All Jobs Damage Simulator And Gear Sets

using Feb 11b I was fooling around with TP sets for PUP and for some reason it gives me a seki shuriken and ninja nodowa +2 JSE neck. I've doubled checked that I have it set to PUP/WAR and those items don't appear on the gear tab as selectable options.

can anyone reproduce this? i'm leaning towards user error, but posting just in case. the gui differs from what the text prints out.




just noticed that's a nin cape too, and when i hover over the images it does display the ninja equipment stats.


edit: haven't been able to reproduce it so i guess i just flubbed something along the way

So the default set that displays on the front page is almost always a NIN set because that is the job the dev plays most.

In order to change it you have to click the button on the second page to copy the best set to the front page. I can't remember the actual button.

yes this was after pressing the "equip best set" button which confused me, but all seems fine now.

Does the gui_wsdist.exe pull data from the gui_wsdist.py? I'm trying to update Maru Kala to get better results but not having much luck. This is what I have thus far:
elif ws_name == "Maru Kala":
base_ftp = [3.092, 7.5, 11.93]
ftp = np.interp(tp, base_tp, base_ftp)
ftp_rep = False
wsc = 0.6*(player_str + player_dex)
nhits = 2
sc = ["Detonation","Compression","Distortion"]

It does, but if you’ve changed stuff you have to recompile

Is SMN data in yet? (For staff WSs)
Thought I read it was getting added a while back

does this apply to adding an item? shulmanu collar is missing from pup's neck selections and I'd like to add it but not sure how.

You can also install python and run gui_wsdist.py instead of gui_wsdist.exe, which will enable you to make direct edits (so you can break it) to all of the data. Anytime you make changes you will need to close and reopen gui_wsdist.py.


No, the gear.py file is read by the exe when it runs. The best way to make edits is to copy an item that is close(ish) the item you want to add and change\add the data as needed. You will also have to add the item to said item's type list, for example:



Yes? As far as I know, all WS are in the simulator. Avatar's favor and special buffs are not able to be applied (at least not in the GUI).

get_delay_timing.py is using average delay between MH and OH when determining the delay cap, it should not.


It shouldn't divide by 2 here, verified by running this function and making a print statement

With the / 2 you get 0.733 and without you get 1.466 which is correct.

I was wondering why some of my sets differ from what is listed in the "All_Sets" pages. Now, after seeing multiple people question the accuracy of the SIM, I understand. I have just been building and testing in real time. It's tedious, but fortunately, I'm testing DRG. 90% of it's physical WS have STR modifiers. Made it real easy to test and confirm.

huh? it only uses the information that you give it to find sets. garbage in, garbage out. not saying the sim in infallible, but its a helluva lot more accurate than some gut feelings.

perhaps the biggest weakness of the sim atm is the lack of enemy stats for things like Odyssey and Sortie boss mobs, but I don't know how much testing has been done there. I think someone linked the JP wiki with Sortie boss info before, but I don't remember when/where.

I believe the sets listed were derived from the SIM. And after testing them against my own, I got different outcomes. When I say "tested," I mean I went out into the field using the listed sets, fought the same group of mobs for awhile - recording the numbers. Then I switched to my sets and repeated. Same trusts, same buffs, same mobs.

this is the criteria for those sets, if your results are significantly different, be prepared to show your work is all I'm saying.


so even in the mid-buff scenario you're getting hm + 2 miunets and crooked chaos, and if you're just using trusts your attack is going to be much much lower. if you run the sim and input your trust buffs instead, I guarantee you won't get the same sets posted on wiki.

No, the "gui_wsdist.exe" file compiles the weapon skill files with the GUI itself. If you make any changes, you'll need to recompile the "gui_wsdist.exe" file or run the "gui_wsdist.py" as you would run any other Python file with Python installed.

Unlike the weapon skill files, the "gear.py" file is read in dynamically each time you open the GUI. Any changes you make to the "gear.py" file will be included in the GUI through the "gui_wsdist.exe" file when you reload it. See drakefs' post for how to add new/custom gear/food to the code. I also posted about this some time ago, but I can't seem to find my semi-detailed guide on adding new gear.

I believe I added SMN stats a while back. Most staff weapon skills are included too. Some jobs are still missing correct stats, such as BST main, and BRD/BST/SCH subjob stats. These have comments to remind me to update them later in the "create_player.py" file.



The code's "Time to WS" value being less than 1.0s for high buffs was suspicious... I'm surprised that an issue as obvious as that survived for so long.

