Forum Replies Created
-
Posts
-
Yes, a simple toggle on / off for each skill is planned to be added in Advanced Settings.
However this is exactly the same as taking unwanted abilities off your hotbar.The original question is about adjusting weights of the algorithm, which would allow Miqobot to identify whether a certain skill should or should not be used automatically. This feature is technically not possible outside of the source code.
The algorithm we use is called State Space Search with Probabilistic Transition Matrices.
The weights are dictated by actions themselves, and they are hardcoded under several layers of AI. Extracting them into advanced settings would break low level optimizations, increase cache miss rate, and result in computation time burst up to several days for a single synthesis.
However we can adjust weights on development side, and for each new version of Miqobot we release a better version of AI.
The problem is that the current version of Miqobot still uses Heavensward AI for crafting which is not suited for Stormblood recipes. Crafting Solver will behave differently in Stormblood Upgrade Stage 2.
We would kindly ask you to wait until the next version, and then let us know whether Maker’s Mark behaviour still needs to be adjusted.Cross Class Actions for 70***
When we released Crafting Solver support for 60-stars, we posted our analysis on cross class abilities in order to determine which ones work best with Miqobot.
(You can find it here: Crafting Solver for 60****)Just like then, we use our special analytic AI to obtain the most relevant and mathematically correct data.
If crafting rotation would be 100% optimal, how much value would each action contribute?This is the answer for the newest Stormblood 70*** recipes:
1. Ingenuity I or II: 38.9% to average HQ rate 2. Byregot's Blessing: 33.3% 3. Maker's Mark: 26.6% 4. Flawless Synthesis: - (required for Maker's Mark) 5. Comfort Zone: 20.2% 6. Piece by Piece: 13.0% 7. Tricks of the Trade: 9.8% 8. Steady Hand II: 6.5% 9. Innovation: 4.4% 10. Reclaim: - Hasty Touch: 2.0% Careful Synthesis II: 1.2% Rapid Synthesis: 1.3% Rumination: 0.0014% Waste Not I: 0.00072% Waste Not II: 0% Manipulation I: 0% Careful Synthesis I: 0% Muscle Memory: 0% (3.7% when no Maker's Mark)
As you can see, the crafting system in Stormblood has reached an almost perfect balance.
The most valuable cross class actions fit exactly into 10 available slots on any crafting job. When you reach level 70 you have to setup your skills only once, and no more swapping for Reclaim is required.We would like to mention Ingenuity one more time.
In most cases, it is better to choose Ingenuity I over Ingenuity II due to lower CP cost. According to the table we posted earlier:- Crafter level 70 converts to hidden level 290.
- Recipe level 70** converts to hidden level 320.
- Ingenuity I lowers this level by 28.
- Ingenuity II lowers this level by 29.
Thus, Ingenuity II gives only a small additional boost which can not be justified by extra 8 CP.
Also, please note that Reclaim works best with Hasty Touch, because it’s the only 0 CP action which does not increase progress and therefore can guarantee a failed synthesis. Starting from level 61, Miqobot will be able to use Hasty Touch II in combination with Reclaim, but in this case she must account for additional CP that is needed to fail the synthesis.
- Reclaim is used when HQ rate is expected to be less than required.
- Decision is taken at point of no return.
- When using Hasty Touch, the point of no return is when the next skill will drop CP below 55.
- When using Hasty Touch II, the point of no return is when the next skill will drop CP below 65-70 CP.
(Depending on how many points of durability left.)
So unless you absolutely need that extra 2.4% of HQ rate, we recommend to take Hasty Touch instead of Innovation.
We consider this feature, but we are not sure whether it will be really useful. The problem is quite easy to explain, and it’s not even about market domination (which is also a factor but not the main one).
Let’s take a simple example.
- You instruct the undercutting bot with a minimum of 900 gil per potion.
- Somebody else instructs the undercutting bot with a minimum of 800 gil per potion.
- Both of you start from 1,000 gil but just within a few minutes your bot stops at 900 while the other one keeps going.
Your potions are not selling anymore. - Now you have to look at Market Board manually and evaluate the situation.
You decide to reset the minimum to 700 gil. - The cycle repeats.
So the question is: How exactly is this any different from undercutting without a bot?
Instead of using the Market Board you now use a tool, but since it’s public you have to assume that everybody uses the same tool.We would like to hear more feedback from our users.
How would you solve this dilemma?Crafting Actions
Miqobot is now capable of using all new Stormblood abilities.
In this article, we analyze their respective mathematical value in comparison to their pre-Stormblood counterparts. With the help of Miqobot’s special analytic AI, we can calculate how much each action contributes to average HQ rate.
