All Jobs Damage Simulator And Gear Sets

In the "Optimize" tab there is a drop-down menu for selecting TP metrics as "time to WS" or other options such as "DPS". Since Izanami has specifically stated that "the code maximizes the average TP gained per attack round", I assume the "time to WS" option corresponds to this, while the "DPS" option may relate to what you just described. But I leave this to Izanami for confirmation.

For some reason on THF, Antitail +1 is in the ammo selection instead of the ranged selection. Boomerangs go in the Ranged slot.

This was mentioned a few pages back when he put it in to the program. Not sure why he went with this, but it is known about.

Izanami:

Noticed while playing around and comparing #s with a friend that the WS simulations provide very different WS avg compared to "quick look WS" (I'm using the beta right now). Any idea why?

This comes back to optimizing three stats at once, which is difficult when only swapping two items. I have no plans to add MEva+PDT+MDT checks simultaneously at the moment.


The code originally only supported Ninja and was written for "throwing" to work from the ammo slot. It was simply easier to include other throwing items in the ammo slot so their "Throwing Skill+" stat could be added in a consistent way as shuriken.


Nope. Sounds like an error. The simulation average and the quicklook average should be close. This must be a recent issue or a rare/niche issue as I remember checking things when I re-introduced the simulation plot to the beta and found things agreed well.

Do you have an example I could check with WS, gear, buffs? This would help figure out where any issues are.

Edit: I'll look into this later today.

main="Masamune",
sub="Utu Grip",
ranged=Empty,
ammo="Knobkierrie",
head="Mpaca's Cap",
body="Nyame Mail",
hands="Kasuga Kote +3",
legs="Nyame Flanchard",
feet="Kas. Sune-Ate +3",
neck="Sam. Nodowa +2",
waist="Sailfi Belt +1",
ear1="Thrud Earring",
ear2="Moonshade Earring",
ring1="Cornelia's ring",
ring2="Epaminondas's Ring",
back="Smertrios's Mantle",

Any WS, 1200-2100 TP. SAM/DRG, ML35, Hasso/Overwhelm, Apex Bats, Dia 3, Songs+8, HMarch VMarch Min Min Aria, Chaos/Sam, Indi-Fury, Geo-Frailty. Crooked + Marcato are used, Grape Daifuku+1.

Quicklook & the chart seem to be in agreement, but the WS simulations shows some really weird numbers when it prints off in the console.

Thanks for the note. The issue was the code not including TP Bonus when running DPS simulations in the Simulations tab. This was a new issue from the 2023 November 10a update since I recently adjusted where TP Bonus was added and forgot to update the Simulations part of the code to re-introduce it properly. The issue has been fixed.

Note that the "Average DMG/WS" statistic printed after running a DPS simulation is based on actual TP received from TP attack rounds and TP returned from WSs. This value will differ from the quicklook average (which assumes one fixed TP value averaged from your min and max values) and the Damage Distribution plot average (which uniformly samples over the range of TP values you provide), unless all of them use the same "average TP" value for their calculations.

To avoid any confusion, I've also modified the DPS simulation output to print the "Average TP/WS" value, which is just the average value of TP used for WSs during the simulation. If you run a DPS simulation and enter this average TP/WS value for both the min and max TP values (to force the average TP value to match), then the quicklook, distribution plot, and DPS simulation values should all agree reasonably well.

are you solely focusing on accumulating dw to reach the 80% reduction cap? if so, stacking dw reduces combined weapon delay which lowers tp return.

In the original Python sets across jobs many showed 2 haste sets once highest DPS and other a hybrid DT set. I am wondering if it is possible to repost those in the BG page along with currently posted WS sets?

Was taking a look at this for RNG, specifically to see the numbers for Arcadian Beret VS Oshosi Mask +1 during Double Shot. It doesn't seem the STP bonus from Recycle is taken into consideration properly for the Arcadian Beret. I did a quick search but didn't find anything yet.

Thanks for pointing this out. There were two issues related to RNG Recycle and how the code handled Recycle procs with Arcadian Beret +3:

• RNG was missing the +25% Recycle from assumed 5/5 Recycle merits.

• I was dividing the Recycle proc rate by 100 twice before calculating bonus TP return.
  

  • With 90% Recycle proc rate and 5/5 merits, you should receive 0.9*50 = 45 TP per shot on average (ignoring multi-shot), but the code was giving 0.009*50=0.45 TP per shot.
  
  

These issues have been fixed in the 2024 Jan 16a version of the beta code. Check lines 1022-1028 of actions.py to see how the code handles TP return from Recycle procs with Arcadian Beret equipped when using the "Quicklook Spell" button. Let me know if you find any other issues.


Other issues fixed sometime in the last month:

• BRD Etudes no longer increase stats when BRD is disabled.

@Izanami

Just an FYI, you have the stats for Terminus as DEX/VIT, it should be DEX/AGI

Thanks for catching this. I have corrected the issue.

I've also updated the WSC and FTP for Imperator (Sword), Fimbulvetr (Great Sword), and Sarv (Bow) based on the values listed on BG Wiki. The code is still missing WSC and FTP values for Maru Kala (H2H), Dagda (Club), and Oshala (Staff). Has there been any testing on these last three? Maru Kala may also transfer FTP, so that'd need to be tested as well.

Oshala: https://www.ffxiah.com/forum/topic/57318/prime-ws-information-testing-discussion/27/#3677869

maru kala didnt seem to transfer from some short testing I was doing on friends char.

Looks like nyame flanchard are listed at 3% haste but in game they say 5%

I just started running this today so please forgive my ignorance.
I'm using 2023 august 06a.