I've corrected the issue and updated the two places in the code that call this function. This issue affected the time between attack rounds for dual wielding situations; H2H, 1h, and 2h melee situations should be unaffected. Weapon skill damage calculations were also unaffected. See the Github page or actions page to download the updated "2024 June 29a" version of the code.

I ran a few generic 0% DT NIN dual wield TP sets through the updated code and compared it to the old code and found largely no change. Haste capped situations found identical sets, but the 30% magic haste set found a "better" set that was 2ms faster than the previous set (7.195s vs 7.197s to reach 1900 TP) and the 0% magic haste set found a set that was 143ms faster (8.804s vs 8.947s per 1900 TP). It seems the correction is most significant when your TP set relies on dual wield to reach the delay cap, but it doesn't seem to be a problem unless you have near zero magical haste.

Let me know if you find any other issues, or if I broke something with this most recent change.


While I appreciate your concern for the reliability of the simulation's outputs, it is largely unwarranted these days after all of the updates the tool has had based on community feedback and suggestions. With the community's help, this tool has become the most accurate(?), complete, and convenient publicly available open source tool for estimating damage dealt for FFXI. The reason that your sets may be different than what the tool is showing is most likely a lack of proper sample size or improper (or generally unrealistic) testing on your part.

For reference, your new DRG community guide currently states:
However, an experienced player would immediately recognize that your "best" Savage Blade set (which is identical to your best Camlann's, Stardiver, Impulse Drive, Double Thrust, and Leg Sweep sets) is likely flawed as it does not use Moonshade Earring. In fact, only swapping the Sherida Earring for Moonshade Earring in that set increases its simulated average damage by 18% in standard buff situations (see this imgur link), while other changes can further improve it depending on the situation. I would not trust any "testing" that failed to identify such a simple huge improvement. At the very least you must state what testing conditions you used to find your "best" sets.

I should note for fairness that the sets I've posted to BG Wiki have not been updated in over a full year (DRG sets last updated 2023 May 14). During that time the simulation tool has gone through many small corrections. I should spend some time over a week or so to update all of the sets I've posted to BG Wiki. I would bet that at least a few sets I've posted would show small changes.


Let me know if you find reliable stats for Odyssey, Sortie, Omen, etc. so I can include them. The enemy stats can be manually input anyway, but I agree it would be nice to have common enemies pre-loaded for easy access.




In other news:
I've added a "Print All Stats" button to the "Stats" tab, which simply prints the full list of stats being used for the player Python class in alphabetical order. There is no context for these, so it may be confusing at first. For example: there will be stats such as "Accuracy," which represent the total "Accuracy+" obtained from gear/traits/buffs which applies to both hands when dual wielding. In contrast, "Accuracy1" and "Accuracy2" represent the final aggregate accuracies for the main/off-hands, after adding all sources together. This should help with looking for specific stats that don't fit in the Stats tab page (such as OAX) and for debugging any potential issues.

Note that stats which are not included in this output list are not used by the current player (you won't see OAX in this list unless you have some OAX gear/buff). Similarly, not all stats shown are being used. For example, NIN/WAR has a Smite trait, which shows up in this output but isn't used unless the conditions for Smite are met.

Yeah they won't be getting the correct "bis" sets as he will be attack starved using Gleti's as theirs is mostly r0 and their Nyamr is r25 which is a massive attack increase.

This is why other people are getting better numbers with the correct sets and buffs but he's saying his testing is making his sets do more damage.

trust/low buff sets are still valuable, just gotta be clear that that is what they are and not try to compare apples to oranges etc.

Oh, no, they were talking about them being BiS sets for the DRG guide they are writing. When they used the actual BiS sets they claimed they were doing less than their BiS sets.

https://www.ffxiah.com/forum/topic/57911/a-collection-of-sets#3709344

Thats the thread.

For dual wielding, the minimum delay should be 20% of the total delay. For Hand-to-Hand (H2H) combat, the minimum delay is 480 + Weapon Delay.

Examples:

1. Hand-to-Hand (MNK with Godhands):
(480+138)×0.2=123.6

2. Dual Wielding (Heishi and a 190 offhand like Taka):
(227+190)×0.2=83.4

With capped magic haste:
(1−0.35)×(1−0.6875)=0.203125

This matches the need for 1 DW to hit the 80% reduction:
(1−0.36)×(1−0.6875)=0.2


Ignore above, I was looking at your old repository. However I did find that Vajra was applying WS bonus to Pyrrhic Kleos instead of Terpsichore, added that commit to my dictionary proposal. I also added AGI for those that had comments that noted they were a guess as well as proposed a dictionary setup for that file as well.

