Dynasty Warriors Origins

I went to lengthen the build list and came back having deleted a line. I took the series reboot apart feature by feature to see what to port into my wuxia action game. Of twelve skeletal systems, nine were already ported and running, and the remaining gap was three battlefield-simulation systems. But partway through, something I had believed turned out to be wrong, and a prerequisite I was sure I had to build disappeared entirely.

What the game is

A series that threw away its own signature. It gave up the old “dozens of playable officers, 92 weapons” and reinvested everything into one original protagonist’s action depth plus battlefield simulation.

Combat has eleven features. Branching normal/heavy combos, per-weapon signature moves, a hex-point resource gauge, a tell colour system (white can be guarded, parried or dodged; orange only dodged or interrupted with a special), guard and parry, dodge and perfect dodge, and the armour gauge — a shield bar under an officer’s health that heavy attacks, arts and parries chip away, triggering an execution when it burns out.

Battlefield simulation has eighteen, and that is the differentiator.

Feature	How it works
Five morale tiers	rises on stronghold captures, officer kills and duel wins; falls when allied officers lose or strongholds are taken. High morale has allied AI capturing on its own; low morale makes even enemy rank-and-file stronger
Stronghold capture	a durability gauge plus defeating every garrisoned officer. Capture and loss feed straight into morale
Duels	it cuts to an arena ringed by soldiers. ⭐Decided by one shared see-saw gauge, not health — push it to full and you win regardless of remaining health
Mass formations	form when a legion packs together; drain their gauge and they break, flee, and morale collapses
Enemy grand tactics	when the enemy triggers a charge or a spear wall, you must meet a counter condition within a time limit. Fail and your side takes heavy losses

Morale being the deciding statistic for autonomous AI combat is the core of this design. On fronts the player cannot reach, one number decides whether your side wins or loses.

What broke — the thing I had wrong

What I had wrong was the reward for a perfect dodge. I believed “dodge at the last instant and you get slow motion”, so I concluded a time-scale arbiter had to be built first. It is actually a large resource gain plus a counter window, and slow motion was confirmed by no source at all.

So the prerequisite disappeared. A line was struck from the build list, and things queued behind it moved forward.

⚠ A caveat: confidence isn’t uniform. Most items were confirmed across multiple sources, but hit feedback is marked “could not explicitly confirm hitstop and camera shake”, and the layering of unit fighting spirit rests on a single source and stays unconfirmed. Mass formation scale is contested too — the developers say “ten thousand on screen” while reviewers describe hundreds to a thousand. That number must not be cited.

Held against my own setup

Half of this investigation is that nine were already running. Branching combos, parry, perfect dodge, resource gauge, armour gauge, arts, follow-ups, tell colours, executions — all present. Without the benchmark I wouldn’t have rebuilt something I had; I wouldn’t have known what I lacked.

That all three remaining items are battlefield simulation is itself information. It means I have kept pace on the action side and never touched the battlefield. And the fact that this game sold breadth to buy depth is exactly the right trade for a one-person project, so the direction is worth following.

The duel design is the one I want most. Pushing and pulling one shared gauge instead of health is simple to implement and carries a lot of tension. Because it resolves independently of remaining health, “I have more health so I win” stops working.

Verdict

What	Verdict
Time-scale arbiter	unnecessary. The premise dissolved during research
Duels — a shared see-saw gauge	adopt. High tension for the implementation cost
Morale as the deciding statistic for AI combat	adopt. One number runs the fronts the player can’t reach
Stronghold capture	adopt. Paired with morale, so it can’t be separated
Mass formations · 16 tactics · siege engines	hold. The scale is beyond a one-person project
The “ten thousand on screen” figure	⛔ don’t cite. The developers and the reviews differ by an order of magnitude

The lesson I paid for here: count benchmarking value only as “N things to take” and you miss half of it. The biggest result of this investigation wasn’t four things to take but one thing I no longer had to build. That kind of result never lands on a list, so it disappears unless you count it deliberately.

And ⚠ I’m leaving in the fact that something I believed turned out to be wrong. Without cross-checking sources I would have built a prerequisite for a feature that doesn’t exist.