FAForever Forums
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Groups
    • Login
    1. Home
    2. maudlin27
    3. Posts
    Offline
    • Profile
    • Following 0
    • Followers 7
    • Topics 21
    • Posts 1,026
    • Groups 2

    Posts

    Recent Best Controversial
    • RE: M28AI Devlog (v229)

      v229 Update
      Smaller update with 6 changes, primarily fixing some naval issues:

      • Text markers that M28 places for teammates should be removed after 90 seconds
      • M28 should be less likely to think it has air control with small numbers of inties, and shouldn't try and attack air targets near the enemy base if it lacks air control
      • Fixed several issues with how M28 calculates the nearest enemy naval unit in a zone (which would lead it to ignoring units in the same zone as it for parts of the logic)

      Acknowledgements

      • Vortex - QUIET naval replay against v227
      posted in AI development
      maudlin27M
      maudlin27
    • RE: T2 torpedo turrets are awful

      Agreed would be nice for a range boost (so they have at least the same range as the best destroyer range but cruisers and battleships would outrange them), - could always buff the Aeon cruiser range slightly since its a weak attack anyway to compensate/provide Aeon with a T2 counter. That way there'd both be more unit variety and unit counterplay.

      Land combat has T2 PD that outranges T2 tanks, but is outranged by another T2 unit (MMLs), which in turn can be defended against with TMD, and the game feels more interesting with that as an option than if say T2 PD were outranged by T2 tanks. Even with fixed shields being buildable on land to cover T2 PD, and T2 arti being available (that outrange any land unit) firebases aren't oppressive (while for naval maps outside of protecting your naval factory and the occasional pond with a chokepoint, there'd be less value from a 'naval firebase' than on land, so less risk of it being too powerful).

      Another nice sideeffect is it'd provide a potential (defensive) counter option to sera subhunters (who currently outrange torpedo launchers other than HARMs)

      posted in Balance Discussion
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      v228 Update aka the 'Vindex beat 1.5 M28 AIx update'
      17 changes, 15 of which were based on a replay where Vindex beat 1.5 AIx on the Ditch, including:

      • 8 improvements and bugfixes to M28's 'telesnipe defence' logic, including making it no longer think the laser can be dodged as though it was a t1 arti; the ACU should no longer stop running after a short while of being chased by a tele-laser com; it should recognise the enemy has a laser ACU sooner (if it has visual of it); PD should be built preemptively to defend (instead of waiting for the teleport to start); and other changes
      • Fixed an issue where M28 would try and drop plateaus but choose a tiny land zone to drop, causing it to drop early (not on the plateau)
      • Reduced the core base T2 builder so it no longer send all engineers to build T2 PD in response to a single heavily outnumbered enemy tank
      • Air facs should be more likely to be built on larger maps once M28 has 4 land facs in its main base
      • (unrelated change to the replay) T1-T2 skirmishers shouldn't be built if the enemy has T3 land.

      Acknowledgements

      • Vindex - 4 replays against M28 AIx on the ditch

      Trophy Award
      A rare update to the trophy holder - @Vindex is now the holder of the Sladow trophy for beating M28 at 1.5 AIx on the ditch - the first time I've seen anyone manage the feat!

      posted in AI development
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      I'm not sure, as the method used by players will be via the UI, I've made a note to potentially look at in a future release

      posted in AI development
      maudlin27M
      maudlin27
    • RE: Thoughts on UEF doctrine and balance from a noob

      "UEF has worst land scouts."
      They have some advantages to the others. While you dismiss their extra DPS, it means they can kill enemy land scouts and radar much faster. Personally I prefer them to selens because they have better radar range (which is why I build land scouts) and are much cheaper. Spirit's by far my favourite scout though.
      "T1 bots...the only scenario where comparably good DPS and speed matter is when raiding poorly defended structures or attacking units that have no ability to defend themselves against land attacks"
      The main use case for T1 bots (with the potential exception of the flare) is to attack units that cant defend themselves (engineers), so speed is very useful

      "T2 heavy tanks: UEF Pillar...Why does it have the lowest amount of HP...Does it really tie well into UEF doctrine?"
      Compared to the other factions' tanks the pillar has more health per mass cost so you can still argue it fits with the UEF doctrine (i.e. a UEF T2 tank army will have more health than a mass equivalent tank army from another faction).