Couple needed changes I found:
1 -- Shining One's stats in gear.py should be 20 DEX, not STR
2 -- Ambuscade armor set bonuses are overstated by giving the bonus an extra time. E.g. Flamma +2 set actually gives +8-32 str/etc for 2-5+ pieces worn. Current code instead gives 16-40 for 2-5+.
[Fix: Subtract 1 from the "set_count" values before multiplying by the bonus constant(8) in lines 671+ of create_player.py.] I spot-checked the abjuration set bonuses and those looked ok but it wouldn't hurt to get another pair of eyes on that whole section.

Separately, I noticed Ginsen isn't in the gear list. It's not amazing but is a decent and easy-to-get TP ammo item. It happened to be the only thing I had available for my new SAM which is why I noticed. Is the criteria for gear.py only those items which have a chance of being top-tier? That could explain it.

True about Ginsen, and this alljobs 1% Double Attack piece https://www.bg-wiki.com/ffxi/Vanir_Battery

Thanks for catching these issues. I've corrected the issue with Shining One and the Ambuscade set bonuses.

Two considerations for adding new gear are

• Does the item actually contribute comparably to other items, or is the item significantly easier to obtain? I would occasionally whm/nin in segment farms so I try to also consider gear for niche situations.

• When using the "Select ALL" button to select gear to test, adding insignificant gear will increase how long it takes to find the best set since the code needs to check the two-piece combination of that new item with all other items in each other gear slot. Because of this, I try to avoid adding obviously inferior gear that is easily replaced. We can always manually unselect items to reduce the runtime, but that can be annoying with many items to unselect.

Ginsen not being included was simply that I had not thought of that item when creating the file. I don't think Vanir Battery is worthwhile (over White Tathlum or Amar Cluster, for example), but I've included it to the gear.py file with Ginsen. I've also added the Ponente Sash, which was apparently recently mentioned in the RNG thread, even though in their situation both the code and their in-game testing show Sailfi/Fotia win for Sarv. It seems like Ponente may be a decent temporary belt option for something else, though.

Finally, I wrote a simple code to help me rapidly add Empyrean+2 armor variants for the Empyrean +3 gear in the gear.py file. This was pretty easy to do after noticing that the difference between +3 and +2 appears to be mostly static based on the stat (base stats are always +5 across all jobs, standard attack is +10, magic attack is +5, etc). I believe I caught the special cases, but let me know if you see any mistakes since this method was automated for the most part. When using the "Select all/ALL" buttons, only the Empyrean +3 variants will be selected so you will need to manually select the +2 variants if you want to test them.

If you modified the gear.py file yourself, remember to create a back-up of your changes before updating the file so you don't lose your specific changes.

Link to the Github page.

Hi Izanami! Nice to see you again. I'm so happy to read that Empyrean +2s have been added. I had to manually add them for a few of my jobs, so it makes me happy to know I can update the program and not have to do the rest of my jobs.

I was using the program last night, and for some reason, after I accidentally clicked on show final distribution damage, and the window with the chart appeared, it made the main window noticeably smaller for no reason. I know this is a UI issue and you're probably already aware of it, but I wanted to let you know just in case you don't. Never seen it happen before yesterday. Cheers!

Sorry if I might've missed it, but was the enspell feature for Crocea (RDM) added btw? (It's because whenever I simulate it the DPS output is horribly low compared to anything else)

Also, does BLU take into account its damage traits or is it assumed it's traitless?

I've actually never seen this issue and can't seem to reproduce it right now.

BLU uses the spell set from the Zahak Reborn set on the FFXIAH thread. You can see the exact stats that I added to BLU on the BG Wiki page I made for the BLU sets here. The sets on this BG Wiki page almost certainly need to be updated, though.

I've just now added enspells using the damage equations on BG Wiki. This would be a very easy test since it's magic damage, but I am not able to log on to test this myself so I assume the equations on BG Wiki are correct.

For enspell enhancing equipment, I assume the bonus only applies to the hand that wields the weapon with the bonus (Pukulatmuj +1, for example) and that the flat bonuses apply before the percent bonuses. The general equation I used is:
Composure adds +200% to enspell damage when selected. RDM job points add +23 flat damage to enspells. I also include the enspell+15 from merits since I assume that if you're testing enspell damage, you'll want max enspell potency. This also lets me add the enspell+5 to the relic+3 pants augment.

