Breakdown of Tempering RNG

After seeing all of the complaints about tempering, I came up with an idea to solve it.

Before getting to that, I wanted to see how bad the tempering system really is (assuming no weighting), to better explain why it’s a problem (if it is indeed a problem), and how to fix it.

So, I wrote an application that rolls up to 100 million items, until it gets an item with the correct 2 tempers both with the maximum value (assuming 10 steps on the values. i.e., 2/4/6/8/10/12/14/16/18/20).

So, here’s some results from the application:

Results
====================
Total Rolls: 2
First item with 1 correct temper: 1
First item with both correct tempers: 2
First item with 1 perfect temper value: 2
First item with 1 perfect and 1 correct temper: 2
Item with both perfect temper values: 2
Results
====================
Total Rolls: 5
First item with 1 correct temper: 1
First item with both correct tempers: 2
First item with 1 perfect temper value: 5
First item with 1 perfect and 1 correct temper: 5
Item with both perfect temper values: 5
Results
====================
Total Rolls: 64
First item with 1 correct temper: 1
First item with both correct tempers: 27
First item with 1 perfect temper value: 17
First item with 1 perfect and 1 correct temper: 52
Item with both perfect temper values: 64
Results
====================
Total Rolls: 31
First item with 1 correct temper: 1
First item with both correct tempers: 1
First item with 1 perfect temper value: 3
First item with 1 perfect and 1 correct temper: 3
Item with both perfect temper values: 31
Results
====================
Total Rolls: 78
First item with 1 correct temper: 1
First item with both correct tempers: 5
First item with 1 perfect temper value: 1
First item with 1 perfect and 1 correct temper: 10
Item with both perfect temper values: 78

Honestly, after seeing these results I actually don’t think there’s that much of a problem.

For clarification, the application does the 2 free rolls first, and then only rolls on a temper if it’s not “perfect”, and it does it alternatively between the 1st and 2nd temper. Read into that what you will.

It also chooses a random item with 0 to 3 GA, which increases the number of roll attempts per item up to 3 additional rolls.

In all tests I’m assuming that both temper recipes have 4 options, and that there are 10 values from worst to best for each. Granted some only have 3, some have 5 and a few (although the ones that you less likely care about which option you get) have 6; 4 seems like a good middle ground for testing.

I’m not sure how many steps actually exist in the game, it could be 5, 10, or 20 for all I know - I’m sure someone does know this information and I can re-run with the value easily once I know the number of steps.

You can see from the above results that getting both ‘correct’ tempers on an item, even when trying to reroll for a perfect stat each time you normally get it within 5 items. 27 is an outlier, but I would need to run this thousands of times and collate and average the data to give better stats (which I’ve not yet yet).

Now, assuming that people think based on this data that tempering is still a problem, because I mean who want to reroll 78 items in order to get both tempers with perfect stat rolls (especially on 3GA items), there’s a potential solution for this. Of course, bricking an item should still be a thing, but increasing your chances of getting what you want should also be a thing, and doing it efficiently is kind of the goal here.


The Solution

When visiting the Blacksmith, you should be able to place your item in and choose “both” of the temper recipes that you want.

Then when you roll, you roll both tempers with a single charge, effectively doubling the number of temper rolls you have, and instantly halving the number of times you have to roll to get a perfect temper.

In addition, when you get one of the tempers you want (even if the roll isn’t good), you can “lock” that temper, and it will still reroll it (to try and get a better value) but it will not replace it with an alternative temper.

This will give you a massively increased chance at getting a perfect or even good roll on each temper and allow you to use ALL of your temper charges without risking bricking the item once you’ve got the right temper affix.

The last change would be that the blacksmith REMEMBERS your temper choices for the item. You can obviously change your mind if you want, but once you’ve made your choices, you can just hit “Temper item” or whatever the button may be, to reroll it. No separate popup/animation, no having to go into each temper option and select the recipe, just simply hit the button to try again.

This will likely mean that you can get the perfect tempers on any gear in a dozen items or so and will almost certainly allow you to get both of your tempers to be the ‘correct’ one in 1 to 3 items.

The benefit here is that your 3GA items will more than likely get the ‘correct’ tempers, they just might not necessarily be perfect. You can then use that item and go farming for another one in hopes to improve the rolls.

If the Scroll of Re-Tempering just increases the number of rerolls on the item to 5 + number of GAs, without wiping the existing temper choices/values etc… Then you’re even more likely to get what you want.

I don’t think this is “necessary”, I don’t think Tempering is broken, but enough people have a problem with it that I think this solution will generally make people happy, without making it so easy that the perfect rolls are guaranteed, and that bricking no longer happens.

2 Likes

Did you consider that affixes have different weights??? Because clearly some are more difficult than others.

1 Like

Nice info OP thanks. bookmarked.

Pretty cool but unless you know how it is coded in d4, I don’t think it will accurately replicate in game tempering results.

1 Like

yeah i’m still puzzled on a pair of gloves I recently “bricked”
bet you can guess which one rolled 5x of the same.

https://imgur.com/a/vRIDRU8
1 Like

Looks just like my experience lol

They don’t, and yes. I have. My code allows me to add weighting to specific affixes, and the results are not what we see in game.

Although I can’t say exactly how it’s coded, RNG at its core is the same in all languages, and given the results of testing in game, it doesn’t appear that there’s anything layered on top of it.

Of course, if you can provide evidence that a particular aspect weights toward a certain affix (and by how much), I can do testing to see how hard it would be to get the affix you want from that temper, and to perfect it, as well as how many items you’d brick in the process (on average).

Your program is busted. You kept getting the both perfect temper rolls on the last roll. 78 rolls you got it on roll 78, 31 rolls you got it on roll 31. That or I dont get what you are trying to say.

I’ll quote what I feel. At the beginning of the character I get good tempers with my characters, when I change builds too. But it seems that when I reach a point, the rng becomes addictive. I have critical on all my items. But to roll critical now it’s more difficult. I have a volley character and another rapid fire character and I see. That the desired skills invert. It’s not just rng.You have to test after using a lot of tempera.

I feel that as I stop the temper in critical, the game determines that I want that affix and it makes it weighted. I write down some releases. It gets addictive over time. Not all affixes.

It ran multiple times, that’s the results from different attempts. The count is the number of items that would need to be tempered in order to get the perfect roll.

Perfect = Maximum value on the affix you want. i.e., 100% [25-100%]x of whatever.

On the first run it only took 2 items to get perfect rolls, on the last run it took 78.

ITs still broken then. Why is is 2 rolls for the 2 total rolls? 5 rolls for the 5 total rolls? 64 rolls for the 64 rolls total? So you just hit a perfect item no matter how many rolls you do??? I dont get it.

It’s not broken.

The first run I only had to temper 2 items.

The first item got 1 correct temper.
The first item got 2 correct tempers with maximum roll.

The second run I had to temper 5 items.

The first item got 1 correct temper.
The second item got 2 correct tempers.
The fifth item got both tempers with maximum values.

And so on, through each test.

I just think I dont understand your table and what it means.

Change your table to be more meaningful and easier to use maybe?

========================
Total Items Rolls: 20
Tempers Chosen: Elemental Surge - Lightning / Shock Augments - Charge Bolts
Number of different tempers: 6 / 3
Number of items with 1 perfect temper: 3
Number of items with 2 perfect tempers: 0
Number of successes (2 successful tempers): 13
Number of Bricks: 7

I would like to scroll through the subtfugio manual, it seems thoughtful. I’ll write down all the temperings now