      "UEF Parashield is straight up worse than Aeon Asylum" If UEF's slight faction identity is units that are generally 'tankier', Aeons is its focus on shields (shield instead of nano for its ACU, shielded T2+T3 tanks, T2+T3 shield generators that focus on health over size), so it's reasonable to me that the 'shield' faction has a stronger shield.

      (Titan+Percy - others have already covered)

      "UEF Fatboy is a very weird and too much of a situational experimental compared to something like Cybran Monkeylord"
      It's relevant for longer in the game than a monkeylord which is usually good as an early experimental and/or to try and surprise the enemy. If you can protect it from air and TML snipes then it ends up being quite a good value over time unit, and can force the enemy to either retreat to T2 arti, or push in (so provided you have enough percies supporting your fatboy that can be good for you). It's not that strong, but it still has a use.

      "Tier 3 stationary shield generator" - Seraphim costs the most but yeah generally Seraphim has the best shield, and then the other factions have different focus (UEF on size, Aeon on strength, Cybran has the worst shields but at least have a very cheap t2 shield which also has use)

      As to whether units follow the suggestions of a UEF doctronine/faction focus, agreed there are some exceptions.

      posted in Balance Discussion
      maudlin27M
      maudlin27
    • RE: Main problem of Supreme Commander

      @thinker I don't see the benefit of slowing the game down significantly as a blanket change - there are already many maps where the game is decided at the tech1 stage (e.g. 5km 1v1) where someone who eco's early is likely to be punished.
      Where slowing down gameplay might have a beneficial impact is on larger maps (40km+) since the FAF gameplay on those strongly encourages focusing on a game-ender or experimentals (due to how long it takes for units you build at your base to reach the enemy/do damage - so there's far less point to T1). I could see the use case of a mod that adjusts FAF balance to work better on such maps by making the T1-T3 stages more important, although I'm not sure what the precise changes would need to be (as a 'finger in the air' I'd guess doubling costs and build times for all HQs and eco buildings might help). However from the description of your changes I dont think it'd help much - you're increasing only the build time, meaning you have a slightly larger cost to build some engineers to assist those buildings with upgrading (e.g. 8 T1 engineers mean a t1 mex upgrades at the same rate to t2 as before; you only have to build those 8 T1 engineers once though, and it shouldnt take that long to get; meanwhile t1 mexes build fast enough that even with 5x you could probably just have 2 engineers build t1 mexes in pairs and quickly get your initial mexes in your base area to fund more engineers). I.e. since the mass cost isn't different I feel like it'd only slow down things slightly, and still lead to 'rush experimentals/game-ender' on an 80km map.

      posted in General Discussion
      maudlin27M
      maudlin27
    • RE: ACU TML too strong for short-range combat

      Unless the enemy ACU is stationary there's no guarantee you'll hit - it's challenging to time it right for a moving target that's moving predictably, let alone unpredictably, especially if the ACU is operating close to gun range (as opposed to much closer than that). By going TML you'll be forgoing defensive upgrades that make the ACU much stronger in combat. A mobile shield can also potentially block the TML missile (unless the ACU is operating close to the edge of the shield bubble). It also will only snipe them if they're sub-6k health limiting its use to either firing when it wont snipe them to try and put some damage on the ACU and force it to retreat, or waiting until they're low enough that they either die or are so low your ACU/units can kill them before they retreat. Finally, the enemy should already have a good idea you have a TML ACU since chances are you'll have used it to attack mexes (in part as it's not good enough given the above limitations that you'd get TML over shield/nano to just use in combat); so they should know to be careful getting too close to you, and to not keep their ACU stationery if 6k unexpected damage on it would lose them the game.

      I'd hate to see this eliminated, since it's great to see when it's attempted (whether it's a success or not) in casts, and for the ACU TML user provides an interesting risk-reward option - the closer you get to them the greater your chance of hitting, but the greater the risk you die if you miss, while to a lesser extent the person considering attacking the ACU also has to weigh up the risks of say getting closer to their ACU when they think they can get a kill (to make it harder for them to retreat/get out of range) vs the increased risk of a TML hitting.

