BasicCAT — Computer-Aided Translation (CAT) Tools

Translate Your Comic Into Multiple Languages: A Guide for Independent Creators Going Global

Fri, 05 Jun 2026 12:52:50 +0000

The internet has completely changed how comics are created and distributed.

In the past, reaching readers in other countries often required a publisher, a localization team, and significant financial resources. Today, independent creators can publish directly to a global audience through online platforms.

However, two major challenges remain:

How do you make your comic accessible to readers who speak different languages?
How do you adapt your comic to different publishing platforms?

The first challenge requires localization. The second often requires format adaptation.

Fortunately, advances in AI translation and publishing tools have made both much easier than they were just a few years ago.

Why Publish Your Comic in Multiple Languages?

If your comic is only available in one language, most potential readers around the world simply cannot enjoy it.

Comics are one of the most globally accessible forms of storytelling. Readers from different cultures can appreciate the same artwork, characters, and stories. Yet language barriers still prevent many great comics from reaching their full audience.

Adding additional language versions can help you:

Reach readers in more countries
Increase social media exposure
Build a larger fan community
Generate more support and subscriptions
Explore international markets

For many creators, English is the first step. Later, expanding into Spanish, French, Portuguese, Japanese, or other languages can open even more opportunities.

Choosing the Right Platforms for International Readers

Before translating your comic, it is worth considering where your future readers are most likely to discover it.

Different platforms attract different audiences and support different business models.

Patreon: Build Sustainable Income

Patreon allows creators to monetize their work directly through memberships.

Many comic creators use Patreon to offer:

Early chapter access
High-resolution downloads
Behind-the-scenes content
Exclusive bonus materials

Since much of Patreon’s audience is English-speaking, having an English version of your comic can significantly improve its appeal to potential supporters.

For independent creators, Patreon can become a reliable source of recurring revenue.

Pixiv: Reach Japanese Readers

If your comic is influenced by manga or anime culture, Pixiv can be an excellent platform for gaining visibility in Japan.

Japanese readers are highly engaged with original comics and illustrations, and many creators use Pixiv as a primary publishing platform.

Even if you are not ready for a full Japanese translation, localized descriptions and tags can help increase discoverability.

WEBTOON: A Global Audience for Digital Comics

WEBTOON has become one of the world’s largest comic publishing platforms.

Unlike traditional page-based comics, WEBTOON is designed for vertical scrolling on mobile devices.

For creators seeking international growth, publishing an English version on WEBTOON is often one of the fastest ways to reach new readers.

The Hidden Cost of Multi-Language Publishing

Creating a translated version of a comic sounds straightforward.

In practice, maintaining multiple language editions can quickly become time-consuming.

For example, publishing a comic in Chinese, English, and Spanish may require:

Translating dialogue
Updating text layouts
Exporting multiple versions
Synchronizing future revisions

Every typo correction, script change, or chapter update must be repeated across all languages.

As a series grows, the workload grows with it.

How AI Is Changing Comic Localization

AI-powered tools have dramatically reduced the amount of manual work involved in comic translation.

Modern workflows can automate tasks such as:

Dialogue translation
Text replacement
Layout adjustment
Batch processing
Multi-language generation

As a result, localization is no longer limited to large publishers. Independent creators can now manage international releases with far fewer resources.

Introducing ImageTrans

ImageTrans is a localization tool designed specifically for comics, manga, webtoons, and image-based content.

Unlike many translation tools that only work with exported images, ImageTrans supports both image-based workflows and source-file workflows.

This makes it useful for:

Comic creators
Manga artists
Webtoon creators
Illustration storytellers
Localization teams
Small publishers

Whether you are translating a new project or localizing older works, ImageTrans can help streamline the process.

Workflow 1: Translate PSD Source Files Directly

If you still have your original project files, there is no reason to start from exported images.

ImageTrans can work directly with PSD files and other source files containing editable text layers.

A typical workflow looks like this:

Source File

↓

Extract Text Layers

↓

Translation

↓

Replace Text Automatically

↓

Export Target Language Version

This approach offers several advantages:

No OCR required
No text removal or image repair
Original layer structure is preserved
Easier future updates
Ideal for ongoing series

For creators producing long-running comics, this can save a significant amount of time.

