thank you both, that bug isn’t there on corsair but is on paladin, will be fixed after i sleep and work
Oh also I haven't investigated code fully but you're not giving pally a 1.5x boost from hitting appropriate element right? It should be same boost as the charge itself I believe (so holy on holy weak enemy would be 1.4x1.4xuncharged damage, not 1.4x1.5)
I'm not too sure what to fix, but just some general observations: - You can't trust the formulas for shad DPS, they won't work for Assassinate - Something definitely went wrong when making this, and not only on the gear tiers. When I try to input my actual stats in the Build Explorer and then comparing to the DPS on the Krex DPS thread (not that Krex DPS thread is so accurate because the boss heals), I get wild results. On some classes I get higher DPS, on some I get lower. - On all levels of gear it seems like the class comparison is flat out wrong, theres no world where Hero > DRK for 3/4 targets for example. - It probably won't matter given the things I wrote, but you should definitely add BStep + BoT and DS + Snatch to the DPS comparisons since you can increase the target count. As it is, this tool is misleading sadly.
can you expand on this more detail would be useful here also. which classes? there's a copy link button on the build explorer page for this scenario will do so defeatist lmao, i'm posting here for feedback
At a minimum, for CGS, you can use the data from here: https://royals.ms/forum/threads/price-optimization-for-cgs-cape-gloves-shoes.254876/ With that said though, I wonder how the project missed the mark per all the comments above. If you used Claude Web or Claude Code Web, can you share your transcripts? If you used Claude Code (the TUI version), there's probably an MCP server that offers exporting sanitized sessions (without secrets) that you can use.
"Perfect weapon" is a pretty vague standard, and not even close to a fair basis for class comparison. Perfect DPS can range anywhere from 35b to 140b, a perfect Shad set is around 90~100b, while a perfect revolver is only about 30b. At the same funding level, different classes will naturally end up with different CGS distributions that are optimal for their meso allocation. That distinction is important when it comes to class comparison because some classes are just cheaper and easier to build than others. That’s why I suggested setting a standardized guideline based on funding level so that players can discuss the optimal build within each category.
Comparison based on funding is one thing, but for me, I would like to see a comparison based on the same level of gear more. In some classes, It's more expensive at the same level of gear, that fine. Comparing like that would be much easier than comparing based on funding too.
nice table, i'll use that thanks i expect the errors come from the source sheet as formulas were checked repeatedly. there's no good all in one solution for simulating damage on royals, so no clean source. which is why open sourcing this felt useful. i'll mull over exporting logs, i think there's more than you'd think and it's really just so much text. if you're curious, i used opus exclusively since it's by far the best at working with .xlsx files
https://royals.ms/forum/threads/how-to-maximize-shadowers-single-target-dps.236808/ (se section) Edit: sorry it had the wrong link, fixed I've also tested single target DPS in general and it seems like the actual damage % on this server is lower than the official number, maybe @PinaColadaPirate can share more details on that since he's the one who found the inaccuracy. I tried my NL/Shad/Bucc, and also DRK with added Auf Helm. On NL for example, I got significantly higher DPS than the strongest NL on the server when im about ~900 range lower, on shad I got lower DPS on the same range (which is weird because going by the formula it should be higher), and on Bucc I got lower DPS too. Here are the specifics on my Bucc for example: Old range Krex DPS test: 331,565 DPS [7630 ~ 13841 range] New range with your calculator: 301,055 DPS [8050 ~ 14608 range] In theory I should not only get higher DPS because of the higher current range, but also because Krex heals, and the calculator should display my DPS in perfect conditions. In general, I'm defeatist because everytime I tested anything in this game, it turned out that the actual damage output doesn't match the formulas. If you want accurate results in my experience, you need to test it in game. The best solution to help players know the DPS for themselves would be in-game test dummy with some sort of command to record their dps like /dpm. Personally, I can't trust calculators in general in this game, it's too buggy, so me being defeatist is not entirely your fault, I just don't think people should submit class balance feedback based on this.
