Initial mavuika implementation by Charlie-Zheng · Pull Request #2301 · genshinsim/gcsim

Charlie-Zheng · 2025-01-01T17:56:24Z

Removed C2 extending initial cast. C2 extension corrected to 6s.

Changed from 10f to 2f after in-game testing.

Syntax fix from previous PR?

- Add C2 DMG% buff when party member takes damage within field - Delay adding field by 1f after final Burst hit to avoid self-triggering

- Reapply remaining sanctum DoT ICD after burst

- Remove normal attacks resetting the boolean for whether skill had been reused - Rename boolean for skill reuse to make more sense - New function to remove field since it is used in several scenarios - Skill recast function now uses character skill tracking variable for skill duration

- Fixed burst casting skill if starting with skill off cooldown. - Set burst status to apply on cast to handle actions that can interrupt the burst being used before first hitmark. Will need to evaluate best way to cancel the first queued attack. - Removed old burstCast tracking variable.

- remove / 100 as well

- implement a1 - implement a4 - implement skill damage mitigation and self DoT - fix burst frames because no basis on sheet - fix burst punch/kick state for iframe semantics - move asc/cons checks into the functions - move c2 to its own event sub - rename remove field to pick up field

- two vars instead of array for flat dmg ratio

- fix c4 icd not being hitlag extended - adjust c4 heal message - c4 heal target should always be dehya

- fix slow punch hitmark - fix burst interaction with attack/skill during punches - add dash -> jump interaction during burst leading to kick - refactor jump canceling burst punches - min 6 punches idle, max 10 punches <C6, max 15 punches C6 applies now - remove burstFrames for constant - cannot cancel before first punch so simplified burst func

- doesn't really matter right now because burst punch/kick do not have hitlag

- fix frames - fix sac pickup timing to happen right before redeploy instead of on action start - remove unnecessary hitlag info for non-hitlag attacks - adjust c2 flag reset to be a bit more future-proof - fix skill damage DoT having icd (should have no icd)

Atk Speed did not apply accurately, adjusted animation length to account for attack speed. Hitmarks were moved up slightly (more accurate) to assist with BoL logic in configs.

Updated frames for Skill and Attack actions.

Added frames for plunge for bike and non-bike version. Support for walk-jump into low_plunge.

Comment update on CAF Delay

Only one left, not sure how to solve.

Fix nested statements to cap allowed hold duration to various limits.

CAF was double-adding partial CA durations when forced into final hit.

Recast can queue faster than normal skill use, so this extends the normal skill frames out of plunge to match total frame times. The ratio is slightly skewed, but I think it's the lesser of two evils. Fixed CAF timing again.

Based on testing, A4 is affected by hitlag.

jwalk-511 · 2025-02-27T23:29:57Z

internal/characters/mavuika/cons.go

+		if a.Target.Type() != targets.TargettableEnemy {
+			return
+		}
+		if c.StatusIsActive(c6IcdKey) {


Does this status have a function, since the ring can only trigger once every 2 seconds? Or is it just an extra sanity in case the ring hit interval is somehow decreased?

Or does the cb from QueueAttack get called for every target hit, so this is required to prevent multiple triggers from a single ring hit in aoe?

jwalk-511 · 2025-03-06T03:37:43Z

internal/characters/mavuika/jump.go

+}
+
+func (c *char) Jump(p map[string]int) (action.Info, error) {
+	if !c.StatusIsActive(player.XianyunAirborneBuff) && c.armamentState == bike &&


Leaving note for reference for plunge code review. If Mavuika has Xianyun Airborne Buff, Mavuika will do a Xianyun-buffed jump, no matter whether Mav has bike armament state or not.

internal/characters/mavuika/plunge.go

Updated the valid target check to align with what it's actually checking. Changed bool for final hit to be more intuitive with the conditions triggering it.

Add support for biked plunges with XY buff.

jwalk-511

Review is not done, just leaving comments so far as a starting point (Notably, I have reviewed all functions called by BikeCharge from line 121 to 147, and functions called by those functions, recursively. But I have not reviewed functions only called by BikeCharge after line 147, including HoldBikeChargeAttack and CountBikeChargeAttack)

jwalk-511 · 2025-03-14T19:04:07Z

internal/characters/mavuika/charge.go

+
+func (c *char) buildValidTargetList() []HittableEntity {
+	enemies := c.Core.Combat.EnemiesWithinArea(combat.NewCircleHitOnTarget(c.Core.Combat.Player(), nil, 8), nil)
+	hittableEnemies := []HittableEntity{}


Very small performance optimization:
Initialize hittableEnemies with a capacity equal to the number of total enemies, so the array under the slice doesn't need to be reallocated every time you append to the slice (on line 485)

jwalk-511 · 2025-03-14T19:08:45Z

internal/characters/mavuika/charge.go

+	case *geometry.Circle:
+		enemyRadius = v.Radius() // Rt
+	default:
+		return false


It might be better to throw an error here rather than return false. If a non-circular target is ever added, this would cause Mav to silently "skip over" them, doing know damage, which would be hard for an end user to figure out.

Added error saying Mav needs circular hitboxes for CA calcs.

jwalk-511 · 2025-03-14T19:32:05Z

internal/characters/mavuika/charge.go

+		if g == nil {
+			continue
+		}
+		switch g.GadgetTyp() {


Can you try to future-proof by adding a default case that checks if a gadget is targetable and adds it to the gadgets list if so (similar to the Tea Lotus)? Or raises an error if you think any new gadget type would need custom handling

I might be interpreting this wrong, but it looks like all gadgets are targettable, so a default case would be handling pretty much every other gadget type and may not know if something is "new" unless the current case select has every gadget type explicitly handled. I think either way would require a bug report in the future, but I might be missing an easier solution here.

I think you're right; I was assuming that most gadgets would have some flag indicating they shouldn't be targeted by player abilities but if that isn't present then it isn't a super huge issue.

It could only possibly matter if some future character can create a gadget that can be hit by any other character, which is a rare case anyway

Generally if a player cannot hit the object, there is no reason to actually code the object as a gadget.

jwalk-511 · 2025-03-14T19:42:39Z

internal/characters/mavuika/charge.go

+		c.Core.Tasks.Add(func() {
+			c.bikeChargeAttackHook()
+		}, bikeChargeAttackStartupHitmark-1)


I think this needs to be called immediately. If a target moves/dendro core spawns between this frame and bikeCASH -1 frame, then no change would be made to the bikeHE list.

If it doesn't need to be called immediately, then you can pass c.bikeChargeAttackHook directly rather than an anonymous function that calls the same function.

Went with the first option, since cores could be missed in the scenario you described.

internal/characters/mavuika/charge.go

Updated docs for Mavuika's actions and params. Added bike dash handling for specific frame counts. Added an error when a target has a hitbox shape that does not work with the CA calculation. Changed hook timing to capture dendro cores that spawn during windup. Updated comments. Fixed CAF not meeting the minimum CA duration if used as the first action. Added 'buffered' param to allow for delaying CA after burst in order to proc N0 triggers.

lint bugfixes

Add note about approximated aoe

Added hitlag and aoe information.

Aftermathrar and others added 27 commits January 13, 2024 20:53

Update Dehya C2 Skill Mechanics

d0756cb

Removed C2 extending initial cast. C2 extension corrected to 6s.

Update Dehya E proc delay

a9456dc

Changed from 10f to 2f after in-game testing.

Update Dehya skill.go

3126658

Syntax fix from previous PR?

Merge branch 'main' of https://github.com/Aftermathrar/gcsim

9b90e5a

Update Dehya C2 Field Mechanics

189d294

- Add C2 DMG% buff when party member takes damage within field - Delay adding field by 1f after final Burst hit to avoid self-triggering

Update Dehya Field After Burst

584f80b

- Reapply remaining sanctum DoT ICD after burst

Merge remote-tracking branch 'upstream/main'