Hasty Touch II (Lv. 61)
Hasty Touch: 29.6% Hasty Touch II: 21.5% HT vs HT2: 7.2%
The values above say:
– Hasty Touch contributes 29.6% to average HQ rate.
– Hasty Touch II contributes 21.5% to average HQ rate.
– If you drop HT and start using HT2 only, your average HQ rate will drop by 7.2%.New Hasty Touch II does not require a cross class slot, but old Hasty Touch is still better in performance.
While it seems that 90% success rate is better than 80%, the extra cost of 5 CP makes a big difference. Considering that Steady Hand II must be present for both skills, HT costs 5 CP per step and HT2 costs 10 CP per step.
Total CP consumption is doubled. Additional 10% rate is not enough to justify this change.Careful Synthesis III (Lv. 62)
Careful Synthesis II: 19.4% Careful Synthesis III: 9.6% CS2 vs CS3: 9.4%
Once again, old Careful Synthesis II is twice better in performance that its Stormblood counterpart.
If you don’t want your average HQ rate to drop by 9.4%, it is not recommended to dismiss CS2. At least not at level 62.Rapid Synthesis II (Lv. 63)
(40 Dur) Rapid Synthesis: 13.5% (40 Dur) Rapid Synthesis II: 9.7% (40 Dur) RS2 vs RS: 0.7% (80 Dur) Rapid Synthesis: 18.9% (80 Dur) Rapid Synthesis II: 23.6% (80 Dur) RS2 vs RS: 4.2%
This is where Stormblood abilities start to get interesting.
When crafting 40 Durability recipes, RS2 does not contribute anything new and is inferior in performance.
When crafting 80 Durability recipes, RS2 contributes additional 4.2% to average HQ rate and is superior in performance.So can you drop Rapid Synthesis for an extra cross class slot? Yes, you can.
You will lose only about 4% and only on 40 Durability recipes.Patient Touch (Lv. 64)
Patient Touch: 41.5% Patient Touch vs HT2: 25.9% Patient Touch vs HT+HT2: 19.4%
Now you can drop Hasty Touch and forget it ever existed.
Patient Touch is an extremely aggressive skill.
It can make half way through IQ stack in just 3 steps. And even the penalty of losing IQ is not very bad, because reusing Patient Touch after failure quickly compensates for the temporary loss.
In addition, when your IQ stack is odd (3,5,7,9,11) the result of division is rounded up (3->2, 5->3, 7->4). This makes Patient Touch very useful at the beginning, because penalty is less severe.Let’s compare two similar scenarios.
- Steady Hand II -> Hasty Touch (x5).
- Steady Hand II -> Patient Touch (x5).
As success rate is 80% in both cases, we should see one failure on average. Let’s assume it happens on 3rd Touch, where it’s rather bad for Patient Touch. It drops from IQ5 to IQ3 but then quickly builds up to IQ7.
- IQ stack = 4 (25 CP used).
- IQ stack = 7 (55 CP used).
In order for Hasty Touch to catch up, it has to do 3 more successful steps which cost 30 Durability. Even if we use Manipulation II, it still takes 24 CP per 10 Durability. And don’t forget that it needs another Steady Hand II. And it can still fail.
- 25 + 25 + 24*3 = 122 CP used.
- 55 CP used.
So with Patient Touch we saved 67 CP and avoided extra RNG.
Manipulation II (Lv. 65)
(40 Dur) Manipulation II: 26% (80 Dur) Manipulation II: 16.9%
Not a lot to say, Manipulation II is a plain boost to HQ rate.
It’s currently the best CP-to-Durability skill. It’s better than Waste Not II and imposes no restriction on what skills you can use. (Waste Not II dictated that you can’t stop to rebuff Steady Hand more than once and you had to ignore Tricks of the Trade.)The only inconvenience is that Manipulation adds a new layer of stacks, which means increased computation time for Miqobot.
Fortunately, we have reimplemented several core algorithms and managed to optimize Crafting Solver more than 20 times. Now we can introduce new CPU intensive abilities and keep system load at bay.Prudent Touch (Lv. 66)
Prudent Touch: 12.2% Prudent vs Patient: 5.8% Prudent vs WN: 3.1% Patient vs Prudent: 12.6% WN vs Prident: 0.5%
Prudent Touch is another great addition of Stormblood.
It gets along with Patient Touch naturally, because Patient is good in the beginning of synthesis while Prudent gets better towards the end. So neither of them gets replaced.
Also, it is much more efficient than Waste Not and replaces it completely, thus freeing another cross class slot.Focused Synthesis (Lv. 67)
FS: 0.09% FS + Observe Combo: 0.86%
Useless without Observe combo.