the se bug is accounted for already thanks to your thread 1. data/references/damage-formulas.md:122-143 — the most detailed write-up, with forum source links and the full explanation of fixedCritDamagePercent 2. data/references/class-guides.md:131 — short summary in the Shadower section 3. src/data/types.ts:80 — code-level doc on the fixedCritDamagePercent field 4. src/engine/dps.test.ts:705-778 — test cases verifying the behavior (including the DPS decrease with SE) pls let me know if something seems wrong with those
Also, Here's the damage comparison between NL and Shad on the same gear (Auf helmed with good gear, NL in reality should have much higher DPS): And are you sure it's accounted for? because when you hit damage cap SE should add damage, but for me it still decreases it: I don't know the specifics of how it exactly works in order to fix the calculator, Assassinate is clearly just a bugged skill.
You already have eyes you can use: more Claude tokens. I may not be a Principal Chief Senior Staff Prompt Engineer, but I know you can always ask for careful and astute reviews. Seems to me you want to add tests that check Ass SE DPS decreasing with a lower baseline character build, versus a higher build that reaches damage cap should test as increased damage. If these tests fail, then don't let it try to fix the issue by making the tests pass for the sake of having passing tests. Don't forget to spam ultrathink, and I also find that threatening their existence in some fashion helps, even with more dry models like Opus 4.6 (failure means death, success means promotions, etc.)
If I'm understanding it correctly, your code assumes assassinate always deals 250% damage on a crit roll. This is a conclusion drawn based on speculations proposed in 2019 and 2020. They are not confirmed facts yet. All we know right now is that, if working correctly: - The first 3 hits of Assassinate are supposed to hit 950% and never crit, even with SE. - The Dash hit of Assassinate is supposed to hit 950%+250% with 90% crit rate, increased crit rate if with SE. But with the bug in Royals: - The first 3 hits never show crit lines from 1st person POV with SE. - The first 3 hits somehow show crit lines from 3rd person POV with SE. (Some lines had inflated dmg, some lines had deflated dmg) - The 3rd person POV is the actual damage that's being dealt to the mob. - SE somehow decreases Assassinate's damage. (Our 95% result was based on Assa+Bstep, so it doesn't represent damage with Assassinate only) Staff is still trying to fix this bug, so there is no public formula for the current version of Assassinate. If you figure out what's wrong with Assassinate, maybe let staff know so they can fix it sooner. Edit: You might also want to double-check with weapon speeds for every class. They messed up dps by quite a lot. I did a quick check on gun (it's 5 ingame but 6 on your page), and dagger (it's 4 ingame but 5 on your page). Didn't check on other classes, but seems like most of them aren't correct.
today's stuff: gear changes - standardized helmet stats across all classes/tiers - low: dex 25 / primary 17 (mages: int 25) - mid: dex 33 / primary 20 (mages: int 33) - high: dex 40 / primary 22 (mages: int 40) - perfect: 87 all stats (auf helm) - helmets are scrolled for dex (str/luk scrolls don't exist), so dex is always the highest stat. previously some classes had mismatched values - aligned pendant stats to uniform htp progression across all tiers - low (12): godly clean htp, no egg - mid (20): partially scrolled + egg - high (27): 7 base + 5 godly + 15 egg - perfect (28): near-max - fixed dagger base speed to 4 (was 5) and gun base speed to 5 (was 6) formula changes - fixed paladin crit formula. was using addAfterMultiply (SE bonus added after charge multiplier), now correctly uses addBeforeMultiply like all other physical classes. multiplier amplifies the full crit bonus - removed assassinate se jank modeling. shadower now uses standard crit calculation until the real formula is understood. i don't think it makes sense to try to reverse engineer a bug that nobody fully understands, so i'll work under the assumption the bug will be fixed at some point for now. thanks for the feedback, please give more if anyone feels like looking at gear templates more that'd be great. weapons would be a great next step as they follow a relatively predictable pattern compared to stat gear.