersive single player experience is crazy.  Eventually, I couldn't be bothered any more.

     

    Having said that, I would love the CM editor to have even a tenth of the power / flexibility of Arma's.

  11. Actually, there are some..

     

    First of all standard graphics are done by one or very few people who have to deal with all the 2d "painting" of the game, and that's a huge task to deal with.

    So, having to work on so many files of different nature it's hard to keep an extreme standard quality.

     

    Increasing quality would require a lot more time, thus asking many more months of development time. Time is money and BFC has to do cuts as well as set a certain quality to assure a good time/development ratio.

     

    Second, BFC sets a certain level over which the game must not go, and this level is about performance on a given machine (minimum computer requirements).

     

    Increasing quality would hurt that choice, force BFC to increase the target level, increase the minimum requirements, thus cutting a potential portion (estimated) of customers away... moreover, all of this would require game computing resources, so changes on the programming side should be done too.

     

    All in all I agree that at least some textures of the original game are not really good (personally, I dislike the buildings, that's where I started modding CM2x games), but we must keep in mind the contraints BFC has to deal with and operate within.

    Yes, this is all true, but there are issues such as the saturation of the textures on the different tree lods causing that nasty draw line which have been present in every CMx2 title and were fixed by modders [you, I think] very quickly http://community.battlefront.com/topic/117497-some-technical-questions-about-trees-rendering/

  12. Hardware specs don't mean much.  I go by game benchmarks, which the 960 wins for the most part, and the card's a little more future-proof.  Again, $200 for a USED card that gets lower FPS is crazy IMO.  Now for $100, I would be tempted.

    Think you need to double check your sources, according to Nvidia the 680 has better performance.  Which you would expect, because it is better hardware.  Additionally, at the moment the 960 isn't available with more than 2GB of RAM, so you have less capacity to anti alias, run at higher than 1080p, or downsample to 1080p from a high res (on my 680 I used to downsample from 4k to 1080p with no degredation in performance) than you would on a 4GB 680.

     

    If the V8 is quicker than the 4 cylinder, do the hardware specs also not mean much?  Don't think so.

     

    http://www.nvidia.co.uk/object/graphics_cards_buy_now_uk.html

×
×
  • Create New...