Useless with Observe combo.Focused Touch (Lv. 68)
FT: 0.05% FT + Observe Combo: 0.77%
Useless without Observe combo.
Useless with Observe combo.Initial Preparations (Lv. 69)
Initial Preparations: 0%
According to our measurements, Initial Preparations reduces CP consumption by 30% with trigger chance about 20%. It means on average we get a 6% reduction rate.
In order for this skill to be useful on its own, you need a pool of 834+ CP. Otherwise, you invest more than you receive.Specialty: Reinforce, Refurbish, Reflect
IP + Reinforce (25 Dur): 15.1% IP + Refurbish (65 CP): 12.4% IP + Reflect (3 IQ): 10.6%
Initial Preparations is useful only when you’re a specialist.
From the three specialty skills Reinforce is usually the most powerful, because 25 Durability has the potential to become 30 Durability if planned correctly (the last 5 Durability is considered a full step).Miqobot will evaluate all possible cases and will adapt to the situation.
Maker’s Mark and Whistle
Maker's Mark: 26.6% Whistle: 6.4% Whistle (after Maker's Mark): 0.29% Whistle (during 60+ Maker's Mark stacks): 10.2%
Maker’s Mark is an extremely useful tool at 70-star recipes.
It can be combined with Piece by Piece to fill the progress bar almost to completion, and also retrieve used CP by utilizing several Comfort Zone cycles.Whistle system, on the other hand, is penalized by reduced condition rates at 70-stars. In addition, Maker’s Mark removes the opportunity to use Trained Hand which is the strongest skill Whistle system has to offer.
But 2.5-star recipes give Whistle one extra usage.
6377 difficulty results in 64 stacks of Maker’s Mark, which is enough to restore all CP spent on Piece by Piece. But we are still left with 40/70 Durability in the end.
According to our research, the chance for condition to become Good is 10% (11% with Quality Assurance). It means that we encounter about 6-7 Good conditions over 60+ steps, which is just enough to bring the Whistle stack down and restore Durability with Nymeia’s Wheel.Whistle usage requires x13 CPU time in Miqobot.
However we are planning to implement a separate algorithm for Whistle + Maker’s Mark combo at no CPU cost.
(This is scheduled for release in Upgrade Stage 3.)Ingenuity I & II
Crafter 61 -> Recipe 62: 6.7% Crafter 61 -> Recipe 64: 60.5% Crafter 62 -> Recipe 64: 43.3% Crafter 65 -> Recipe 68: 45.8% Crafter 70 -> Recipe 70**: 38.9%
Ingenuity is literally the only way to remove quality penalty imposed by level difference, therefore it’s irreplacable when crafting recipes above your level.
Which is better, Ingenuity I or II?
If you take a look at the table from our previous article, you can see that values in third and fourth column are very similar. In terms of efficiency, it results in almost identical output. Sometimes Ingenuity II may be even worse than I, because of additional CP cost. And in most cases the difference is only by 1-2%.If you have one spare slot, choose Ingenuity I.
If you have two spare slots, put both Ingenuity I & II on hotbar and let Miqobot decide.Please wait until Stormblood Upgrade Stage 2.
It is already being prepared for release.You can read about Bountiful Harvest II here:
Gathering
In addition to full support for underwater gathering, Miqobot will be able to use new Stormblood abilities.
Essentially they are direct upgrades from their Heavensward counterparts, since they cost the same amount of GP but result in improved performance.No modifications to gathering rotations will be required.
New skills will be 100% compatible with the current rotations designer. Whenever Miqobot detects an upgraded ability on your hotbar, she will automatically use it instead of an old one.- Impulsive Appraisal II will be used instead of Impulsive Appraisal.
- Unearth III will be used instead of Unearth II.
- Leaf Turn III will be used instead of Leaf Turn II.
- Bountiful Yield II will be used instead of Bountiful Yield.
- Bountiful Harvest II will be used instead of Bountiful Harvest.
First of all, let us mention (and we will do it only once) that this behaviour of Engage is almost unanimously declared an exploit by Reddit community. However, there has been no official comment from Square Enix yet, and until then we can neither discourage nor encourage its usage.
Please do so at your own risk.Now that it’s out of the way, let us talk about good news.
Of course, we were expecting this request and have already unlocked Squadron dungeons in Miqobot. You can create navigation grids and use all available functions from Scenario Engine inside these dungeons.