      That said, I'm not saying the TML upgrade as a whole isn't powerful, just that this particular potential usage of it IMO makes the gameplay more varied and interesting (so if TML upgrade was considered too powerful as a whole taking the more niche ACU snipe potential into account, I'd rather the upgrade was made a bit weaker more generally by say increasing the cost; also because the proposed changes like speed and damage would make it inconsistent with a normal TML which wouldn't be intuitive)

      posted in Balance Discussion
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      v227 Update
      13 changes, including:

      • Various adjustments to air anti-air (i.e. asf) logic, increasing the number M28 wants to think it has air control (including a bugfix) and having them be more likely to follow strat bombers to protect them
      • If M28 wants to build an experimental, but first wants more T3 units, and lacks power, it should be more likely to build a pgen instead of a land factory
      • Fewer pre-emptive SAMs should be built if the enemy lacks a nearby air threat and M28 could be spending the mass ecoing.

      Acknowledgements

      • Vortex – QUIET v225 and v226 replays
      posted in AI development
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      v226 Update
      12 Updates, including:

      • Refinements to Inties/ASFs to make them more cautious if the enemy has an air threat in an adjacent zone to the one it plans on defending, or if lots of airAA units have been sent for refueling
      • Several tweaks to the first bomber for some rare cases where they'd get stuck not attacking a unit, and to consider hover-bombing before firing their first bomb
      • ACUs should be more likely to run from approaching land experimentals
      • 4 mod specific changes, including removing AA threat from ACUs in QUIET, recognising if bombers in QUIET/LOUD fire salvos, and increasing the AirAA M28 wants to defend friendly targets

      Acknowledgements

      • Vortex - QUIET replay on big wonder
      • Little Tank – Replay and highlighting a strange M28 message (where it just said “1” in chat, when it should’ve been saying it had gifted a mex)
      • Zhanghm – Log with some error messages relating to a unit mod
      posted in AI development
      maudlin27M
      maudlin27
    • RE: Trainer Team 2025 - The Wiki | Part One

      Thanks for the feedback, we’ve made some adjustments to clarify the references to monkey spam accordingly

      posted in General Discussion
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      v225 Update
      17 changes, including:

      • New game options to make M28's micro more configurable. You can now specify the maximum number of units it can simultaneously apply dodge micro to, and/or hover micro to (previously there was just an option to enable/disable dodge micro, or to use M28Easy to disable). Refer to the "M28: Use hover air micro?" and "M28: Use dodge micro?" settings in Game Options. @Swarm133 this means you can disable its hover-micro with asfs without affecting other micro.
      • Gunships should use a more precise check of if the enemy has nearby groundAA when deciding whether to retreat (or search for further away targets)
      • T1-T2 gunships should be more cautious when taking damage so they're less likely to die (e.g. meaning jesters have a chance of surviving when fighting a single flak)
      • Several refinements to how ASFs decide whether to attack targets, including factoring in nearby enemy ground AA that isn't large enough to make the asfs run in its own right, but when combined with the enemy air force could turn an air win to an air loss
      • Changed the use of markers to a general chat message where long ranged units that M28 hasn't seen start firing at it and it wants to warn its teammates

      Acknowledgements

      • Votex – Highlighting M28 building t3 shields to protect 1 mex from a t3 arti
      posted in AI development
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      @Saver If it's the last game you played then the replay should be available, e.g. in:
      C:\Users\USERNAME\Documents\My Games\Gas Powered Games\Supreme Commander Forged Alliance\replays\FAPROFILENAME\LastReplay.SCFAReplay

      I expect in your case M28 sent an air scout near the enemy units at some point which allowed it to detect them (e.g. since both those units have an anti-air attack, it could've been that they fired this at the air scout; alternatively M28 could've got intel with no Omni coverage, or subsequently stalled E, meaning the unit was visible for a brief period of time to it)

      However in the case of nukes I'll tone down it's approach to highlighting units for enemies so if it doesnt have a visual and it is a long ranged unit (like a nuke) it should just warn in chat (since while a player might have a good idea where the unit is from the direction the missiles/arti shells come from, on some larger maps it can still be hard to infer)

      posted in AI development
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      @Swarm133 Thanks for the feedback; I can look at how feasible it would be to add further customisation into how it uses hover-micro as a game setting (e.g. the number of units it's able to hover-micro simultaneously); in the meantime you can play using M28Easy which should disable all hover-micro.

      For ASF vs ASF combat I've seen it lose lots of air fights that have looked relatively close in separate replays, but there is a high degree of variability in asf matchups even with the same tactics being used by both the player and the AI, so it's certainly possible that you can have more asf and lose an air fight, although I'd expect with a significantly larger force you should overcome the variability/micro difference.