Workflow 2: Translate Published Comic Images

Not every project still has its original source files.

You may only have:

JPG files
PNG files
Scanned pages
Archived web releases

In these situations, ImageTrans can use OCR to extract text and generate translated versions.

Image

↓

OCR Text Extraction

↓

Translation

↓

Text Removal

↓

Typesetting

↓

Export

This workflow is particularly useful for:

Older comics
Completed series
Re-releases
International editions

Workflow 3: Convert Page Comics into Webtoons

Language is not the only barrier when expanding to new platforms.

Many creators originally design comics for traditional page-based reading.

However, platforms such as WEBTOON favor vertical scrolling formats optimized for smartphones.

Adapting a page comic to a webtoon often requires reorganizing panels and restructuring layouts.

ImageTrans can assist with page-to-webtoon conversion workflows, making it easier to prepare existing comics for modern digital platforms.

This allows a single comic project to reach readers across multiple publishing ecosystems.

Which Languages Should You Prioritize?

If you are just starting your international expansion, consider these languages first.

English

The largest and most accessible global audience.

Spanish

A massive reader base across Spain and Latin America.

French

France has one of the world’s strongest comic-reading cultures.

Portuguese

Especially valuable for reaching Brazilian readers.

Japanese

A natural choice for creators targeting manga audiences.

From Local Creator to Global Audience

Not long ago, publishing a comic in multiple languages required translators, editors, and production teams.

Today, AI-powered localization tools have made global publishing accessible to independent creators.

Whether your goal is to build a Patreon community, reach Japanese readers through Pixiv, or grow an audience on WEBTOON, multi-language publishing can dramatically expand your opportunities.

Great comics do not lack readers.

More often, they simply need a way to be discovered beyond the boundaries of a single language.

OCR for Republican-Era Chinese Periodicals: Text Extraction and Translation into Simplified Vernacular Chinese

Wed, 03 Jun 2026 10:36:50 +0000

Most Republican-era Chinese periodicals were printed in vertically arranged Traditional Chinese and written in Classical Chinese. This article demonstrates how to use ImageTrans to recognize the text and translate it into Simplified Chinese vernacular language.

For example, the article “Women’s Virtue” from Volume 1, Issue 1 of Chinese Women’s World:

Configure MinerU

To achieve better OCR accuracy, we use MinerU as the OCR engine.

Configuration steps:

Download the plugin from mineruOCRPlugin.zip and extract it into the plugins directory of ImageTrans.
Apply for an API Key and enter it in the MinerU section of ImageTrans’s API preferences.

You can obtain an API Key from the official website:

https://mineru.net/apiManage/token

OCR and Translation

Create a new project and import PDF files or images. ImageTrans can directly extract images from scanned PDFs and import hundreds of pages within seconds.
Select MinerU as the OCR engine, then click Edit → Detect text areas and recognize text to process the current image. The recognized text regions will be highlighted on the original image, and both the positions and text contents can be edited manually.
To process all images in the project, choose Project → Batch → Detect text areas and recognize text for all pictures.
After OCR is complete, proceed with translation. You can use the DeepSeek large language model to translate the text into Simplified Chinese vernacular language. In the preferences dialog, modify the DeepSeek prompt and replace the {langcode} parameter with Simplified Chinese Vernacular.

Export as Markdown

Next, you can export the results as Markdown for reading, searching, or AI analysis.

The following is an example of the exported result, formatted with one line of original text followed by one line of translation:

婦德

妇德

汪長祿

汪长禄

問曰婦德何爲而作也昔者周公制禮以九嬤隸於天官實掌婦學之法其立敎之目有四曰婦德婦言婦容婦功德者何貞順之謂言者何辭令之謂容者何婉婉之謂功者何絲案之謂鄭玄既以此解周官其注禮記昏義亦循是說（二）然則當日后王君公卿士大夫之家女師所講閑門所習文質粲然斯已備矣載稽內則女子十年不出姆教婉婉聽從執麻東治絲繭紆組綱學女事以共衣服觀於祭祀納酒柴鑠豆茫蘸禮相助奠鄖云婉謂言語婉謂容貌與前說頗有異義孔類達釋之曰鄭意以此上下備其四德以婉爲婦言婉爲婦容聽從爲婦順執麻桌以下爲婦功從知德言容功四者析言之則有此別合言之可通名之曰德范史后紀九嬤掌敦四德（二）其說蓋有所受之矣班昭者生長名門博極羣書中更憂患以反說約所著女誠七篇務爲平易近人其言曰婦德不必才明絕異也婦言不必辯口利辭也婦容不必顏色美德也婦功不必技巧過人也幽閒貞靜守節整齊行己有風動靜有法是謂婦德擇辭而說不道憑語時然後言不厭於人是謂婦言盟浣塵穢服飾鮮潔沐浴以時身不垢辱是謂婦容專心紡績不好戲笑潔

问：为什么要写妇德呢？从前周公制定礼法，把九嫔归入天官管辖，实际掌管女子教育的法则。其中设立的教育条目有四项：妇德、妇言、妇容、妇功。什么是德？就是贞洁顺从的意思；什么是言？就是辞令应对的意思；什么是容？就是温婉柔和的意思；什么是功？就是丝麻纺织的意思。郑玄用这些来解释《周官》，他注解《礼记·昏义》时也沿用这个说法（二）。那么，当时王侯公卿士大夫家的女师所讲授、闺门所练习的内容，文采和实质都很完备了。考察《内则》记载：女子十岁以后不出门，由女师教导她们温婉、听从，学习处理麻葛、缫丝织茧、编织丝带、缝纫衣服，以供穿着，并观察祭祀礼节，送酒浆、摆放笾豆、调和肉酱，协助礼仪。郑玄解释说：婉是指言语，婉是指容貌，与之前的说法略有不同。孔颖达解释说：郑玄的意思是，上下文中已经包含了四种德行，用婉指代妇言，用婉指代妇容，用听从指代妇顺，用处理麻葛以下指代妇功。由此可知，德、言、容、功四者，分开说就有这些区别，合起来说都可以统称为德。《后汉书·后纪》说九嫔掌管教导四德（二），这种说法是有所传承的。班昭出身名门，博览群书，中年经历忧患后返璞归真，所著的《女诫》七篇，力求平易近人。她说：妇德不必才明绝世，妇言不必巧舌善辩，妇容不必容貌美丽，妇功不必技巧过人。幽静贞洁、守节整饬、行为有度、动静有法，这就是妇德；选择言辞说话，不传恶语，适时说话，不令人讨厌，这就是妇言；洗涤污秽、服饰整洁、按时沐浴、身体不脏不辱，这就是妇容；专心纺织、不好戏笑、洁净

期一第卷一第

期一第卷一

As you can see, once the Classical Chinese text has been translated into punctuated Simplified Chinese vernacular language, it becomes much easier to read and understand.

OCR for Ancient Chinese Books: Converting Scanned Ancient Texts into Searchable PDF and Markdown

Tue, 02 Jun 2026 12:55:50 +0000

Most classical Chinese ancient texts already have modern versions, whether digital or newly printed. However, a large number of ancient documents have yet to be digitized, such as local chronicles, genealogies, stele rubbings, rare library editions, and newly discovered or newly compiled versions of ancient books. To enable full-text search, version comparison, content proofreading, AI analysis, and long-term preservation, these documents still require digitization, a key step of which is OCR. This article introduces how to perform OCR on ancient books using the desktop OCR software ImageTrans, converting scanned ancient texts into searchable PDF and Markdown formats.

Challenges in Recognizing Text from Chinese Ancient Books

Recognizing text from ancient books presents difficulties such as vertical layout, complex page structures, image blemishes, tilted text, and recognition of variant and traditional Chinese characters. ImageTrans integrates OCR engines like PaddleOCR, mineru, and Qwen, which perform well with common traditional Chinese documents. For more challenging ancient texts, the Kandianguji Ancient Book API can be used.

Using ImageTrans

Below, we introduce how to use the Kandianguji Ancient Book API to recognize an image from Shishuo Xinyu (A New Account of the Tales of the World).