Furthermore, you can utilize sendKey() function to target enemies, trigger Engage action from your hotbar, and interact with dungeon NPCs and objects. Our developers confirm that it is possible to create a working scenario prototype with the current version of Miqobot, by carefully planning waypoints and commands throughout the dungeon.The only thing you would probably require our help with is stability. We will have to implement a few additional functions which will prioritize enemy order, handle KO events, and certain other mechanics.
Unfortunately, the exploit has already gone public and extremely viral on Reddit.
Because of this we expect an incoming fix in Patch 4.11 this Tuesday (Oct. 24). There is very little value in doing something that becomes obsolete in two days.So yes, we accept your feature request.
But doing it before the patch is simply impossible. If this behaviour stays intact after the patch and Square Enix confirms that it works as intended, we promise to implement the necessary stability functions in the next version.Until then, here are a few tips that will help you create a scenario prototype:
- Split your dungeon run into several chapters.
A single chapter should reference to a single monster fight, boss fight, treasure chest or NPC interaction. - Each chapter should start with navigating to a certain waypoint. Use goToWaypoint() function.
This way you can easily reset and continue from the point of accidental KO. - To target enemies, use holdKey() instead of sendKey().
This is because of how the game handles target actions. Use delay of 0.25 seconds for holdKey(). - To target enemies, use “Cycle through Enemies” keybind.
- Make sure to position your camera properly so that all enemies are on your screen.
Alternatively, you can try “Cycle through Enmity List” keybind to target monsters which may be accidentally obstructed from view. - Put Engage action on hotbar and use sendKey() to activate it.
Send it about 10-30 times for each enemy. You should find the best matching value for your Squadron DPS. - Use repeatChapter() for as many enemies as there are in the pull.
- To interact with NPC, use “Confirm” keybind.
You might have to do it several times depending on NPC. - When confronting bosses, keep in mind that some mechanics may prioritize you as main target.
Because of this, you should reposition yourself after pulling to bait AOEs away from your Squadron (for example, poison puddles in Brayflox’s Longstop). - Some bosses become invulnerable and will require switching targets.
Since it usually happens at specific HP percentage, you can find the correct time intervals depending on your Squadron DPS.
Of course, if this request remains viable in the near future, all of above will be implemented by Miqobot Team and combined into a single feature so you won’t have to do it manually.
We apologize for temporary inconvenience.
Crafting Condition Rates
As Crafting Solver calculates all possible states and outcomes during synthesis, knowing exact rates at which one may encounter each condition is of vital importance.
Furthermore, Quality Assurance trait acquired at level 63 claims to alter these rates, but doesn’t say by how much.Just like in Heavensward, we use the data we collected ourselves and apply certain filters to weed out unreliable samples. From all possible step transitions we consider only the ones that are Normal condition to something else.
- Normal -> Normal
- Normal -> Good
- Normal -> Excellent
The results are as follows:
Recipe Level Without QA With QA Normal Good Excellent Normal Good Excellent 1 – 54 73% 25% 2% 71% 27% 2% 55 – 59 83% 15% 2% 81% 17% 2% 60 – 60**** 89% 10% 1% 88% 11% 1% 61 – 64 78% 20% 2% 76% 22% 2% 65 – 70 83% 15% 2% 81% 17% 2% 70* – 70*** 89% 10% 1% 88% 11% 1% As you can see, Quality Assurance increases rate of Good condition by 2% on non-star recipes and by 1% on star recipes. Also, it does not change rate of Excellent at all.
So what does it mean when compared to pre-Stormblood rates?
(You can read our previous analysis here: Heavensward Condition Rates.)The good news is that Whistle system finally becomes useful, albeit only for lower 60+ level range. As we approach endgame 70-stars, value of Whistle drops significantly once again.
Before we reveal data on Whistle, we have to mention another important change arrived in Stormblood – Trained Hand.
CP cost of Trained Hand was reduced from 32 to 16 and it no longer requires Steady Hand. Because of faster IQ stack growth made available by Patient Touch, the optimal Whistle rotation now includes 2 usages of Trained Hand:- During normal IQ stack growth, whenever it meets Whistle.
- After IQ reaches 11. At this point you can reapply Whistle and spam Trained Hand several times. Unfortunately by this time you will be struggling for CP so it’s very unlikely for this trick to be performed more than once.
And of course, Trained Hand potency is negated by Maker’s Mark, since it leaves no extra space on progress bar.
Maker’s Mark remains superior by a long shot.Recipe Level Whistle adds to average HQ rate Without MaMa With MaMa 61 – 64 +14.9% +3.8% 65 – 70 +12.0% +2.7% 70* – 70*** +6.4% +0.29% There is one special case concerning combination of Whistle and Maker’s Mark.
More details in the next article.miqoradar.exe is a separate EXE-file which resides in your %temp%\miqobot\ directory.
