Six realms, quests, and returning characters
Carries worldbuilding and unlocks new content.
Perfect World · Interaction Planning Internship · 2026
Zhu Xian 2 Equipment Reroll Redesign
During a three-month interaction planning internship with Perfect World’s Zhu Xian 2 team, I audited the equipment reroll system, identified 13 interaction and information issues, redesigned four core interfaces, and validated the flow through two rounds of user testing.
Decision support for high-risk rerolling
Make every
reroll decision
understandable.
The redesign keeps the underlying values and drop logic intact while making recommendations explainable, material goals traceable, and irreversible actions safer.
SYSTEM UX · INTERACTION PLANNING · USER TESTING
PROJECT BOUNDARY
Improve decision quality without changing the game economy
The proposal does not alter values, drop rates, or equipment rules. It restructures the interaction flow and information hierarchy so players can understand materials, risk, recommendations, and results before committing.
01 · System Audit
The growth loop worked, but its decisions lacked evidence
Zhu Xian 2 is a multi-platform MMORPG whose character-growth layer connects equipment, realms, artifacts, formation spirits, and the broader resource economy. Equipment rerolling sits directly inside the daily growth loop: acquire resources, invest, gain combat power, enter harder content, and improve acquisition efficiency.
I mapped the current system and documented 13 interaction and information issues. The recurring pattern was not missing functionality; it was missing evidence at the moments when players had to decide.
PRODUCT POSITION
A cultivation-themed MMORPG built from six connected systems
Set one thousand years after the original Zhu Xian story, the multi-platform MMORPG combines IP narrative, an open world, real-time combat, character growth, life skills, multiplayer relationships, and long-term events. Its three design goals are to extend the IP, build character growth through realms, equipment, artifacts, formation spirits, and related systems, and sustain stable multiplayer relationships under one cultivation-themed language.
Class combat, team dungeons, bosses, and exploration
Creates the repeated moment-to-moment play loop.
Level, equipment, skills, artifacts, spirits, and stones
Converts resources into combat power and build choices.
Production, crafting, trade, and upgrade spending
Moves resources between play, exchange, and growth.
Teams, societies, families, homes, and auctions
Builds persistent multiplayer relationships.
Daily, weekly, limited events, and rankings
Sets the cadence for returning and progressing.
Dependency chain · World narrative opens activities → activities produce resources → resources enter growth systems → growth supports competition → scheduled content regulates the rhythm.
GROWTH MODEL
Rerolling belongs to build growth and feeds the daily retention loop
Level and story progression
Unlocks systems instead of simply stacking values.
Equipment, enhancement, skills, and realms
Forms the baseline combat-power layer.
Artifacts, branches, traits, and stones
Creates configurable builds; equipment rerolling sits here.
Formation spirits, treasures, cosmetics, and collections
Adds long-term collection goals.
The game contains single-battle, daily-growth, periodic-cooperation, and long-term-content loops. Daily growth carries foundational retention: players acquire resources, invest them, raise combat power, unlock harder content, and return with better acquisition efficiency.
Acquire resources
Story · daily tasks · dungeons
Invest in growth
Equipment · artifacts
Raise combat power
Values and build strength increase
Enter harder content
Higher challenge becomes available
Unlock content
The available activity set expands
Improve efficiency
Better returns restart the loop
System tensions observed · free cultivation versus a unified combat-power score; open world versus quest containers; free class switching versus specialization; life skills versus combat attributes; and long-term goals versus system expansion.
CURRENT EQUIPMENT FLOW
The audit covered the whole equipment module before narrowing to rerolling
Entry and overview
Main screen → More → Equipment → overview and equipment schemes → equipment detail, bag, comparison, and settings.
Forging
Open forging → preview the effect and special effect → choose whether to use protection → forge → inspect the result.
Transfer
Choose the source equipment → choose the receiving equipment → confirm transfer → inspect the result.
Mystic stones
Open stones → select a stone → review upgrade requirements → upgrade → inspect the result.
Rerolling
Open rerolling → start → filter by equipment level → select material equipment → reroll → inspect the result.
Supporting screens
The module also contains material selection variants, reroll preview, equipment detail, inventory, equipment comparison, and settings screens.
13
Interaction and information issues across selection, confirmation, results, and repeated attempts.
4
Reroll main screen, material selection, consumption confirmation, and result preview.
2
A V1 diagnostic round followed by a fresh V2 validation round.
4 / 4
Every participant completed all four core validation tasks.
SYSTEM POSITION
Rerolling sits inside a dense growth network
Equipment, resources, combat power, and content progression form one loop, so each reroll decision affects more than a single screen.