API Configuration

Install the Kandianguji Ancient Book OCR plugin. Extract the plugin into the software’s plugins directory. Download
Open ImageTrans, go to Preferences → API Settings, and fill in the API and account credentials (email or phone number) for the kandianguji entry.

The API can be obtained from the Kandianguji Ancient Books official website: https://kandianguji.com/shuzihua?page_mode=ocr_api

Text Recognition

Create a new project and import PDF or image files. ImageTrans can directly extract images from scanned PDFs, importing hundreds of images in seconds.
Select the OCR engine as kandianguji, then click Edit → Detect text areas and recognize text to recognize text on a single image. The results will be highlighted on the original image, and both the recognition positions and text can be edited.
To process all images, click Menu → Project → Batch Process → Detect text areas and recognize text for all pictures.

Export

After recognition, you can export the results as a searchable PDF or Markdown file. A searchable PDF retains the original scanned images while embedding a hidden text layer, making it easy to perform full-text search and copy text.

The Markdown format is more suitable for subsequent editing, version management, and AI processing, such as importing into knowledge base systems, generating training data, or performing automatic translation.

Note that due to the large number of rare characters in ancient books, a font with a large character set, such as Source Han Sans (Download), should be used to generate the text layer of the PDF.

Post-processing

Using large language models, you can further proofread the recognized text, add punctuation intelligently, translate, convert variant characters to standard forms, and perform traditional-simplified Chinese conversion. These features are yet to be integrated.

Conclusion

Digitizing ancient books is not just about converting paper documents into images; it is more important to transform them into searchable, editable, and analyzable data. With the Kandianguji Ancient Book OCR and ImageTrans’s visual proofreading capabilities, users can quickly convert scanned ancient texts into searchable PDF and Markdown files, providing a foundation for further research, organization, and AI analysis.

Free Alternatives to SDL Trados

Tue, 26 May 2026 12:55:50 +0000

SDL Trados is a computer-aided translation tool. It is the industry standard. Almost everyone in the localization industry has learned it.

It features translation memory, terminology management, support for various file formats, millions of options, and a community with industry expertise.

However, it is a bit complex and not cheap ($43/month for freelancer). There are various alternatives, like memoQ which has a lighter UI; Phrase (previous Memsource), which is a cloud platform. All have their uniqueness. But still, they do not meet my needs:

Cross-platform desktop app supporting Linux, macOS and Windows. Completely offline.
Easy to use interface.
Easy extensibility.

OmegaT stands out as a free and open-source alternative and is cross-platform. But I don’t like its UI.

In the end, I created my own CAT tool with B4J, a RAD development tool. It is named BasicCAT.

I have been developing BasicCAT since 2018. I’ve completed many translation tasks with it. Since it can directly translate Trados’s SDLXLIFF file, I can accept tasks requiring Trados even though I don’t have Trados.

In recent years, I mainly use it to translate the UI, website and documentation of my tools.

Its development has been stable. With the advent of large language models, I’ve added translation plugins to use ChatGPT, Gemini, DeepSeek, etc.

BasicCAT.org also provides other tools:

Aligner: Bi-text aligner.
ImageTrans is a computer-aided image translation and transcription (OCR) tool.
Silhouette is a computer-aided translation tool for videos and audios.

These tools are a good company for freelance translators, translation hobbyists, teachers and students as well as companies.

In the AI era, every one can write his own translation app. But computer-aided translation is a complex field. Since BasicCAT is open source, we can create our own tool with the help of AI and BasicCAT as the foundation.

Browser Extension for Manga Translation

Sat, 16 May 2026 12:22:50 +0000

There are many manga that never get translated. When we browse manga websites through a browser, we wish to see translated versions of manga images directly, but browsers don’t have this feature built in. That’s where browser extensions (add-ons) come in.

The computer‑aided image translation software ImageTrans provides such a browser extension, supporting Chrome, Edge, Firefox, and other browsers.

Manga translation typically involves the following steps:

OCR text recognition
Layout analysis – merging text based on semantics and adjusting text order
Translation (using large language models such as ChatGPT, Deepseek, etc.)
Text removal (inpainting)
Back‑filling the translated text (lettering)

Depending on the language and purpose, we can use different software combinations and translation strategies.