General question: Is this the current version? I saw comments saying some of the Prime WS are included but can't find them.

Is Aria included in any releases?

I'm looking to be able to account for things like Temper 1&2 and Gain Spells. Is there a way to do that?

Can you add subjob traits and buffs? Example /DRK gives access to Last Resort and higher PDL traits for many jobs. /DNC gives a lesser version of building flourish.

Also this is pretty great. Thanks for making it

Thanks for catching this.

I can't fix this for a while, but since it's just a typo in the gear.py file, you should be able to manually adjust it yourself and see the changes in the GUI.

The gear.py file is not included in the pyinstaller command to make the .exe file. Anyone can modify or add gear to the gear.py file to test with the main code.


The original version of the code is not being updated anymore and shouldn't even be linked on the main post of this thread. The "beta" version of the code was last updated 2024 February 11 and includes everything you asked for and more (except /dnc giving flourish). This updated code is in a different repository (https://github.com/IzaKastra/wsdist_beta). Scroll down on its Github page to see screenshots of the updated GUI.

The beta version of the code should have no/few remaining bugs. There are issues with the GUI window scaling when using custom Windows DPI scaling settings. I've been wanting to rewrite the GUI part of the code to allow for varying text sizes and window resizing, which would also deal with the DPI scaling issue. After I've found time to do this, I'll likely replace the original GUI entirely.

Hiya, great tool, I use it a bunch!

Suggestion if possible to add a bunch of the elemental affinity pieces like Perfervid Sword, Donar Gun and Levante Dagger in? I believe that items like the Quanpur neck and archon ring and such are already in and functioning properly.

Also, super random appeal to add the White Tathlum into the tool - it's the best ammo slot I have found to use for BLM Laev AM3 TPing. Similarly, the Battlecast Gaiters are probably the best feet slot for mages using AM3 to TP.

Hey bud- if you really wanna go crazy on BLM AM3 TP sets, consider enjoying the Skirmish hell and grab some Telchine Pigaches. Naturally better MEVA than Battlecast, and augs make possible acc+20, STP+6....5 more acc and 1 more STP than Battlecast. ;)

Hello @Izanami I'm using the tool for WAR with Mighty Strikes ticked, and it doesn't seem to like crit gear for WS. I.E. Yetshila +1 or Boii Calligae +3.

Is that because of DPS loss or another factor?

I am looking through the window and it does seem to be trying to use Yetshila +1 but it has it losing to even items like Seething Bomblet +1 and stuff.

Edit: I'll add, set "damage" doesn't deviate from a set where mighty strikes is not ticked either.

This has been fixed now.

The Fire/Wind/Thunder affinity Sword/Dagger/Gun are already in the code. I've just added Battlecast Gaiters and White Tathlum, though.

This is because I forgot to tell the code to actually use Mighty Strikes for weapon skills, so non-crit weapon skills were still not getting crit rate with Mighty Strikes active. This should be fixed now.


Note that I do not have access to my Windows computer right now, so I've fixed the issue by manually adjusting the code on the GitHub website and testing the result on my MacBook (through the gui_wsdist.py file; I can't check the EXE on my Mac). This just means that the EXE file that I usually upload to the GitHub repository is not updated since I did not compile the updated code myself this time. The GitHub-generated EXE file on the Actions page should be updated. The GUI title should show "March 01a" if it is updated. You should technically be downloading the Actions page EXE anyway rather than trusting some EXE that you can't confirm the origin of.

I really like this. Awesome job, but I think in current state of game, if you want to have accurate avg numbers and DPS you need to add


near the end of average_ws method, because otherwise DPS is calculated of some impossible damage weapon skills. For example I had Tornado Kick doing up to 230k damage and way above 99k median, which boosted dps to 25k. With 99999 limit, it dropped down to 20k.

The problem is that the calculation of average damage becomes difficult when you also include a damage limit. For example: determining how much the average damage is increased by increasing DA from 0% to 5% is no longer a simple 0.05 multiplier to the base damage (if the average damage was close to 99999 already).

As you mentioned, adding a damage cap to the actual damage simulations is as easy as adding one line to cap damage at 99999, but the automatic set-finder itself uses average damage to find "best" sets.

For now, I've introduced the 99999 damage limit as an option in a new "Settings" drop-down menu at the top of the GUI. Selecting this will cap damage at 99999 as the last step before returning total damage dealt by the simulation weapon skills, but does not affect the "Quicklook" buttons or the automatic set finder. This is not a problem until you have enough buffs that you would frequently overshoot 99999, which is not a typical situation for the set finder anyway.
This new setting will be added to your "defaults.txt" file and remembered after you click "Save Defaults" in the File drop-down menu.



Finally, while I'm here:
Do we know how long the skillchain window is after each consecutive SC? I assume we do since I think there's an add-on that displays a countdown timer for multi-stepping. I currently have my test code set to allow 10s for the first SC, with -2s for each additional SC step down to a minimum of 3s. I do not think this is correct, though.

My testing environment is setup to handle skillchain damage for DPS simulations when spamming one weapon skill, which I've tested for Heishi Shun (2step, but requires one WS to first obtain Aeonic Aftermath), Kikoku Metsu (2step), Naegling Savage (no skillchain), and the 3-step from Dokoku Zessho. Before I introduce multi-step skillchains with different WSs at each step, I'd like to confirm the SC window timing that I'm using is correct.

window is 10 seconds initially, but you lose 2 when solo do to the animation. each step further lowers that by 1 second. so when solo you have 10(8) > 9(7) > 8(6). 3 is the min

Are you using the beta and have you updated the beta?

I see it using DEX and AGI: