AreaAudit

Which languages each generator says it supports, and where

Why writing systems break generated pictures first

Letters drawn inside a generated picture are the only language layer nothing can repair afterwards. Subtitles sit on top of the picture and dubbing does not touch it, so a wrong sign means generating the shot again. As of 2026-09-12.

Four attempted fixes for writing inside a generated pictureSubtitles sit on top of the picture and cannot change it. Dubbing does not touch it either. Changing the script before generation costs nothing, and shooting the writing as its own short insert means each market regenerates seconds rather than a scene.Try asubtitleIt lands ontop of thepicture andleaves thesignunderneathunchanged.Still wrongTry a dubIt replacesthe soundtrackand does nottouch anythingdrawn in theframe.Still wrongChange thescriptKeep legiblewriting out ofthe frame, orshoot it asits own shortinsert.Fixed, for freeOr regenerateProduce theshot again forevery marketthat needsdifferentwriting.
Fig. 1 The two fixes that work both happen before the picture exists, which is what makes this a script-stage decision rather than a post-production one.
Which language layers can reach writing inside a picture. Recorded 2026-09-12.
LayerCan change itWhat would be needed instead
A subtitleNoA new render of the shot
A dubbed lineNoA new render of the shot
A script change before generationYesNothing further
Shooting the text as its own insertYesA short regeneration per market

Inclusion rule. The ways a production can change writing that appears inside a generated picture. Two of the four require the picture to be produced again. Order. Cheapest attempted fix first, which is also the one that fails.

1It is a rendering problem rather than a language one

Drawing legible characters is about glyph shapes, spacing and joining, not about grammar or pronunciation. A model can speak a language convincingly and be unable to write it, and the reverse is possible too.

So a language list says nothing about it. A product that speaks forty languages may be able to draw legible text in two, and nothing published in this category would let a reader tell.

2Some scripts fail sooner than others

Marks above and below letters are small, sit outside the letter body, and disappear or land wrongly first. Scripts whose characters change shape depending on their neighbours fail differently and more visibly.

Latin text without diacritics is the easiest case and the one most demonstrations use. A claim naming only easy cases cannot be read as covering the hard ones or as excluding them.

3The cheapest arrangement is to avoid it

Where the story does not need legible writing, keeping it out of the frame removes the problem completely. That is a script decision, taken before any picture exists, and it costs nothing.

Where a sign has to be readable, shooting it as its own short insert means each market regenerates a few seconds rather than a scene. It also isolates the one shot most likely to need several attempts.

4It is the argument for deciding language early

Every other language layer can be changed after the picture exists. This one cannot, so it is the layer that decides whether a market version is a file edit or a re-production.

A production planning several markets should therefore treat visible text as a script-stage decision, in the same bracket as character names and settings rather than as a post-production detail.

An explainer about method and structure, not a count. No vendor number appears here that is not also in the register. The sourced material is on the register. Related: Voices for a series, Shapes of a set.