For example, the text removal step can be time‑consuming if processed with AI. If our goal is merely to understand the meaning, we can use a white background fill instead. The same applies to OCR and translation.

For OCR, if the language is a mainstream one like Chinese or English, local OCR such as PaddleOCR works quickly with good recognition accuracy. For vertical Japanese, traditional Chinese, etc., higher‑accuracy models are usually needed – for instance, mangaOCR for Japanese.

For translation, ordinary machine translation often performs well for languages with strong logical structures like English. For high‑context languages like Japanese, especially when OCR may contain errors, a high‑quality large language model is required – sometimes even feeding the image to the model for better semantic understanding.

The browser extension of ImageTrans supports freely combining various OCR and translation technologies to meet the translation needs of various manga images. The entire process can run on your local computer, free to use, no subscription required, no pay‑per‑page fees, with faster response times and better controllability. This is the method I recommend for translating images in web pages.

Below is a demo video:

The video demonstrates how to use ImageTrans’s browser extension to translate manga images from Pixiv, from Japanese to English. It shows purely browser‑based local processing methods – region‑based screenshot translation, full‑image translation – as well as using the ImageTrans desktop edition for higher‑quality translation.

Remove Text in Manga and Preserve Screentone

Mon, 11 May 2026 11:21:50 +0000

In the lettering step of manga translation, the original text needs to be erased and replaced with the translated text.

One of its tedious tasks is restoring screentones. Most tutorials use Photoshop’s clone stamp tool, which requires careful alignment with the screentone and struggles with complex patterns like gradients. Photoshop’s built-in content-aware fill is also unpredictable and may fill in irrelevant image content.

Another method involves using custom patterns and overlaying them on a separate layer, as explained in the Krita documentation, but the supported patterns are limited. More professional software like CLIP STUDIO PAINT and ComicWorks also require significant manual effort.

With advances in image processing and AI technology, several software tools now offer effective screentone restoration.

The computer-assisted manga translation software ImageTrans provides an all-in-one manga translation platform, covering processes from scanning, text recognition, translation to typesetting.

It integrates several methods for text masking and text removal. Below is a comparison of the results produced by some of its methods that handle screentones well.

Original image:

Text mask:

PatchMatch (traditional image processing):

Flux Klein 9B：

Lama Inpaint (original):

Lama Inpaint (fine-tuned for Japanese manga):

Gemini 3.0 (Nano Banana):

As you can see, the results are quite good, especially the fine-tuned Lama’s, and even when there are minor flaws, only a small amount of correction is needed.

These generative AI models have powerful repainting capabilities. Besides restoring screentones, they can also reconstruct various backgrounds covered by text, such as human figures, landscapes, and other content.

How to Translate Images on a Web Page

Fri, 08 May 2026 12:42:50 +0000

Browsers all have built-in webpage translation features, but they usually can only translate selectable text, not images. Yandex supports image translation, but with unsatisfactory results. Safari allows copying text but has no translation function. The computer-assisted image translation software ImageTrans provides a Chrome browser extension that supports direct translation of images on web pages, erasing the original text, filling in the translated text, and restoring colors, rotations, and various styles.

Below is its processing workflow:

Through the browser extension, an image translation request is sent to the server.
The server calls the connected ImageTrans program to perform the translation.
ImageTrans carries out text recognition with OCR, text translation, text inpainting and typesetting, generating a translated image.
After receiving the translation result, the browser extension replaces the original image with the translated version.

Because this operates within the browser, some technical limitations must be overcome:

Browsers restrict the use of local programs. The local server needs to add headers that allow local invocation. Older versions of Chrome also require HTTPS.
If an image is cross‑origin restricted, it cannot be downloaded directly, and fetching image data via canvas will result in a tainted canvas error. In this case, the extension must modify headers to remove CORS restrictions. If referrer information is also restricted, or if the image is not an <img> element but a canvas, making direct download impossible, a screenshot feature can be used—but content outside the viewport will be lost. Pixiv, for example, restricts cross‑origin access.
Some websites overlay additional content on images, so replacing the image alone does not display the translated version. For instance, X.com.

