Deterministic render engine · v2

Any graphic, every language.

imato re-renders your finished designs in 39+ languages — same fonts, same layout, same pixels. Only the words change.

Translated poster in DeutschOriginal poster in EnglishEnglishDeutsch
Poster · EnglishDeutsch · real render
39+ languages
RTL and CJK included, rendered natively
Text is typeset, not painted
real fonts, matched per element
Backgrounds untouched
pixels outside text never change

How it works

Your design, taken apart — and put back together.

The same trattoria menu, at each stage of the pipeline. These are the engine's actual artifacts, not illustrations.

Menu source with detected text elements outlined — headings, items, body copy, and verbatim prices

1.Decompose

Every text element is detected and classified — headings, items, body copy. Brand names and prices are marked keep-verbatim and never touched.

The same menu with all translatable text erased — brand name and prices preserved on a clean background plateAntipastiPředkrmyGrilled bread, ripe tomatoes…Opečený chléb, zralá rajčata…SecondiHlavní chody

2.Erase & translate

Original text is lifted off the background, leaving a clean plate. Meanwhile your strings are translated — with your glossary applied — ready for review.

The finished menu re-rendered in Czech with matched typographyRendered · Čeština

3.Re-render

Approved translations are typeset back onto the plate in matched fonts, auto-fitted to each element's box. Download, or render the next language for one credit.

Built for real work

Menus. Posters. Screenshots.

Domain profiles tune the engine per graphic type — what wraps, what stays verbatim, how lists breathe. Browse the same pipeline across three of them.

Menu in English · source
English · source
Menu in Čeština
Češtinarendered
Menu in Français
Françaisrendered

Item names stay Italian (verbatim glossary), descriptions translate, prices never move a pixel.

Typography fidelity

Look closer.

Zoomed to the pixel: the same crop from the source and the render. This is where overlay tools and generative editors fall apart.

Diacritics, correctly typeset — source close-up in English
English
Diacritics, correctly typeset — rendered close-up in Čeština
Čeština

Diacritics, correctly typeset

Czech carons and acutes are rendered by a real font engine — not painted by a diffusion model. Same weight, same size, same baseline as the source.

CJK inside UI chrome — source close-up in English
English
CJK inside UI chrome — rendered close-up in 日本語
日本語

CJK inside UI chrome

The button pill is untouched — only the label is re-set, in a matched CJK face, auto-fitted so 予約 never overflows the capsule.

Right-to-left, done right — source close-up in English
English
Right-to-left, done right — rendered close-up in العربية
العربية

Right-to-left, done right

Arabic runs right-to-left with correct shaping and joins. The photo, the phone, the lighting — every non-text pixel identical to the English source.

An honest comparison

Why not just ask a generative AI editor?

You can — and for one-off edits they're great. For publishing localized assets at scale, the failure modes matter.

Generative image editors

GPT-Image, Nano-Banana-style “edit the text” prompts

  • Re-synthesize the whole image — photos, faces, and logos drift on every run
  • Text is painted, not typeset: kerning wobbles, diacritics get invented
  • Non-deterministic — the same prompt gives a different image tomorrow
  • No review step: you proof-read finished pixels, per language

imato's deterministic engine

decompose → erase → typeset, same result every run

  • Only text regions are touched — every other pixel is carried over unchanged
  • Real font engine: matched faces, true diacritics, RTL shaping, auto-fit
  • Reproducible — re-render a language a month later and it matches
  • Element-level review before rendering; edits persist in translation memory

The one honest caveat: photos with heavy baked-in text need AI repair for the erased background — imato detects those regions, fixes them automatically, and flags anything it's unsure about in review.

Glossary that sticks

Pin brand terms to stay verbatim, or map them per language once — applied on every future job and shared across your team.

Batch a whole campaign

Drop up to 10 files, pick your languages, and watch the render matrix fill in. One click per market, not one afternoon.

Review before rendering

Element boxes over your source, suggested translations per language, inline edits. Nothing renders until you approve it.

Pricing

One credit, one render.

1 credit = 1 image rendered in 1 language. Start free with 20 a month; the Launch Pack (300 credits, $19) never expires.

Free

Try the engine on your own designs.

Free
  • 20 credits / month, forever
  • All 39+ languages, review mode included
  • Watermark on outputs
  • No credit card required
Get started

Starter

Creators shipping a few assets a week.

$120/yr
  • 250 credits / month
  • No watermark, full resolution
  • Brand glossary
  • ≈ $0.05 per rendered image
Start Starter
Recommended

Pro

Teams localizing campaigns every month.

$390/yr
  • 1,200 credits / month
  • Glossary + translation memory
  • Batch translation — 10 files at once
  • Projects & priority processing
  • ≈ $0.03 per rendered image
Start Pro

Studio

Agencies and heavy localization volume.

$990/yr
  • 4,000 credits / month
  • Everything in Pro
  • 5 team seats, shared glossary + memory
  • API access
  • ≈ $0.025 per rendered image
Start Studio

Full detail — packs, rollover rules, team seats — on the pricing page. Bigger volume or API? hello@imato.ai

FAQ

Straight answers.

Ship it in every market.

20 free renders every month, no card required. Your first localized asset is about a minute away.