dkates
  • dkates
  • Advanced Member Topic Starter
2016-09-22T20:24:06Z
So, having had the chance to do some AI Duels with the latest experimental version of the AI, I've been able to compile a few notes so far.

  • Constellar Pleiades doesn't seem to chain its effect to removal anymore; it did in previous versions.
  • Speaking of Constellar Xyz, I'm not sure, but I don't think Constellar Omega uses its effect when it should. For example, I seem to recall activating Dark Hole when the AI had out Constellar Omega, and it did not use the effect.
  • The unaffected checks for various monster effects apparently don't see First of the Dragons as being unaffected. I've had the AI try to take out First of the Dragons with a number of effects -- just about every Monarch that could possibly try to affect it, Castel, and Moralltach all come to mind, and there may be others. For cases like Caius, it would make sense for the AI to still target First of the Dragons if there is no better target, but the AI is clearly expecting First of the Dragons to be affected when it won't be.


That's all I've got for now; everything else I can think of so far has been recently reported and acknowledged. I will, of course, be sure to report anything else I come across and can reasonably replicate.
bostero-arg
2016-09-22T22:28:54Z
Light and Darkness Dragon's negation effect still applies even if LaDD doesn't have enought points to lose. I mean, if LaDD's ATK and/or DEF is under 500, it does not lose points but still negates effects.
Snarky
2016-09-23T05:34:06Z
Originally Posted by: dkates 


  • Constellar Pleiades doesn't seem to chain its effect to removal anymore; it did in previous versions.
  • Speaking of Constellar Xyz, I'm not sure, but I don't think Constellar Omega uses its effect when it should. For example, I seem to recall activating Dark Hole when the AI had out Constellar Omega, and it did not use the effect.


  • I cannot confirm these. Both Pleiades and Omega seem to chain their respective effects just fine. Are you sure, nothing on your side of the field was preventing the activation of their effects?
    Quote:


  • The unaffected checks for various monster effects apparently don't see First of the Dragons as being unaffected. I've had the AI try to take out First of the Dragons with a number of effects -- just about every Monarch that could possibly try to affect it, Castel, and Moralltach all come to mind, and there may be others. For cases like Caius, it would make sense for the AI to still target First of the Dragons if there is no better target, but the AI is clearly expecting First of the Dragons to be affected when it won't be.


  • I noticed that as well, but First of the Dragons is a problem card for the AI anyway, as the only way to get rid of it would be removal in S/T form, which not every deck has reliable access to. I'll try to improve the handling, but the AI will always struggle with that card.

    Originally Posted by: bostero-arg 

    Light and Darkness Dragon's negation effect still applies even if LaDD doesn't have enought points to lose. I mean, if LaDD's ATK and/or DEF is under 500, it does not lose points but still negates effects.


    I cannot confirm that, either. Are you sure, the effects still get negated? LADD will still activate the negation effect in response to any effect, but if he is unable to lose the stats, his negation would resolve without effect and the original effect would go through.

    dkates
    • dkates
    • Advanced Member Topic Starter
    2016-09-24T16:13:11Z
    I may have to recheck on the Constellar Xyz. If I see it happen again, I'll note it. If not, just chalk it up to me going from flawed memory.

    It looks like First of the Dragons is even more of a problem card for the AI than I thought, since it also seems to mess with the attack logic. It doesn't even behave as if the AI simply doesn't see it as not being destroyed by battle. Frequently, the AI will launch an attack against First of the Dragons, clearly intending to use a stat-changing effect during the battle, then not use that stat-changing effect, leading to pointless suicides. What appears to be happening (from the standpoint of someone who is just observing the behavior, at least) is that when it launches the attack, the AI doesn't realize that First of the Dragons won't be destroyed by the battle, but at the point where the AI would activate the stat-change effect, it rechecks, sees that First of the Dragons can't be destroyed by the battle, and therefore doesn't use the stat-change effect.

    Honestly, First of the Dragons seems like it just may have to be a "special case" for the AI. Basically, make it a higher-priority target for the appropriate S/T effects, blacklist it for the appropriate monster effects (preferably with an exception for if First of the Dragons' effect is currently negated), and code for it in the attack logic (probably the hardest part).

    Incidentally, the deck I've been testing with that includes First of the Dragons also includes Evil Hero Dark Gaia, and the AI seems to handle that just fine just as a result of its general logic. It might not be a bad idea for the AI to give Evil Hero Dark Gaia a slightly higher priority for effect negation (since negating Dark Gaia's effect makes its ATK 0), but considering that Dark Gaia isn't used all that often (and when it is, it's usually as a finisher anyway), that may or may not be worth the effort. Come to think of it, this might be something that can be done semi-generically, depending on what information the AI has access to -- if the AI is able to see that a monster's printed ATK is ?, it should be possible to code things so that it will "know" that negating that monster's effect will make its ATK 0.

