How this is put together

Four ways a film becomes 3D

These are not shades of the same thing. Two of them mean the depth was never photographed, and one of those is still, in the sense a reader means, real 3D.

Shot in stereo also called native 3D

Two cameras, or one camera with two lenses, running at the time of filming. The depth in the finished film is the depth that was in front of the lens.

441 films in this catalogue

Converted also called post-converted, 2D-to-3D

Shot flat, with depth added afterwards by a conversion house: every frame cut into layers and displaced. Good conversion is expensive and slow, which is why the house that did it is worth recording.

349 films in this catalogue

Hybrid also called part shot, part converted

Both, in one film. Usually stereo rigs for the controllable sequences and conversion for the rest. Where a source states the split, this catalogue records the split rather than just the label.

35 films in this catalogue

CGI re-render also called rendered in stereo

An animated film whose original 3D models were rendered a second time from a second virtual camera. Nothing was photographed and nothing was converted: the depth is real, and it was computed rather than filmed.

15 films in this catalogue

A fifth state is the commonest one of all: not established. 281 films of 1,097 have nothing recorded about how their 3D was made, and those pages say so. An empty field is a fact about the record, not a gap to be filled with a guess.

What is in this catalogue, and why

This is not a survey. Nobody has watched 1,097 films to see whether they are in 3D. The catalogue is the union of a handful of published lists, and a film is here because one of them says it was released in 3D.

From the published lists

1,037

From a storefront or a shared sheet

60

1,037 films are here because Wikipedia’s lists of 3D films carry them. The other 60 arrived from a community-maintained sheet of titles that play in 3D on the Vision Pro, or from Apple’s own storefront — films the lists had missed.

So “released in 3D” is the one claim here that is not independently checked, and it cannot be. It restates the reason the film is in the catalogue. Every other fact on a film page can be held against a second publisher; this one is the edge of the record rather than something inside it, and holding it against itself would prove nothing.

Two things follow, and both are worth saying plainly. A film missing from this catalogue is not a film that was never released in 3D — it is a film none of these lists carries, which for anything obscure, foreign or very old is a real and ordinary gap. And a film here that never actually had a 3D release would be somebody else’s error that this site has inherited and repeated.

The second is the harder one to catch, because nothing about a wrong entry looks wrong. It is also why a correction is worth more here than a confirmation: the lists are the floor this record is built on, and the only way to find a hole in a floor is for somebody to stand on it.

Two sources, or it is not a fact

A claim needs two independent publishers before this catalogue records it. The word doing the work there is independent: two machines reading the same Wikipedia row agree with each other about nothing except what the row says. That is one source read twice.

The rule exists because it was broken once. A single research pass published a Blu-ray 3D that has never been pressed, and nothing in the site could tell that page from a real one. A later pass produced 453 confident claims that no 3D release existed — every one of them citing a list of theatrical releases, which cannot establish anything about a disc. All 453 were withdrawn.

So a reading that has been done and not corroborated is published as a reading, with the page it came from named and linked, on under review: 250 films sit there now. It is somebody else’s research until a second publisher agrees with it.

A fact printed on a film page with one source beside it is recorded but uncorroborated. It is the weakest thing this site prints, and it is marked so rather than quietly mixed in with the rest.

Three kinds of claim, and what each can be checked against

The rule above is not the same rule for every field, and this site spent a long time writing one word over three different meanings. Saying so is cheaper than pretending otherwise.

Claims publishers can disagree about

How the 3D was made, who converted it, what it was shot on, whether a disc exists. These are the claims the two-source rule is for, because two people who have both looked can and do come back with different answers.

381 of 816 films have a provenance two publishers agree on. 23 more have two that contradict each other, and those pages print both. The rest carry one source, which is what that badge has always meant.

What the rule cannot see

It counts hostnames. Two addresses republishing one upstream list look like two publishers to it, and the only way to find out is to ask how often they differ. Two editors making a judgement call on “shot in stereo or converted” differ somewhere.

Two publishers	Films both hold	They agree on
en.wikipedia.org and realorfake3d.com	231	212 91.8%
3dmovielist.com and realorfake3d.com	195	194 99.5%
3dmovielist.com and en.wikipedia.org	189	189 100.0%
docs.google.com and en.wikipedia.org	127	127 100.0%
3dmovielist.com and docs.google.com	113	113 100.0%
docs.google.com and realorfake3d.com	102	101 99.0%

Three of those pairs have never once disagreed. That is suggestive and it is not proof, so no row has been deleted or downgraded for it — inferring a copy from a correlation is exactly the step this page rules out. Read the figures and weigh a confirmed badge accordingly: where it rests on a pair that has never dissented, it is weaker than the word suggests.

Claims read from the only place that would know

Whether Apple sells a film in 3D is answered by Apple’s storefront and nowhere else. A second publisher cannot corroborate it; it can only report what the storefront said on some other day. These carry the address that was read and the date it was read, and their worth is entirely in how recently that was.

689 claims of this kind.

Claims about which film this is

A TMDB or IMDb identifier is not a fact about 3D at all. It is an assertion that this record and that one describe the same film, and the thing that can be checked is the match itself — whether the year agrees, whether the title is a coincidence. Everything hanging off a match stays hidden until it holds, because a wrong match prints another film’s plot under our title.

1,137 identifiers held.

The conversion house is the field where this went wrong longest. Every credit in the catalogue came from one encyclopedia, all of them marked as though they had been checked against something, until trade press was finally read against them.

When sources disagree

The catalogue prints the disagreement. Both readings, both linked, neither promoted. A reference work that silently picks a winner has converted somebody else’s uncertainty into its own confidence, which is the one thing it is for not doing.

This happens more than it looks like it should. The honest answer to “was this shot in 3D?” is sometimes that two people who have both seen it do not agree.

Saying that something does not exist

“There is no 3D Blu-ray of this film” is a claim, and a harder one to support than its opposite: a single listing proves a disc exists, and nothing short of exhaustive looking proves one does not. So an absence is held to the same two-publisher rule as anything else, and it carries the date it was checked, because an absence goes stale the moment somebody presses a disc.

Where this catalogue simply has not looked, the page says nothing rather than saying no. Those are different answers and the difference matters.

What the dates mean

Every fact carries the date a person or a script last read its source, not the date the fact became true. A price checked in March is a price from March, and the page says March rather than implying today. Availability goes out of date faster than anything else here, which is why those rows print their date beside the value instead of tucking it into the source list at the foot.

Whether the addresses still work

A citation stops being one the day its page goes away, and nothing about the fact it supports changes to say so. So every outbound address here is fetched again and the answer written down: 2,653 of them on 18 October 2026, of which 10 no longer resolve.

Those 10 are not on the site. An address this repository records as gone cannot be linked from a page, and the build refuses to ship one that is. 30 more could not be read either way — a shop that turns away an unfamiliar browser, a server that did not answer in fifteen seconds — and every one of those is left where it is. A page we could not reach is not a page we know is missing.