Finally: the estimated DPS simulation output now splits the average damage per TP attack round into physical and magical damage. RDM has also been given a new spell "EnSpell" at the top of its spell list. This can be used to print the average EnSpell damage and TP return for a single successful attack (not a full attack round, no multi-attacks, and ignoring accuracy checks) for the main-hand only.

Let me know if I messed something up. My quick tests found no errors and produced reasonable values, but it's probably pretty easy to miss something simple.

Thanks Izanami, I tested the enspell and it seems a lot more reliable with the numbers so I think this works well!

and great to know about that consideration about BLU and that a relevant trait set is being emulated.

So far this app has saved me weeks / months of potential grinding, and will continue helping me judge certain gear so really appreciate all the effort you put into this.

Hey Izanami! Thanks as always for keeping this updated. Much appreciated.

Couple questions -

1. Do the Prime weapons have Occasionally Deals Double Damage / Occasionally Deals Triple Damage hidden effect? I believe it's 30-33% proc rate for both stages. I know there might be some questions regarding if the proc rate differs between 1H vs 2H. See - https://www.ffxiah.com/forum/topic/57318/prime-ws-information-testing-discussion/63/#3700922

2. Are Prime Aftermaths for PDL applying? See - https://www.ffxiah.com/forum/topic/57318/prime-ws-information-testing-discussion/35/#3680658

3. Are there plans / is it feasible to include additional party buffs, specifically from DNC main including Haste Samba, or SMN buffs?

1. I remember specifically not adding the Prime weapons' auto-attack "Occasionally double/triple damage" effect back when they were new since the effect was not well characterized yet. I'll add this quietly later and add an "Edit:" line to this post when I do, but I'm traveling soon so if it isn't done by Monday, then it'll have to wait.

If you wanted to add this yourself, then you should look at line 307 (and similar lines) of the actions.py file with the Relic/Empyrean effects.


2. Prime aftermaths are applying, but only if you turn on aftermath from the drop down menu where you select the weapon skill and player job. I believe that, in game, the prime/REMA aftermath effect applies to the weapon skill that activates it. If you want to test prime weapon skills properly, then make sure to turn on aftermath before you check the outputs.

Having the aftermath value change automatically based on the weapon equipped, weapon skill name, and TP value was surprisingly annoying to add, so I never got around to finishing it.


3. Somebody else requested this a long time ago for SMN, BLU, and BST. Adding more universal buffs should be easy enough. The only "issue" is that then the buffs list would become mostly filled with buffs unrelated to the player's main job, which may make it annoying to scroll through just to find the one buff you want enabled. I'll play with this idea again soon.

Is there a convenient list of desired buffs and debuffs that would be beneficial to add?

Awesome, thanks for the info.

I'm not aware of an existing list of buffs. For party buffs / additional buffs, I had the following in mind:

Haste Samba from DNC main - 10% JA haste
Haste Samba from DNC subjob - 5% JA haste
Box step from DNC main - 23% def down (assuming 10 stacks)
Box step from DNC subjob - 13% def down (assuming 5 stacks)

Mighty Guard from BLU (can be self or AoE) - 15% magic haste
Nat. Meditation from BLU (can be self or AoE) - 20% attack

Corrosive Ooze from BST jug pet - 33% def down
Rage from BST jug pet (applies to BST only, not party) - 50% attack
Frenzied Rage from BST jug pet (applies to BST only, not party) - 25% attack (not listed on wiki, see - https://www.ffxiah.com/forum/topic/45830/killer-instinct-the-beastmaster-compendium#enhancingready)

SMN Favors (assuming +9 Favor)-
Ifrit Double Attack - 25% DA
Shiva's Favor - 39 MAB
Ramuh's Favor - 23% Crit hit rate

SMN Blood Pact Ward -
Crimson Howl - 11% attack (cannot be used with Warcry from WAR)
Crystal Blessing - 250 TP Bonus

Edit -
WAR Party Member
Warcry - 700 TP (Assumed 5/5 merits & reforged Relic head)
Blood Rage - 20% Crit hit rate (cannot be used with Warcry or Crimson Howl from Ifrit)

While I was adding this to myself, I found a small error with relic hidden proc. Not all relics has ODT with 13% chance. Some of them have 16% for x2.5 and some of them have ODD with 20% chance.

Another error is that Relic ranged weapons also only proc on first hit and you coded it on all hits. I haven't fixed that, because I don't care about it for now.

Here is the code, that fixes melee relics and adds ODD/ODT for primes, if anyone wants to update a local version.