      To help illustrate what I mean, in sandbox you could create 50 asf for two players, and have 1 group attack-move towards the 2nd group, and the 2nd group move and then press stop when near the 1st group; I'd expect the 2nd group to win the fight. M28Easy will disable the main micro M28 uses in asf vs asf fights, which I'd expect to make it weaker (although I've not actually tested to confirm!)

      @Laso Please can you provide a replay - the only scenario I'dve thought M28 would have knowledge of a nuke it hasn't scouted is if that nuke has already launched its missile; M28 aims to approximate what a competent human player can infer by recognising certain units exist when they fire, to avoid scenarios where an unscouted unit can fire at M28 and it continues as though that unit doesnt exist (another example would be a monkeylord firing outside of visual range - a human player would see the big laser and know there's a monkeylord there). I'd need a replay to check or (if it's marking the nuke when it shouldn't) to fix the bug causing this).

      As for the first bomber again I'd also need a replay to assess, but I can assure you it is very much limited by intel, and doesn't see everything despite not having a scout. For example, in the recent rainbow cup semi-finals you can see the dark blue early t1 bomber do a number of suboptimal plays due to its lack of intel:
      https://www.youtube.com/watch?v=GQI_jUrivfk&t=120s

      • Once it gets near the enemy T1 AA (bombers have a radar+visual range although you can't see this in the cast due to observer view) it decides to turn and try and seek a different target
      • It then proceeds to go towards various empty mexes with no units, because it's trying to find enemy engineers to target (2m10 of the replay/2m32 of the video)
      • It's not until 3m20 of the replay (3m47 of the video) that it approaches a location with enemy units, having wasted a long time trying to find some
      posted in AI development
      maudlin27M
      maudlin27
    • RE: Optional Split Teams Option for TMM

      @Sainse said in Optional Split Teams Option for TMM:

      It won’t make it much easier. You underestimate how easy it already is lol. Just alt f4 whenever one group found game but other didn’t. Abuser also doesn’t need to “feed group info”, he can just intentionally lose and it’s game over

      Both of which would be more detectable, and are bannable

      posted in General Discussion
      maudlin27M
      maudlin27
    • RE: Optional Split Teams Option for TMM

      @Firv said in Optional Split Teams Option for TMM:

      you can already do this right now just by queueing up as two seperate groups. So this change would not change anything for that regard

      To not change anything means assuming the two separate groups will always be assigned to the same game which won't always be the case. The change would make things much easier and more reliable for someone seeking to do this. Whether that makes it a big enough issue to outweigh the benefits I dont know, but it clearly makes it easier and increases the risk of such behaviour.

      posted in General Discussion
      maudlin27M
      maudlin27
    • RE: Still being called benny, insulted in every lobby and moderators removing the evidence

      If you genuinely think you're being insulted then you can make a report to the mod team, or create a ticket to discuss further with the mod team (e.g. if the report is discarded due to not meeting the reporting requirements involving sufficient evidence of a rule breach and you want to understand this further).

      posted in General Discussion
      maudlin27M
      maudlin27
    • RE: Optional Split Teams Option for TMM

      I quite like the idea, but my main concern would be the potential for abuse that would be hard to detect (e.g. you're on VC with your group, and one of you gets put on the enemy team, and decides to feed your group info)

      posted in General Discussion
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      v224 Update
      15 changes, including:

      • M28 should place markers when it spots high value enemy units if it has a human teammate
      • Engineers should be more likely to reclaim wrecks within a second of being in range of the wreck.
      • Decreased bomber aggression early game and where M28 lacks intel
      • T2 radar outside the base can be upgraded to omni on larger maps when M28 has very good eco
      • 2 QUIET specific changes, including an infinite upgrade bug and fixing factory enhancements

      Acknowledgements

      • Relent0r - identifying triggers relating to a brain gaining visual of a unit that I was able to use to add in some taunts (although after working on it some more I decided to go with a custom approach for greater functionality)
      posted in AI development
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      @NomenNominandum You can (try using a text marker to ask)

      posted in AI development
      maudlin27M
      maudlin27
    • RE: M28AI Devlog (v229)

      v223 Update
      8 changes mostly relating to improving M28's response to a comdrop attempt:

      • More inties should be built if an enemy T2 transport is detected
      • Various adjustments to inties+asf micro when engaging enemy air targets
      • Emergency PD's build location should be reassessed if the PD is going to be built in range of existing enemies and the engineer is sitll moving to the build location.

      Acknowledgements

      • Chucups replay where 4 M28 lost to a 2 player T2 comdrop
      posted in AI development
      maudlin27M
      maudlin27