Forum Replies Created
-
Posts
-
And I believe everyone can see today very well where this Assist Mode could be very useful, Eureka Pagos, where everyone right now are destroying their hands by doing their rotations non stop for hours with no breaks for chaining mobs.
I would not mind taking care of moving myself, and switching target myself there, and just let Miqobot doing the rotations, because we really are damaging our hands and fingers with this content right now, which is stupid.As active players, we are also disappointed with the release of Eureka Pagos.
New content is supposed to be fun, at least for the first week. The fact that it becomes extremely grindy in the first hour means that something went wrong in development.We sincerely hope that this is just a simple misunderstanding.
We assume that the only adjustment that was implemented during maintenance on Aug 13 is a hint to players.
Lowering HP of monsters in certain Pagos FATEs can mean only one thing – Square Enix wants players to focus on these FATEs. Perhaps this is a key to increasing the NM spawn rate.This Reddit thread looks very promising: Pagos Bunny Fate Theory
So if I understood correctly, with the Assist Mode, Miqobot would be able to pick up the DPS there anywhere during my rotation, keep a high enough DPS (better than fillers), letting me focusing on dealing with mechanics on my own, until I feel safe to go back on full manual mode?
For example, for BRD, if she sees a Trick Attack, or Chain Stratagem, or Battle Litany, would she refresh DoTs to snapshot them? Maybe even double snapshoting them?
Yes, absolutely.
Miqobot is fully aware of currently active buffs / debuffs and uses a complex priority system for battle rotations. Priorities are calculated on the fly and allow Miqobot to pick up at any moment in the fight, pretty much like Crafting Solver does. Remembering DoT snapshots and analyzing the best moment to refresh them would require some extra work on our side, but it’s also possible.I didn’t tested yet Miqobot with the VPN and Fastconnect. But when I watched her, I’ve seen her spamming keys for oGCD, but not for GCD, and she’s not spamming the keys a lot.
So my question is, would it be enough in my case to avoid her to clip abilities? I’m trusting you guys a lot when it comes to theorycrafting and maths, but I’m so used to spam keys before an oGCD trigger, or even with GCD, that Miqobot look overconfident. For GCD, the reason she’s not spamming is to not queue the GCD in case she would find something better to do?No, Miqobot is not doing it the right way at the moment. Queueing GCDs is indeed better and would result in more DPS. Spamming keys however would not.
We did not implement queueing yet because it was easier to make the first version stable this way.
But of course we will.
For Miqobot it will result in about +2% DPS.Pressing a GCD key about 400ms before it becomes active creates an action queue and sends a network packet, which is the intended way to overcome ping-related issues. It forces an action to go on cooldown as soon as possible both on client and on server at the same time (assuming your ping is lower than 400ms).
Pressing a GCD key more than once within that 400ms window has absolutely no effect.In addition, every GCD or oGCD action initiates a soft animation lock of about 750ms. During a soft animation lock, you can move your character and queue another action but it will not trigger until the lock ends. Some actions create a hard animation lock (like NIN Assassinate) during which you can not move, however its length is also 750ms. The only exception is DRG jumps which are twice longer – 1500ms.
So during any kind of animation lock, the first action you put in the queue is guaranteed to go on cooldown at the end of the lock. The queue can contain a maximum of one action and can not be overwritten.
That’s why spamming keys more than once does not result in more DPS.Miqobot spams keys only because an action queue may be interrupted by lag spikes or certain debuffs (like Paralyze). But she tries to do it as sparse as possible.
At the moment we use an empirical delay of 200ms which seems to work pretty well. During this 200ms window Miqobot does other important things. (Manage squadrons, rescan enemies, plan positionals, etc.)In my case, I know I can double weave with VPN and Fastconnect, but I’m only able to achieve it in openers, or some specific calm moments during a fight, because it requires me to be so quick and to spam so hard, so, my question is, will Miqobot be able to perform double weave constantly on a connection that is at the bleeding edge limit of allowing that?
To summarize all of the above, yes, it’s absolutely possible to perform double weave as long as your ping stays constantly below 200ms. “Constantly” here is the key.
For example, if your ping is 170ms on average but then you experience a lag spike of 500ms, this will inevitably break the double weave and result in clipping GCD. Connection stability is the most important factor.You guys said your theorycrafting allow a 90% efficiency for now, which is already incredible, are you confident you can reach higher efficiency down the line?
Yes, we have a lot of ideas to reach higher than 90%, however we will make no promises about 100%.
Pareto principle is inevitable in this case. The higher in efficiency we go, the more complex algorithms are required for the next +1%. After 96-97% it would be more efficient to implement a custom rotation designer and use it with specific encounter in mind.Would Miqobot be able to refresh the DoTs each time I’m switching on a target when the DoTs are about to fall off?
Would Miqobot be able to see that the 2 targets are close enough to each other to use Rain of Death instead of Bloodleter?
Would Miqobot be able to see that I’m close enough from the 2 targets, that are also close enough themselves, to use Quick Nock instead of Heavy Shot?Yes, yes, and yes.
You can already observe Quick Nock on 2 targets during Halatali runs.I definitely swiftcast protect at the start of dungeons sometimes. It comes off cooldown fast enough that it doesn’t matter and that way people don’t run off without it.
We will implement advanced options to allow Swiftcast customization for personal preference.
However this particular behavior was not planned and is caused by an issue which may potentially lead to more serious bugs. We are glad to know that it hasn’t caused critical problems yet.Thank you for your feedback!
Had a thought. Maybe make use of sprint? Seems like it would speed up the run. Or is that reserved for getting out of a large AoE?
Please enable the following setting: Navigation tab -> Auto-Sprint.
It’s a global setting for all features, but we can see why it’s not intuitive. Perhaps we should copy it on Combat tab for easy access.Thank you for your feedback!
Had an issue (see attached). Not sure what happened. Killed two bombs, went for the Rudis Beak and apparently got out of range and the bot just spammed the attached message over and over. I was on run 4/10. All party members were alive, and my inventory was not full. If that helps at all. I believe I was in the room just before the last boss.
Thank you for reporting!
This bug is caused by insufficient coverage of dungeon area by navigation mesh. Apparently these monsters can roam too far to the edge of the room and we did not encounter this case during testing.
It will be fixed in a couple of days.Edit: Also as a healer the bot will Swiftcast Protect at the start of the first encounter. It’s not really an issue, but it is a very odd behavior that no human would ever do. Just thought I point it out if this is meant to have more human like behavior.
Thank you for reporting!
This is indeed a bug, Swiftcast should be used only for raising and summoning pets.Would you please clarify, have you encountered this issue on Scholar healer?
Does it happen consistently at the start of each dungeon run?Yes, we will implement more fishing functions for comfortable scenario design. We can not significantly alter existing functions since it would break all scenarios created before, but we can add new functions instead.
Scenario Engine is still in Beta stage, therefore all feedback is highly appreciated.
Thank you very much!