MATERIAL JUDGMENT
Material value is hard to judge
Players see selectable equipment, but filters, protection, recommendation reasons, and consumption consequences are not sufficiently visible.
RESULT JUDGMENT
The result reports change without supporting a choice
Score feedback appears after rerolling, but the interface does not provide a direct old-and-new comparison or a clear next action.
PRIORITIZED ISSUE INVENTORY
Thirteen findings connected risk, comprehension, efficiency, and flow continuity
P0
Asset-loss risk
Material destruction had no confirmation
After players selected material equipment, there was no warning that it would be destroyed and no second confirmation. Selection immediately consumed the item and could not be undone; the frequency of accidental loss still required live-data validation.
Selected materials could not be changed
Different equipment and reroll materials provide different attempt counts, yet the chosen item could not be replaced or cancelled. A single mistaken selection could consume a valuable item without recovery.
P1
Core comprehension and task blockers
Trait ownership created the wrong mental model
Traits appeared on a single equipment page, suggesting that each item owned its own traits. In fact, rerolling refreshes the whole trait set, but the interface never explains that rule.
Trait quality and type were invisible
Traits showed only text and values. With no quality or type hierarchy, the total combat-power increase was the only reference and was not enough for a newcomer to judge whether a trait was worth keeping.
Two core reroll rules were hidden
The random attribute range depends on the level of the invested equipment, not directly on the material level; and rerolling changes the whole set rather than one equipment part or one isolated trait. Both facts were absent from the interface.
The attempt-count label was ambiguous
After filtering by level, the label “Level-30 equipment reroll attempts” did not explain who owned the attempts, whether each level had a separate limit, or how results carried across levels.
P2
Efficiency and guidance
Results lacked save and activation feedback
The preview summarized all attributes but did not connect the result to a specific item or confirm when it had been saved and applied. Result ownership and timing remained uncertain.
The overview had no equipment-acquisition route
Players could act only on equipped or stored items. Newcomers who needed replacement equipment had no visible route to obtain it, so the growth path stopped at the overview.
Transfer mixed equipped, best, and obsolete items
The transfer-source list displayed all forged equipment without distinguishing the equipped or strongest set, making obsolete source equipment harder to find.
Equipment state appeared only after selection
Equipped status and scheme ownership were shown only as small text after an item was selected. Players had to open items one by one instead of judging from the list.
Recommendation badges had no explanation
The interface could label an item or trait as recommended, but did not explain why it matched the current build or decision.
New and old results were not compared
A total value replaced a side-by-side comparison, forcing players to remember the previous result before deciding what to keep.
The repeat path broke when materials ran out
Players had to leave the panel to look for materials and reconstruct the reroll context before trying again.
Core diagnosis · Players could not judge whether a trait was good, which materials were safe to consume, whether the new result was better, or how to retry efficiently. The system therefore replaced player judgment instead of supporting it.
02 · Decision Gaps
Four moments made the same question hard to answer: what should I do next?
Which material is safe?
No filters, locks, recommendation reason, or clear warning made accidental equipment loss possible.
Is the trait valuable?
Trait quality and the logic behind recommendations were not legible enough to support a choice.
Which result should stay?
A total score replaced a clear old-versus-new comparison, and the word “Save” was ambiguous.
How do I try again?
Running out of materials forced players to leave the panel and rebuild context elsewhere.
03 · Competitive Analysis
Yanyun Sixteen Sounds showed how a reroll system can teach while it asks
I compared only the equipment-affix reroll flows in Zhu Xian 2 and Yanyun Sixteen Sounds. The useful difference was not visual polish. Yanyun exposed the trait pool, tier, recommendation logic, old-and-new comparison, confirmation, and material sources at the point of need.
Recommendations must be explainable
Show why an item or trait is recommended instead of asking the player to trust a badge.
Goals must remain traceable
Keep material sources and acquisition paths connected to the current task.
High-risk actions need safeguards
Use clear warnings, confirmation, comparison, or recovery wherever the cost is irreversible.
Borrow logic, not surface styling
The analysis informed decision support while preserving Zhu Xian 2’s own rules and interface language.
Decision supportCurrent Zhu Xian 2 flowTransferable insight from Yanyun
Trait targetTrait pool and quality hierarchy are not visible.Expose possible outcomes, tier, and the current value-to-limit relationship.
RecommendationA badge gives a conclusion without its reason.Explain why the item or trait is recommended at the decision point.
Result comparisonPlayers infer change from a total value and memory.Show old and new results together with visible differences.
Irreversible costMaterial equipment can be destroyed without a dedicated safeguard.Name the item, cost, and consequence before confirmation.
Material sourceThe player leaves the task to search elsewhere.Attach source channels and acquisition routes to the current goal.
GROWTH GUIDANCE
Expose the build target
Trait overview and growth guidance show what can improve before the player invests.
SOURCE TRACEABILITY
Keep material sources one action away
Dungeon, shop, and reward sources remain attached to the current upgrade objective.
RISK CONFIRMATION
State the consequence before confirmation
The confirmation identifies the irreversible cost and the exact action being approved.
04 · Flow Redesign
The new flow keeps players informed from material selection to the next attempt
SOLUTION GOAL
Give players system evidence at every reroll decision
Before rerolling, players should know what the equipment can produce, what counts as a good trait, and where the current result sits. Afterwards, they should know whether the result improved and feel confident keeping, discarding, or rerolling it. The interaction sequence remains based on Zhu Xian 2’s existing flow.
The original path delegated too much judgment to the system and broke when resources ran out. The proposed path adds recommendation explanations, filtering, a material-source shortcut, a consumption confirmation, a direct old-and-new comparison, and a clear route into another reroll.
DECISION AND CONFIRMATION MATRIX
Every decision receives evidence proportional to the cost of being wrong
Decision pointPlayer questionSystem evidenceCost of a wrong decision
Are the current traits good enough?Should I reroll this equipment?Quality tier, recommendation, and current value versus maximum.Spend materials on a result that did not need rerolling.
Are materials sufficient and disposable?Do I have enough, and can I afford to destroy this item?Level, category, attributes, recommendation, and acquisition sources.Destroy valuable equipment or leave the task when supplies run out.
Confirm material selectionAre these the correct materials?A review step before materials enter the conversion flow.A mistaken choice becomes irreversible.
Confirm equipment destructionDo I want to destroy these items for reroll attempts?A second confirmation states the conversion and irreversible consequence.Permanent asset loss.
Is the result better?Should I keep the new or old traits?Side-by-side comparison, difference markers, tiers, and recommendations.Keep a worse result or discard a better one.
Reroll again?Do I want another attempt?The old traits remain a reversible fallback and remaining attempts stay visible.None; the choice remains reversible.
PROPOSED PLAYER JOURNEY
Enter and judge
Open the reroll panel and review the overall trait panel before deciding whether the equipment can improve.
Change · quality color, recommendation badge, and value ratioFilter and select
Filter by level and quality, identify usable materials, and distinguish equipped items from safe choices.
Change · recommendations and equipped items shown disabledAcquire missing materials
When materials are insufficient, use Get More to see where they come from and how many can be earned.
Change · source shortcut and empty-state guidanceConfirm the cost
Review the selected materials, then confirm that equipment will be destroyed and converted into reroll attempts.
Change · two confirmation safeguardsCompare and decide
Compare old and new traits, then keep the old result, keep the new result, or reroll again.
Change · differences, tiers, recommendations, and remaining attempts05 · Core Interfaces
Four interfaces carry the complete decision journey
01 · REROLL HOME
See the current state before acting
The main screen keeps equipment, total traits, remaining attempts, selected materials, and the primary action within one stable workspace.
02 · MATERIAL SELECTION
Filter, inspect, and understand recommendations
Level filters, multi-select states, recommendation explanations, equipment details, and an acquisition shortcut make sacrifice decisions reviewable.
03 · CONSUMPTION CONFIRMATION
Make irreversible cost explicit
A dedicated confirmation names the equipment being consumed, the attempts it creates, and the fact that the action cannot be undone.
04 · RESULT PREVIEW
Compare first, then choose what to keep
Old and new traits appear side by side. The actions say “Keep old traits” and “Keep new traits,” while the repeat action stays visible.
06 · Interactive MVP
A clickable prototype made the complete V1-to-V2 change testable
I turned the proposal into a clickable MVP and used small-scale user testing to iterate. V1.0 was the pre-test version; V2.0 incorporated the findings from V1. The browser-based MVP lets reviewers switch versions, complete material selection and consumption, compare reroll results, and test the added acquisition loop.
Reduce decision cost throughout rerolling
Understand what the reroll result means
Find material sources without assistance
Explain what a recommendation means and why
V1 delivered a complete browser-based wireframe of the main reroll path, including material selection, consumption confirmation, and result preview. It proved that “destroy equipment → convert it into reroll attempts → spend attempts to reroll” could work, while leaving three gaps: Get More was only a placeholder, recommendation reasons were absent, and Cancel / Save did not state their consequences.
INTERACTIVE PROTOTYPE
Launch prototype Compare V1 and V2 in the browser
The prototype opens as a standalone subpage in a new tab, so the full task flow can be tested without leaving this case study.
07 · User Testing
V1 exposed language and risk failures; V2 closed the loop
Each round involved six participants: five players with MMO experience and one newcomer. In V1, the experienced group included two frequent reroll users and three general players. Participants completed an individual think-aloud task walkthrough followed by a short interview, with each session lasting about ten minutes. V2 was tested with six fresh participants who had not joined V1.
Validate three critical decisions
Can players explain why a recommended trait is worth keeping, understand that selected equipment will be consumed, and correctly choose between the old and new result?
Explain, select, and decide
Read a recommended trait and explain it; select a material and predict what happens next; then interpret the result actions and choose what to keep.
Behavior, not preference alone
I recorded task completion, time spent at each step, blocker locations, mistaken actions, and spontaneous interpretation of interface labels.
Individual think-aloud walkthrough
Each participant completed the flow alone, verbalized their thinking, and joined a short post-task interview. All blockers were recorded.
V1 TEST EVIDENCE
Interview questionObserved answerEvidence and failure point
What does the recommendation badge mean?5 of 6 shook their head or guessed that it “might be good.”They saw the badge but could not explain its reason or decide whether to keep the trait.
What would you do when materials run out?Participants wanted to replenish them, but Get More did nothing.The placeholder interrupted the flow before an acquisition route existed.
Would you inspect the detail card while selecting?Most participants skipped the detail card.4 of 6 did not understand that the selected equipment would be destroyed.
What would you do if the acquired reward was unsatisfactory?V1 had no acquisition page, so the question could not be completed.The product flow was incomplete rather than merely misunderstood.
What does Save keep?4 of 6 hesitated or interpreted it as saving every result.The label did not distinguish keeping old traits from keeping new traits.
Misread “Save”
Participants could not tell whether the button kept the old or new traits.
Missed equipment destruction
Participants did not understand that selected equipment would be consumed.
Recommendation was opaque
The badge existed, but its reason was unavailable at the decision point.
Acquisition was a dead end
Players could not find more materials and return to the same reroll task.
Add a material acquisition screen
Dungeon, shop, and task channels create a route out and back to rerolling.
Explain every recommendation badge
Hovering or focusing the badge reveals why the item is recommended.
Separate equipment consumption from reroll spending
Equipment first becomes reroll attempts; the attempts are then spent, matching the mental model shown on screen.
Replace ambiguous actions with explicit outcomes
“Keep old traits,” “Keep new traits,” “Start reroll,” and “Reroll again” state the result of each action.
V2 completed the acquisition loop: when materials are insufficient, players can open an independent source page, choose Secret-Realm Dungeon, Shop Exchange, or Training Tasks, preview rewards and remaining attempts, complete acquisition, and return with the new materials. The result page also supports Claim and Reroll the Challenge, while the reroll screen changes its primary action between Start Reroll and Reroll Again.
V2 · MATERIAL DECISION
Make material selection reviewable
Selection states, equipment details, recommendation explanations, and acquisition routes provide evidence before commitment.
V2 · RISK SAFEGUARD
Separate conversion from spending
The confirmation shows which equipment becomes reroll attempts before any attempt is spent, making irreversible loss explicit.
V2 · RESULT CHOICE
Name the result each button keeps
Side-by-side traits and explicit old-or-new labels turn an ambiguous save action into a deliberate choice.
08 · Validation Outcome
All six V2 participants completed all four core tasks
V2 validation taskCompletionObserved success
Understand the recommendation badge and its reason6 / 6All six hovered to see the reason, explained why it was recommended, and decided whether to keep it.
Identify available material acquisition routes6 / 6All six named Secret-Realm Dungeon, Shop Exchange, and Training Tasks and completed the return loop.
Inspect equipment details and understand consumption6 / 6All six checked the detail card and understood that the selected equipment would be consumed.
Choose a result and return to rerolling6 / 6All six could choose Reroll the Challenge to refresh rewards or Return to Rerolling.
The project completed an interaction-optimization case from problem definition to validation: 13 audit findings, a focused comparison with Yanyun Sixteen Sounds, three design principles, four redesigned interfaces, a clickable wireframe, and two rounds of testing. V2 repaired the acquisition loop, made recommendation logic visible, clarified consumption consequences, and replaced ambiguous result labels; all four core validation tasks were completed by all six fresh participants. The result does not prove that every live-game edge case is solved, but it shows that the revised prototype supplied the missing evidence at the decisions tested.
ZHU XIAN 2 · INTERACTION PLANNING
Back to all projects Explain the recommendation, trace the goal, and protect every high-risk action.