Ziqi Zhou

Never marry your ideas, flirt with them.

Month: May 2018

Creating a Localized Poster for Fortnite Using PhotoShop

May 17, 2018 / / 0 Comments

Fortnite is a game developed by Epic Games and was officially introduced to mainland China on April 23th, 2018 by Tencent. Thus, the game needs a lot of localization work to make sure it is well adapted to Chinese culture with well translated UI strings, understandable names of heroes and items, and locally relevant videos or graphics.

Interested in this game, I decided to create a localized poster for it using Photoshop. The original poster shown below is quite simple: there are several heroes in the poster, the only text needs to be translated is the name of the game, and it has a clean background. Most of the times, when we talk about localizing something by doing desktop publishing, we are aimed at translating all the text in it without losing its original style or format. However, in this case, simply translating the text is far from what can be called “a localized poster”. So, I decided to recreate one based on the existing poster and make it more locally relevant, friendly and attractive to Chinese users.

(Original Poster)

Before I started everything, I wanted to be clear of my goals, which include adding in Chinese style roles, background and other elements.

Workflow

Step 1: Resource material collection

  1. Separate all the heroes so that I can move them around.

(png Files for Each Hero)

2. Add in a hero with a Chinese style outfit (Wukong outfit). Everyone in China knows Wukong, the Monkey King in the story “Journey to the West”.

(Wukong Outfit)

3. Add in a Chinese dragon, which is actually a glider in the game used by players to land onto the battle field.

(Royale Dragon Glider)

4. Create a background with Chinese characters “你好”, which literally means “hello” in Chinese. These two characters are made up by the forts in the game.

(Background Picture)

Step 2: Process pictures in PhotoShop

  1. Resize each picture and group them together in a V line, with the Wukong figure in the center so that it can be emphasized.

(The New Squad)

  1. Remove the background color of the Chinese dragon glider using the Magic Eraser tool and flip the canvas horizontally.

(Dragon Glider with Transparent Background)

3.Capture a screenshot from an advertising video of Fortnite posted on its Weibo account, clean the screenshot and resize the picture

(Screenshot from the Video)

4. Translate the text “Fortnite”.

Although there was only one word that needed translation. It did took me a long time to figure out how I can properly output the Chinese text into the right font. The problem is, I couldn’t find the same font with the game in PhotoShop. Then I decided to get a picture with the official Chinese name of the game and use it to replace the original logo.

(Bilingual Logo)

5. Put every layer together onto one canvas and make adjustments.

(Final Product)

Challenges and Solutions

  1. Resource Material Collection

Resource material collection is the main challenge I’ve encountered in the whole process. I need to get the png file for each character in the poster so that I can reposition them. Although I can find these resources from Google Images by searching “Fortnite characters transparent”, the resolution and the size of these pictures are different. Once I gathered them onto one canvas, they didn’t appear like a whole due to resolution difference.

Solution: Resize each picture carefully so that they look the same big when being put together.  Use the sharpen tool to sharpen the blurry layers to make each hero in the squad stays in the same level of definition.

Another challenge was the lack of ideal resource materials. Most of the time, I need to recreate what I’ve got and then use them for my final product.

For example, it took me a long time to find a satisfactory background picture. As I was exploring every post on Fortnite’s official Weibo page, I was impressed by its advertising video for Chinese market and decided to capture a frame from it. It was not hard to do a screenshot, but some work needs to be done for cleaning up the screenshot.

To remove the game logo and irrelevant text, I used the Content-Aware Fill tool under the Edit tab of PhotoShop.

  1. Font, font, font ……

As I mentioned above, I had trouble finding the same font used by Tencent or Epic. The walkaround solution I took was using another picture with the official Chinese translation text and making it into a layer in my final product. But since it is not a real text layer, this would be problematic if there’s a huge amount of text or the file needs to be recreated in InDesign.

In real life, however, if you are a member of the localization team of Fortnite, you probably can access to their resource library and get the font.

Conclusion

The recreation of the poster is not a difficult task and doesn’t need a lot of high-level skills in PhotoShop. But compared to the original one, the final poster looks more relevant to the Chinese market. As a Chinese player, I will appreciate Fotnite’s devotion in adapting its products for a specific market. As a localizer, I want to prove, by doing this project, that localization is much more than translation, it needs creativity, familiarity with the product, and attention to details from the localizer.

Customizing your own machine translation engine: an MT training project using Microsoft Translator Hub

May 16, 2018 / / 0 Comments

Introduction: This blog post is about a SMT training project using Microsoft Translator Hub. In this portfolio, you will read about the description of the project, lessons learned from the project, my ideas on how we can improve translation quality by leveraging the power of MT, and my experiment of other techniques to improve the overall localization workflow.

SMT Training Project: Developing a MT engine for translating Chinese Government Work Report

Project Overview: This pilot project is designed to establish a Statistical Machine Translation (SMT) engine from English to Chinese in order to support the Chinese government initiative to provide public documentation in both Chinese and English. Our data source will be extracted from Chinese government public report websites.