The ImageTrans extension can already translate images on common websites. Tested sites include Read Comics Online, Pixiv, MangaDex, Lezhin, Tencent Comics, and more. It supports various image types: manga, webtoon, manhua, comics, illustration, scanned books, etc.

Many browser extensions on the market rely on remote servers and require a subscription. ImageTrans, by contrast, runs entirely on the local computer, allowing unlimited translation, using various OCR engines and large language models, directly downloading images to the local machine, preserving translation results, and freely defining text styles.

Extension link: https://github.com/xulihang/ImageTrans_chrome_extension

Demo video:

PS:

ImageTrans’s extension also supports OCR right in the browser through WebAssembly. But its pocessing speed is slow and the ability is limited. It is provided as a free option along with the public ImageTrans server. It is recommended to use the extension with ImageTrans desktop version.

Real-Time Screen Translator

Sat, 11 Apr 2026 12:39:50 +0000

Recently, I want to play the raw Japanese GBA game Ace Attorney. I like the feeling of the original Japanese version and want to improve my Japanese skill. My Japanese level is limited so I still need a screen translator to help me.

ImageTrans is such a program. It integrates with various OCR and translation engines and can work as a real-time screen translator to play games like visual novel and read manga.

Here is its workflow:

The program detects whether the screen capture of the region specified by the user has changed and is now stable.
Use OCR to recognize the text in the region.
Use AI translation engines like ChatGPT and Deepseek to translate the text.
(Optional) Add pronuciation annotation (Furigana for Japanese or Pinyin for Chinese) and use text-to-speech engine to read the text.
(Optional) Save the recognition results for learning afterwards.

The whole process is automatic. You can also make the screen capture by yourself and ask the program to translate, since you may also need to read the text outside the region you specified.

The following is a demo video. Although it takes a while for the OCR and translation to complete and there are mistakes, it is playable.

ImageTrans can also translate other medias, like books and manga. It is an integrated translation tool, like Photoshop for translation.

ImageTrans can run on desktop platforms like Windows, macOS and Linux. Mobile operating systems like Android and iOS are currently not supported.

Optimizing Translation Speed and Quality

Real-time translation processes text only within the current input box, lacking contextual information, which leads to lower translation quality. Additionally, performing a translation operation each time results in slower speeds.

For many games, the text can be extracted, or transcriptions are already available. In such cases, we can pre-translate the text using a large language model. Later, during on-screen translation, the system can match the source text with OCR results and directly retrieve the pre-translated text, achieving both high speed and high quality. This feature is called translation memory in ImageTrans.

Below is a simple comparison using DeepSeek for translation. It shows that more context leads to higher translation quality.

Original Text	Full-Text Translation	Sentence-by-Sentence Translation
（うう‥‥。キンチョーするなあ‥‥）	(Ugh… I’m so nervous…)	(Ugh… Don’t be so nervous…)
‥‥なるほどくん！	…Naruhodo-kun!	…Mr. I See!
ふう。なんとか、間にあったわね。	Phew. I managed to make it in time.	Phew. We made it just in time.
どうかしら？　初めての法廷は。	How do you feel? Your first time in court.	How about it? Your first time in court?
こ、こんなにドキドキするの、小学校の学級裁判のとき以来です。	Th-This is the first time my heart has raced like this since the class trial in elementary school.	My heart’s pounding like this for the first time since the class trial back in elementary school.
‥‥それはそれは。ずいぶん、ごぶさたしてるのね。	…Well, well. It’s been a long time, hasn’t it?	…My, my. It’s been quite a while, hasn’t it?

Additionally, ImageTrans supports fuzzy matching between the source text and OCR results. Even if OCR makes errors, it can still find matches, preventing translation mistakes caused by faulty OCR.

For example, the sentence 借りがあるんですよ。ヤツには。 might be misrecognized as 借りがあるんですよ。ャツには。, where the character “ヤ” becomes the small kana “ャ”, potentially leading to translation errors. For instance, DeepSeek might mistakenly translate it as “I owe you a favor” instead of “I owe him a favor.” Matching against pre-translated text extracted from the original avoids such issues.

