Why do two sites list completely different filmographies for the same name?

Usually because one of them has the wrong person. Romanised names collide frequently, and stage-name changes split a single career across two records that neither site has linked.

How can I confirm two names belong to the same person?

Compare face vectors from images attributed to each name. Two names for the same person produce vectors that sit extremely close together — closer than any coincidental resemblance.

Is the Japanese spelling worth searching if I can't read Japanese?

Yes. Copying and pasting the Japanese name string into a search box requires no reading ability and often returns the official listings that romanised searches miss entirely.

How to Find a Performer's Titles by Name

Searching by name fails more often than searching by face, because one performer can carry a Japanese name, several romanised spellings, and one or more stage aliases. Establish the canonical name form first, then search each variant, then verify identity visually before trusting a filmography.

Searching by name is less reliable than it looks, because a performer rarely has one name. There is the Japanese form, one or more romanisations of it, and frequently a stage alias adopted at some point in the career — and catalogues disagree about which one is canonical.

The fix is to treat the name as a set of variants rather than a string, and to verify identity visually before trusting any filmography you find.

Why does one performer have several names?

Four distinct mechanisms, each producing a different kind of miss.

Mechanism	What it produces	Typical symptom
Japanese ↔ romanised	Same name, different script	Official listings invisible to a Latin-script search
Competing romanisations	`Yuu` / `Yu`, `Ryo` / `Ryou`, `Ohno` / `Ono`	Two partial filmographies, neither complete
Name order	Family-name-first vs given-name-first	Search returns nothing until the order is flipped
Stage alias change	Two separate career records	A career appears to stop, then start again elsewhere

The romanisation problem is measurable in our own data: 197 performer records have an English rendering that differs from the primary name form (our index, 2026-08 snapshot). That is 197 people who are findable under two spellings depending on where you started, and whose filmographies split accordingly.

Alias changes are rarer but more damaging, because they hide half a career rather than fragmenting a search. In our data 13% of performer records carry a recorded alias — and that figure is a floor, not a ceiling, since it only counts aliases we have been able to link.

What are the steps?

Establish the canonical form. Find one authoritative listing — an official catalogue page or a performer profile — and take the name exactly as written there, in both scripts if both are given.
Search the Japanese string. Copy and paste it; you do not need to read it. This usually surfaces official listings that romanised searches miss.
Search each romanisation. At minimum, try long and short vowel forms (yuu/yu, ou/o) and both name orders.
Look for a recorded alias. Profile pages often list previous or alternate names. Each alias is a separate search.
Verify visually before trusting the results. Compare the images attached to each record. If a filmography's images do not look like the same person, they are not.
Cross-check with a code. Once you have a candidate title, its code is the identifier that confirms it across catalogues.

Step 5 is where careless research goes wrong. A name search returns a record, not a person. Records get merged incorrectly, and two performers with the same romanisation get one shared page.

How do you confirm two names are the same person?

By comparing faces, not by comparing biographies.

Face vectors give a direct answer here. Two names belonging to the same person produce vectors that sit very close together — far closer than any coincidental resemblance between two different people. It is the one method that does not depend on whether a catalogue happened to record the link.

Evidence	Strength	Weakness
Two names' face vectors nearly identical	Strong	Requires images attributed to both names
Catalogue explicitly lists the alias	Strong	Only exists where someone recorded it
Matching biographical fields (debut year, measurements)	Moderate	Fields are sparsely populated and sometimes copied between sites
Similar appearance by eye	Weak	Look-alikes are common; this is how records get merged wrongly
Forum or social claim	Weakest	Unverifiable, frequently recycled

On field coverage, be aware how thin the biographical route actually is. Across our 2,333 named performers, debut date is present for 59%, cup size for 59%, birth date for 56%, height for 57%, measurements for 56%, and birthplace for only 31% (our index, 2026-08 snapshot). Matching on fields that are absent for a third to two-thirds of records is not a reliable identification method on its own.

What if you only have a face and no name at all?

Then the name is the output, not the input — and that is usually the faster direction.

A frame with a visible face returns the closest indexed identities with similarity scores. Above our calibrated threshold of 0.40 for video stills, a match is treated as probably the same person. From the name you get the filmography; from the filmography you narrow to the title.

Boundary worth stating: our index contains 241,792 faces but only 2,333 are linked to a name, and 2,246 of those have their representative image from a single source. A high-scoring result with no name attached means the face is indexed and the identification is not — a real and common outcome, not a malfunction.