In order to be considered “fully trained”, the post-edited machine translations (PEMT) from this engine must meet the following target criteria for efficiency, cost savings and quality:

  • Efficiency: PEMT 20% faster than human translation
  • Cost: PEMT 25% savings over human translation
  • Quality: PEMT with an acceptable score of less than 30 based on the Multidimensional Quality Metrics (MQM). The acceptable score was increased from 10 points to 30 points due the length and complexity of the product review compare to the government reports.

A total of 13 rounds of successful training were completed with an initial data set of approximately 56,916 segments used as training data, approximately 1,270 segments used as tuning data, and approximately 1,559 segments used as testing data. A BLEU score will be given after each round of training by the Microsoft Translation Hub. The initial round of training achieved a BLEU score of 12.6 and over the next three weeks, 12 more rounds of training were done with a result of 18.61 (~48%) as the best BLEU score achieved. Finally, the MT system with the highest BLEU score was deployed and the translation of this system was compared with the human translation team to reach our final conclusion of this project.

Data: If you feed the machine with garbage, your machine will output garbage. To maximize the quality of the input data, the official bilingual files from Chinese government were used and each file was carefully segmented and aligned using CAT tools.

  1. Timeline

*GWR: Government Work Report; WP: White Paper

Date Task Duration
04/02 – 04/03 Project Planning 3 hours
04/04 Data Collection: 3 TMX (GWR 16-18) + 2 TMX (WP) 9 hours
04/05 Data Collection: 3 TMX (GWR 13-15) + 2 TMX (OPUS and WP) 6 hours
04/06 Realign Tuning Data: GWR 13/15/17 3 hours
04/08 Data Collection: 2 TMX (GWR 11/12) + 3 TMX (WP) 5.5 hours
04/10 Data Collection: 3 TMX 0.5 hour
04/11 Data Collection: 4 TMX 1 hours
04/12 Data Collection: 8 DOC (GWR Monolingual) 1 hour
04/13 Refine Training 1 hour
04/15 Deploy System 0.5 hour

 

2.Conclusion

A 500-word excerpt from Government Work Report was used to test the deployed engine. The MT output was post-edited by editors and errors from the output was identified. On the other side, the same file was human-translated. After comparing the time, quality and cost, we reached the following conclusion.

 

Based on 500 words Rate Saving Post Edit Time Time Save
Goal:

Price: 20%

Time: 25%

 HTEP $0.25  

20%

 60 min  

25%
PEMT+EP $0.20 30 min
 

Final Result

 HTEP $0.25  

    0%

 60 min  

0%
PEMT+EP $0.25 60 min+

(Comparison of errors found in HT and SMT using MQM as the criteria)

The outcome of our SMT did not meet any of our goals for this project. Nevertheless, the team has put tremendous effort into the allotted time and completed the entire process from initial planning to deploying the trained SMT.

Lessons Learned

The experience and knowledge gained from this pilot project resulted in the analysis to accurately project the requirements to fully train this SMT to support the government work report initiative.

Training Data Requirement:

Based on our pilot project result, we project at least 1,000,000 segments are required to fully train the SMT engine. The major quality issue resulted from our SMT engine are grammar, more specifically sentence structure problem. We conclude that increasing the quantity of training data and tuning data will significantly improve the grammar quality of the SMT engine.

Additionally, well-aligned source documents are key to the success of the SMT training. We only accepted the official government translations and conducted multiple alignment and reviews of our data. The result was less quantity but better quality, the result was evident in our 9th round training result. Our BLEU score increased from 13.4 to 18.6. We believe that adding a dictionary and monolingual data would have significantly improved our BLEU score and the overall quality of our SMT engine.

Time and Cost:

We project 783 hours and about $35,000 are required to fully train the SMT engine. A team of at least 4 full-time members would ensure the SMT can be fully trained to meet the goals. Our team significantly increased the rate of SMT training rounds after the fourth round, which has helped us quickly identify a deficiency and the direction we should focus on.

Tools:

Initially, our team only used the TMXMall as our primary tool for favoring its simplicity and accessibility. But, quickly we found the quality of our aligned documents have suffered from our tool or choice (see first four rounds). By implementing TRADOS Studio and Okapi Olifant, we were able to align all of our documents into sentence segments and drastically improved the quality.

In conclusion, we do not recommend using the SMT instead of the human translation for the government work report initiative. We believe, with the proposed plan outlined above, the fully trained SMT engine would meet the established goals in order to be utilized for future government work report initiative.

Meet the future: Customizing a Neural Machine Translation engine?

When identifying error patterns of MT outputs, we captured some causes of these errors. Word order is a typical error of the MT output. The MT system tends to translate word by word, which ignores the fact that following the same word sequence in the target language may actually doesn’t make any sense. This can be partly improved if we create a dictionary or term database that is linked to this system. But the NMT system may be better at delivering a more accurate translation as NMT better captures the context of full sentences before translating them.

