Best Of

Hello everyone.

Recently posted some thoughts in Discrod but then I was asked to create a forum thread.

I conducted several tests on the item grading system and was quite puzzled by the data I received. I want to share a brief summary with you and hope that someone can confirm or refute my understanding of the "quality" mechanics.
The quality system is based on a linear dependency, where grades range from 0 to 600, spanning from common to legendary. Each hundred marks a change in color: 100 is white, 200 is green, 300 is blue, and so on.

I managed to create a model for some items that predicts the outcome with fairly good accuracy, so I want to use it as an example.

Let's suppose - creating an item requires 4 ingredients in different proportions, for instance, 1+1+10+1 = 13. In this case, each ingredient would account for approximately ~8% (100%/13) of the total. Depending on the color, a grade coefficient is applied to the item.
So, if I take 3 legendary ingredients and 10 common ones, the final quality would be calculated as follows:

3 * 8% * K_leg + 10 * 8% * K_com = 144% + 80% = 224%
where 
     K_leg = 6 for legendary items
     K_com = 1 for common items

This would result in a green item with an internal quality offset of +24% from the possible range of values(RoV).

Let me clarify upfront: this model currently does not account for certain variables present in the game, as I haven’t figured them out yet. When creating the same item with a total quality of exactly 400% by rearranging some ingredients, I sometimes get a heroic item with a minimal range offset and other times a rare item with a maximum offset.

I suspect the issue lies either in the type of data used for the formula on the game-backend or in floating-point arithmetic, which introduces a bit of chaos into the predictions. However, there might also be other explanations.

What is the RoV I mentioned earlier?
Let’s say an item has a "power rating" parameter ranging from 100 to 120. A 25% quality within that range would correspond to a parameter value of 105. Similarly, 50% would correspond to 110.

Better quality in this RoV can be increased in two ways: by using even more valuable ingredients or by equipping items that provide a "Craftname Quality Rating." Through experimentation, I discovered that ~20 Quality Rating is equivalent to 1% of RoV, and this upset me greatly. Let me explain why.

Let’s take, for example, the "Apprentice Leatherworker's Shirt" and the "Journeyman Leatherworker's Shirt."

Their Leatherworking Quality Ratings are:
Apprentice: 56–66 for the white grade, 211–275 for the legendary grade.
Journeyman: 92–107 for the white grade, 282–359 for the legendary grade.
The difficulty and effort required to craft these two items are currently disproportionate, but the final difference in terms of position in RoV appears to be very insignificant—only about ~4%.

Of course, when we’re talking about min-maxing, an 8% difference provided by two items is substantial, but is it really that significant in practical terms?
Let’s take a closer look.

Imagine we magically assemble a fully legendary Journeyman set (ignoring the rarity of leather) consisting of three items—jacket, pants, and belt—at the minimum RoV point. In total, we get:
282 + 307 + 267 = 856 rating, which equals approximately 42.8%. Now, we’re fancy and can afford to craft upgraded versions of these items.

Let’s start with the pants, which have a range of 307–392. With our 42.8%, the new value becomes 343, adding about ~1.8%. A bit of simplified mathematical approximation leads us to conclude that if we spend an extraordinary amount of resources on repeatedly crafting legendary items in leatherworking, tailoring, and armor smithing, we can eventually create a set whose total bonus approaches 50%(though it’s unlikely to actually reach that value).

But what do these 50% mean in relation to legendary items?
Let’s consider something like the Rosethorn Tunic.

Armor: 63–68
Magic Resist: 63–68
Intelligence: 13–16
Magical Critical Chance Rating: 25–31
Mana Regeneration Rating:  20–25
Magical Penetration Rating: 20–31
Wisdom:  3–6

Since we’re slightly below 50%, rounding will likely go downward, resulting in the following stats for our Rosethorn Tunic:

Armor: 65 (+2)
Magic Resist: 65 (+2)
Intelligence: 14 (+1)
Magical Critical Chance Rating: 27 (+2)
Mana Regeneration Rating: 22 (+2)
Magical Penetration Rating: 25 (+5)
Wisdom: 4 (+1)

The result is roughly equivalent to a single +1 enchantment. However, enchanting costs millions of times less than preparing a full legendary-quality crafter set.

Naturally, we wear several items, and each one provides a bit of a benefit for character min-maxing. But even in this case, an approximate increase of +8 Intelligence compared to ~250+ is merely 3%, and the scale of this benefit seems catastrophically disproportionate to the required effort.

I really hope that I’m missing something and that you’ll tell me about it, but for now, I’m in a state of complete confusion. The only consolation is that we’re currently in 1/3 of the game’s progression, and the target items are, of course, level 50 gear. Perhaps a level 50 legendary set will provide +100% to quality, and in that case, it will be an entirely different conversation.

=====================================================
UPDATED 2025/01/16

The initial assumption was incorrect. The game uses a non-linear formula.
So instead of 100%, 200%, 300%... 600% we have thresholds:
1 - сommon
6 - uncommon
16 - rare
31 - heroic
51 - epic
71 - legendary
100 - legendary with best stats.

That means our Crafting Quality Rating would give us different impact on different tiers. If we will use Legendary Journeyman set, for example on uncommon reagents, we will get rare result. But if we will use it while crafting legendaty item, we will get only 30% of possible stat bonuses.

We have some basic calculator: https://docs.google.com/spreadsheets/d/1QbSOHPcWPri-UE-NU_tEiZ1YI3-D-A_b1My-0_WTwYw/edit?gid=843461881#gid=843461881 feel free to use.