Update on crafting solver?

[rileyParticipant]

Hi Miqo,

Just curious if there are any updates on the crafting solvers progress? It’s been a while since we heard anything but from the last post it seemed pretty close to completion. Was hoping to mass stockpile some mats before 3.4.

[MiqobotKeymaster]

Yes, Miqobot Team is working day and night to deliver new version as fast as possible 🙂

Solver algorithms are stabilized and being prepared for release.
Here’s a preview of new UI for Crafting Solver:

Thank you for your concern very much!

[Carl ArbogastParticipant]

Wow nice, it looks so complete, all those CPU intensive options are great, I’m always glad to be able to choose the most advanced algorithm regardless of the time spent to solve the craft.

I hope we will get that before 3.4 get released, in order to maximise the gils income on day one.

[rileyParticipant]

Agreed it does look amazing. Thanks for your continued hard work.

[MiqobotKeymaster]

We would like to share some information on new options coming for Crafting Solver.

CPU Intensive

This section refers to abilities that require a lot of computing power but give very little increase to Quality bar in return.

First of all, each crafting ability with buff stacks multiplies the complexity of solution by its maximum number of stacks.
For example, Innovation has 3 stacks. Therefore the amount of possible states during synthesis is multiplied by 4 (since 0 must be considered too). But in practice, especially when crafting 60*** recipes, Innovation is used mostly as filler to bait Good condition for Byregot’s Blessing, and in the end it’s only slightly better than Observe. According to our research, having Innovation in your arsenal gives you about +2.8% to average HQ rate.

Same story with Manipulation. This one is even less useful for 60*** recipes, because Master’s Mend II becomes superior at 70 durability. Miqobot predicts +0.41% to average HQ rate when using Manipulation.

Don’t get this wrong, these skills can still come in handy in special cases, such as 35/40 durability recipes or low level character leveling (when you simply don’t have any alternatives). But when it comes to end-game recipes, all abilities can be sorted in a strictly decreasing order by their respective mathematical value.

Byregot's Blessing:   +38.6% to average HQ rate
Byregot's Brow:       +16.3%
Comfort Zone:         +15.9%
Steady Hand II:       +12.8%
Precise Touch:        +5.1%
Innovation:           +2.8%
Waste Not:            +1.7%
Waste Not II:         +0.42%
Manipulation:         +0.41%

So whenever you find yourself wondering which cross ability to drop for Maker’s Mark + Flawless Synthesis, you can use this information as a reference.
Please note that this data is slightly outdated because we had collected it before final optimizations were implemented. We’ll be recalculating all abilities for the final release and will post a complete reference table in the Guides section.

For CPU Intensive settings, we would recommend keeping Innovation enabled and the rest of them disabled (just like it’s shown on the screenshot above).

[MiqobotKeymaster]

Whistle System

We’ve got to admit, specialist skills seemed to be a rather promising dark horse.
But a new fact emerged that turned the whole system upside down – condition rate.

Over the past months we have collected more than 100,000 samples of data. From all possible step transitions we considered only the ones that were Normal condition to something else. (Since Good always transits to Normal, Excellent to Poor, and Poor to Normal, and thus they do not require statistical analysis).
So only these cases were considered:

• Normal -> Normal
• Normal -> Good
• Normal -> Excellent

And here are the results:

Recipe Level  1 - 54:      Normal - 73.5%,  Good - 24.5%,  Excellent - 2.0%
Recipe Level 55 - 59:      Normal - 83.0%,  Good - 15.0%,  Excellent - 2.0%
Recipe Level 60 - 60***:   Normal - 89.0%,  Good - 10.0%,  Excellent - 1.0%

Whistle output depends mainly on Goods and Excellents which are required to bring its stack down. This fact heavily impacts the respective mathematical value of Whistle While You Work.

Good Rate = 24.5% -> Whistle adds +15.2% to average HQ rate
Good Rate = 15.0% -> Whistle adds +5.6% to average HQ rate
Good Rate = 10.0% -> Whistle adds +0.48% to average HQ rate

In other words, Whistle becomes utterly useless when crafting 60*** recipes, simply due to decreased Good rates.

Basically the game forces you to utilize Heart Of Class abilities (for which we don’t have enough data yet, but we believe that it restores nerfed Goods to default rate of 24.5%). But you are not guaranteed to encounter even a single Good during the next 7 steps. It still remains a game of chance. And the additional cost in Delineations renders it a rather mediocre ability.

We expect this data to change in the upcoming patches of the game, so we might go back and reiterate our research once a change in crafting mechanics is introduced.

For now, Whistle system is useful only for recipes below level 55. The fact that Nymeia’s Wheel (which is an essential part of the Whistle system) becomes available only at level 57 makes it a discouraging tool to master.

[Carl ArbogastParticipant]

As always, this is some very interesting data with a relevant analysis. It’s always tricky to find out the percentages the game is using somewhere, but you guys managed to get each time enough samples to remove any doubts, fascinating.

[Author]

Posts