82b92c0

Merge branch 'master' into dehya

d0a1105

fix dehya n1 to n4 and reposition lv.4 multipliers

1f92ddb

- remove / 100 as well

fix dehya n1 to n4 hitlag and small frame adjustments

f2a9ffe

Merge branch 'master' into dehya

071ef26

Merge branch 'master' into dehya

d78f830

Merge branch 'master' into dehya

c33e2c6

Merge branch 'master' into dehya

10af7b8

fix dehya a1 missing trigger icd

eb11d47

refactor dehya c1

c06ca08

- two vars instead of array for flat dmg ratio

refactor dehya c2

6ac972d

refactor dehya c4

84484b4

- fix c4 icd not being hitlag extended - adjust c4 heal message - c4 heal target should always be dehya

refactor dehya c6

5b5b320

fix dehya q kick frames

f380652

fix dehya q not being hitlag extendable

0c37152

- doesn't really matter right now because burst punch/kick do not have hitlag

fix dehya e

8fa9550

- fix frames - fix sac pickup timing to happen right before redeploy instead of on action start - remove unnecessary hitlag info for non-hitlag attacks - adjust c2 flag reset to be a bit more future-proof - fix skill damage DoT having icd (should have no icd)

Initial mavuika implementation

9d5b606

Charlie-Zheng marked this pull request as draft January 1, 2025 17:57

Aftermathrar added 2 commits January 26, 2025 20:21

Fix Clorinde eNA AtkSpd Not Applying

60de875

Atk Speed did not apply accurately, adjusted animation length to account for attack speed. Hitmarks were moved up slightly (more accurate) to assist with BoL logic in configs.

Merge branch 'main' of https://github.com/Aftermathrar/gcsim

ff95553

Aftermathrar and others added 2 commits February 18, 2025 00:39

Update Frames

7df76a8

Updated frames for Skill and Attack actions.

Merge branch 'gcsim_main' into mav-dev

55a1147

Charlie-Zheng marked this pull request as ready for review February 18, 2025 16:19

Aftermathrar and others added 12 commits February 18, 2025 08:34

Update NA String Logic

abdfd4e

Merge branch 'mav-dev' of github.com:Aftermathrar/gcsim into mav-dev

ec37ac5

Add Plunge Attack

d6b5985

Added frames for plunge for bike and non-bike version. Support for walk-jump into low_plunge.

Update charge.go

0d0c575

Comment update on CAF Delay

fix lint issues

4d54ab4

Only one left, not sure how to solve.

Update charge.go

bdb8127

Fix nested statements to cap allowed hold duration to various limits.

Fixed CAF Timing

6a349ad

CAF was double-adding partial CA durations when forced into final hit.

Update Skill use from Plunge

90655a4

Recast can queue faster than normal skill use, so this extends the normal skill frames out of plunge to match total frame times. The ratio is slightly skewed, but I think it's the lesser of two evils. Fixed CAF timing again.

Merge branch 'mav-dev' of github.com:Aftermathrar/gcsim into mav-dev

b1c9f06

Add Hitlag Extension to A4

cda8c4b

Based on testing, A4 is affected by hitlag.

Merge branch 'mav-dev' of github.com:Aftermathrar/gcsim into mav-dev

1cd418c

Made a4 and nightsoul reduce affected by hitlag

e9c8021

jwalk-511 reviewed Feb 27, 2025

View reviewed changes

jwalk-511 reviewed Mar 6, 2025

View reviewed changes

internal/characters/mavuika/plunge.go Outdated Show resolved Hide resolved

Aftermathrar added 2 commits March 9, 2025 11:05

bool name changes

8a7730f

Updated the valid target check to align with what it's actually checking. Changed bool for final hit to be more intuitive with the conditions triggering it.

Update plunge.go

d2e78a0

Add support for biked plunges with XY buff.

jwalk-511 reviewed Mar 14, 2025

View reviewed changes

imring and others added 7 commits March 16, 2025 14:27

bugfixes pt. 1

16cc5b9

Fixing lint issues

3781ab0

lint bugfixes

Update character_data.json

9e80a63

Add note about approximated aoe

Update docs

b1bbf9d

Added hitlag and aoe information.

fix docs

2d660c0

pipeline

c447911

imring approved these changes Mar 17, 2025

View reviewed changes

imring merged commit 2744a41 into genshinsim:main Mar 17, 2025
3 checks passed

Conversation

Charlie-Zheng commented Jan 1, 2025 • edited by imring Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jwalk-511 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Charlie-Zheng Mar 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Charlie-Zheng commented Jan 1, 2025 •

edited by imring

Loading

Charlie-Zheng Mar 16, 2025 •

edited

Loading