It’s an essential part of Miqobot.Please understand that false positive is a bug in antivirus.
We have no affiliation with antivirus vendors, therefore we can not help them fix it. If your antivirus does not allow whitelisting for some reason, we recommend that you report this behaviour to their tech support.Miqobot is not a virus.
Please read FAQ at our website carefully: https://miqobot.com/#faqFirst of all, verify that your antivirus software and firewall don’t interfere with Miqobot. Miqobot is not a virus, but antiviruses are usually trained to sic at anything that tries to interact with other programs. Even if you disable them, that might not help, because antiviruses leave background services behind. You might have to uninstall it.
Alternatively, you may try to white-list your Miqobot, but you’ll have to do that after every patch.
Ingenuity Math
According to skill description, Ingenuity lowers recipe level, but it does not say how much.
Neither does it mention the fact that crafting math is based on hidden recipe level, which is not visible anywhere in Crafting Log. Hidden level is a property of recipe just like item level is a property of gear. They are very close but do not match.We would like to share the data we’ve collected during our research.
Values are exclusively tabular, there is no general formula that could describe dependencies between them.
For the sake of clarity we shall omit certain exceptions, but keep in mind that some recipes may match by their recipe level but differ by hidden level. Miqobot keeps a full database of hidden levels and is able to identify them correctly.Recipe Level Hidden Level Ingenuity Ingenuity II 10 10 5 3 20 20 15 13 30 30 25 23 40 40 36 33 45 45 40 38 50 50 45 42 50* 55 50 47 50** 70 51 48 50*** 90 58 56 50**** 110 59 57 51 120 100 99 52 125 102 101 53 130 110 109 54 133 111 110 55 136 112 111 56 139 124 123 57 142 130 129 58 145 135 133 59 148 139 136 60 150 140 139 60* 160 151 150 60** 180 152 151 60*** 210 153 152 60**** 250 154 153 61 260 240 239 62 265 242 241 63 270 250 249 64 273 251 250 65 276 252 251 66 279 264 263 67 282 270 269 68 285 275 273 69 288 279 276 70 290 280 279 70* 300 291 290 70** 320 292 291 70*** 350 294 293 Crafting Math
Stormblood has significantly changed the crafting formulae for progress and quality.
Changes concern only cases when recipe level is above crafter level.
Unfortunately this includes all endgame recipes: 70*, 70**, 70***, and 70****.While crafter level is capped at 70, recipe level keeps growing with each new star added.
(These are approximate values but they are very close.)- 1st star = +10 levels
- 2nd star = +20 levels
- 3rd star = +30 levels
- 4th star = +40 levels
Level modifications are combined together for each star. So when you’re crafting a 4-star recipe, in fact you’re crafting a recipe of +100 levels above yours.
Before Stormblood, it affected progress mostly. Quality efficiency received very little penalty.
After Stormblood, it has been turned upside down. On average, progress efficiency is increased by +30%, and quality efficiency is decreased by -60%. Progress is still affected by level penalty, but not as much.
As a result, crafting HQ items above your level is much harder than it was before.In order to perform correctly, Crafting Solver must know exact formulas used by the game.
Unfortunately they can not be found anywhere in the game client, since all math is evaluated server side. The only way to find them is to collect many data samples at different levels with different stats, and then use clustering and regression methods to deduce the formulas.Over the past months, we have collected about 500,000 samples of data and managed to reach a rather accurate representation of crafting math.
The formulae itself is omitted in this article, but you can see it in Miqobot on monitor tab.This is how progress and quality are affected by level difference.
Level Diff Progress Efficiency Quality Efficiency 0 100% 100% +1 97% 95% +2 94% 90% +3 91% 85% +4 89% 80% +5 87% 75% +10 75% 50% +30 74% 50% +60 73% 50% +100 72% 50% -1 105% 100% -2 110% 100% -3 115% 100% -4 120% 100% -5 125% 100% -10 136% 100% -15 146% 100% -20 150% 100% -80 154% 100% -100 155% 100% In addition to this, quality is also affected by recipe level itself.
Each level tier introduces a 1% penalty (with a few exceptions).Recipe Level Quality Efficiency 0-50 100% 50* 98% 50** 98% 50*** 97% 50**** 97% 51-60 96% 60* 95% 60** 94% 60*** 93% 60**** 92% 61-70 92% 70* 91% 70** 90% 70*** 89% In this light, Ingenuity functions like a second Great Strides and becomes an imperative ability to succeed on 70-stars.
That’s why in addition to all Stormblood skills we had to design an extra module that is now responsible for Ingenuity calculations.