    On a much more positive note, I will say that the ABC deck seems to have come out very well. I haven't yet seen it have the opportunity to pull off the Karakuri play on me, but I can say that, from what I've seen so far, it does a good job handling ABC itself, ABC's pieces, and the Transmodify play.
    Snarky
    2016-09-25T09:57:43Z
    There seems to be a problem with the unaffected check in general, which is not limited to First of the Dragons. I started working on a Raidraptor script, and I noticed while playtesting, that Zapdos (Ultimate Falcon) is not detected as being unaffected by everything at all. Strangely enough, he doesn't even have some sort of conditional immunity, that might not be detected correctly by the system, which I thought to be the problem for First of the Dragons. Instead, Zapdos just has plain, unconditional effect immunity, which should make it an easy case for the unaffected check to detect. Except, that it doesn't seem to work at all. I'll have to check this one more thoroughly.

    Originally Posted by: dkates 


    On a much more positive note, I will say that the ABC deck seems to have come out very well. I haven't yet seen it have the opportunity to pull off the Karakuri play on me, but I can say that, from what I've seen so far, it does a good job handling ABC itself, ABC's pieces, and the Transmodify play.



    Glad to hear that. I didn't even spend that much time on the script, and I know, that there is stuff to improve. But considering the limited work I've put into it, I agree, that ABCs seem to perform reasonably well.
    Urban-Shadow
    2016-09-27T13:58:02Z
    Originally Posted by: dkates 

    On a much more positive note, I will say that the ABC deck seems to have come out very well. I haven't yet seen it have the opportunity to pull off the Karakuri play on me, but I can say that, from what I've seen so far, it does a good job handling ABC itself, ABC's pieces, and the Transmodify play.



    the AI did the combo on me first turn first try, amazing to watch, not so much to deal with, lol

    so far the I've not encounter any problem with the update and the decks just run fine on my end
    dkates
    • dkates
    • Advanced Member Topic Starter
    2016-10-13T14:50:43Z
    Got a new one. Only had it happen once so far, but...

    I was running a Monarch variant, and the AI was running one of its decks that can bring out PSY-Framelord Omega. I forget which, but I think it may have been the Lightsworn. I used the banish effect of Pantheism, and later (possibly the following turn, again not sure), the AI brought out Omega. It had no banished cards at this point, and my only banished card was the Pantheism. During my Standby Phase, you would think the AI would opt not to use the effect of Omega to return a card to the Graveyard, but it did, returning my Pantheism. While there are some cases where the AI might want to use Omega to return one of the player's banished cards to the Graveyard, that was certainly not one of them.
    Snarky
    2016-10-15T11:42:30Z
    Originally Posted by: dkates 


    I was running a Monarch variant, and the AI was running one of its decks that can bring out PSY-Framelord Omega. I forget which, but I think it may have been the Lightsworn. I used the banish effect of Pantheism, and later (possibly the following turn, again not sure), the AI brought out Omega. It had no banished cards at this point, and my only banished card was the Pantheism. During my Standby Phase, you would think the AI would opt not to use the effect of Omega to return a card to the Graveyard, but it did, returning my Pantheism. While there are some cases where the AI might want to use Omega to return one of the player's banished cards to the Graveyard, that was certainly not one of them.



    Omega should only activate, if the AI has any cards in the banished zone. Dunno what happened here, I will try to reproduce it.
    dkates
    • dkates
    • Advanced Member Topic Starter
    2016-10-15T16:33:07Z
    Originally Posted by: Snarky 

    Originally Posted by: dkates 


    I was running a Monarch variant, and the AI was running one of its decks that can bring out PSY-Framelord Omega. I forget which, but I think it may have been the Lightsworn. I used the banish effect of Pantheism, and later (possibly the following turn, again not sure), the AI brought out Omega. It had no banished cards at this point, and my only banished card was the Pantheism. During my Standby Phase, you would think the AI would opt not to use the effect of Omega to return a card to the Graveyard, but it did, returning my Pantheism. While there are some cases where the AI might want to use Omega to return one of the player's banished cards to the Graveyard, that was certainly not one of them.



    Omega should only activate, if the AI has any cards in the banished zone. Dunno what happened here, I will try to reproduce it.



    Indeed, that's how I thought Omega was scripted. Considering that the cases where one would want to return the opponent's banished card to the Graveyard are incredibly niche (especially since none of the AI decks run more than one Omega), I wouldn't expect those plays to be scripted. Is it possible that the AI is scripted to want to return Pantheism to its Graveyard with Omega (which would make sense), and that somehow took priority over "don't activate if I have no banished cards" and/or "don't return the player's cards"?
    Similar Topics
    Users browsing this topic