Manga Typesetting Tool Comparison: TypeR and ImageTrans

Tue, 07 Apr 2026 13:06:50 +0000

In the world of manga translation, typesetting is where everything comes together. After OCR and translation, you still need to place text back into the artwork in a way that feels natural, readable, and faithful to the original design. This is where tools like TypeR and ImageTrans come in.

Both target similar users—scanlators, translators, letterers, and comic localizers—but they represent two very different philosophies: one centered on Photoshop, the other on an integrated pipeline.

Overview

TypeR is a Photoshop extension (forked from TyperTools) designed specifically for manga and comic typesetting. It enhances Photoshop with features like smart alignment, style management, and multi-bubble handling, aiming to speed up manual work without leaving the design environment.

ImageTrans is a standalone computer-aided translation tool that covers OCR, translation, and text reinsertion. It can automatically detect text regions, remove original text, and inject translations, while still supporting export back to Photoshop for refinement.

Workflow Comparison

TypeR: Photoshop-Centric Workflow

TypeR assumes Photoshop is your main workspace:

Open image in Photoshop
Insert text using TypeR tools
Adjust layout manually

Its biggest strength is eliminating repetitive actions (copy-paste, alignment, style switching) inside Photoshop.

However:

OCR and translation must be done elsewhere
Workflow is fragmented across tools

ImageTrans: End-to-End Workflow

ImageTrans covers the full pipeline:

Load images
Detect and extract text (OCR)
Translate using multiple engines
Remove original text and reinsert translation
Export to PSD for final refinement

It integrates OCR, large language model translation, text cleaning, typesetting and batch processing into one system.

This reduces tool switching and enables large-scale workflows.

Feature Comparison Table

Feature	TypeR	ImageTrans
Core Type	Photoshop plugin	Standalone application
OCR (text extraction)	❌ Not supported	✅ Built-in multi-engine OCR
Translation integration	❌ External only	✅ Multiple MT engines + CAT features
Text detection	❌ Manual	✅ Automatic detection (AI + heuristics)
Text removal (cleaning)	❌ Manual in Photoshop	✅ Automated (masking, lama inpainting, nano banana, etc.)
Text alignment	✅ Yes	✅ Yes
Text reinsertion	✅ Manual typesetting	✅ Automatic + adjustable reinjection
Multi-bubble handling	✅ Yes	✅ Yes
Style management	✅ Advanced (folders, presets)	✅ Global styles
Batch processing	❌ Not designed for it	✅ Supported
Photoshop integration	✅ Native plugin	✅ Export to PSD
Automation level	Low (manual control)	High (semi/fully automated)
Cross-platform	Depends on Photoshop	✅ Windows / macOS / Linux

Which One to Use

Choose TypeR is you prefer to do the typesetting entirely in Photoshop.

ImageTrans is a better choice in almost all aspects:

An integrated environment for document scanning, image extracting from CBZ/PDF, image editing, OCR, translation, cleaning and typesetting.
Compatible with basiccaly all Photoshop versions (CS2+) and any other image editors like Krita and GIMP.

How to Convert Vertical Japanese PDFs to Horizontal Layout

Wed, 25 Mar 2026 06:25:50 +0000

Japanese is typically written in vertical, which may be difficult for people accustomed to horizontal text. We can convert it to a horizontal layout.

PDFs can be divided into two types: those with selectable text and those that cannot be copied. For the selectable type, since the PDF simply renders text at fixed positions, the text often cannot be copied in reading order and requires layout analysis. For non-selectable versions, such as scanned PDFs, OCR software is needed to recognize the text, followed by layout analysis to determine the correct text order.

The computer-aided image translation software ImageTrans supports processing both types of PDFs, exporting vertically oriented text into a horizontal format.

Example:

Original PDF File:

Converted horizontal text:

もちろんです以下にランダムな日本語の文章を生成しました。

一匹の子犬。王都が見えてきた。でかいのは態度だけで後は何もかもが小さい。

It also supports adding furigana annotations:

もちろんです以下いかにランダムな日本語にほんごの文章ぶんしょうを生成せいせいしました。

一匹いっぴきの子犬こいぬ。王都おうとが見みえてきた。

でかいのは態度たいどだけで後あとは何なにもかもが小ちいさい。