By the time we finished this project, Microsoft has not opened its NMT system for customized training. But the good new is, THEY JUST DID THIS a week ago. Click here to learn more.

Personally, I’m quite interested in training an NMT engine and can’t wait to see the results. What I do believe is that no matter how unsatisfied we are with the output by existing MT, being able to utilize various techniques is a basic skill for any ambitious translator.

Technology is not always about some unreadable codes or algorithms, sometimes utilizing a simple tool like this will make your translation work much easier.

https://youtu.be/hau8DHpKoVY

There’s much more to be explored. Always being curious, eager to try, and open to latest technology is perhaps something most valuable I get from the CAT course at Middlebury Institute. Thank you, professor Adam Wooten.

(You can access the files of our SMT project here.)

Experience with TMS: lessons learned and a look into the future

May 15, 2018 / / 0 Comments

Introduction: This portfolio is a summary of what we’ve learned from the Translation Management System (TMS) course. It includes the description of a group project I’ve participated in, lessons learned through the project, and my personal ideas on how I would like to design a different translation management system.

The course covers general concepts behind TMS software. Using the SDL WorldServer web-based TMS, students were able to explore the functions and features of a translation management system from the point of view of a translator, project manager and administrator.

A pilot translation project for Foreign Affairs Journal

Basic Information

Client: Foreign Affairs Journal

Language Service Provider: Fish&Chips Localization.Inc

Source Files: Abstract of three articles posted on the website of Foreign Affairs Journal

Source Language: English(US)

Target Language: Chinese(ZH); Spanish(ES)

Word Count: 905(en-zh); 289(en-es)

Translation Management System: SDL WorldServer

The Use of WorldServer in This Project

WorldServer has been applied to each phase of the project, from preparation to finalization. For the preparation, WorldServer was used to create locales, workflow, workgroup, cost model, quality model, translation memories (TM), TM group, term database (TD), TD group, and project.

For the production, WorldServer was used to assign translation files to translators, translate files, and do quality assurance. For finalization, WorldServer was used to deliver products, inform the client and update TM and TD.

Obviously, with a TMS, the project workflow can be automated once the project is launched, all the related parties of this project can check in at any time to view or submit tasks, and relevant data can be updated in real time.

While using a TMS like WorldServer can bring the benefits listed above, not every step went smoothly as we were trying to incorporate WorldServer with our project.

Challenges and Solutions

  1. Project Creation

Launching a project needs a series of steps, which can be buggy since the localization engineer needs to set up everything from scratch without forgetting any tiny step or ignoring the sequence of these steps. When our team has set up everything and finally reached the project creation step, we got the following message:

(Error Message from WorldServer)

To fix this, we first thought there might be some errors in the setting of workflow. Did we forget to add assignee when we were setting up the workflow?

Then we added a new workflow role (see the screenshot below), but the creation of our project still failed. We were guessing, there must be somewhere in the workflow that allows us to change the setting of assignee.

(Setting New Workflow Role)

It turned out that when setting up the workflow, in the human step, i.e. translate/review, engineers can set the properties including who should be assigned to this step.

(Setting Human Step Properties)

It took our group a long time to figure this out, since we could not find much clue about this from WorldServer itself. For any project, large or small, spending much time on fixing bugs can be frustrating, which may arouse a question: do we need to use a TMS for a small and urgent project?

  1. Translation and Review

If a translator logs onto WorldServer, he/she can claim the assigned task, and start translation using WorldServer’s workbench. Translators can click on the “complete” button after finishing translation and the file will automatically go to reviewers’ task list. Similarly, reviewers can start QA checking using WorldServer’s workbench and deliver the file after completion. In WorldServer, the interface of translation workbench and the review one are the same, and once click on the “complete” button, translators and reviewers can no longer see the tasks through their portal.

We believe these may cause some troubles when translators or reviewers want to double check their work or track their task history. The reviewer’s workbench can also be problematic since QA check is different from translation. Features like adding notes, tracking changes, giving scores, identifying error patterns may be more helpful to a reviewer than the current feature, which is editing.

Reflection

What to consider when choosing a TMS?

Questions like “Should we adopt a TMS for your project/company?” “Which TMS should I choose” really don’t have exact answers. Based on the readings I’ve done this semester, there are several aspects you may consider when selecting a TMS.

(Thing to consider before choosing a TMS)

How would I like to design a TMS?

So far, I have worked with such TMS tools as WorldServer, LingoTek and XTRF. Some CAT tools, for example Trados and Wordfast, also have features related to translation management as they can perform tasks like automated file passing, and TM sharing.

These TMS tools have many great features such as access to server-based translation memories, terminology repositories and portals, customization and automation of translation workflows. But I’ve also discovered some defects as I’m using them.

Besides the issues mentioned above, WorldServer requires a dedicated installation environment and is very processor-intensive. It also has a limitation on supported web browsers and java versions. Additionally, WorldServer stores its data in an SQL database, which requires plenty of memories in a computer and data manipulation skills for people working with WorldServer. So, using such tools requires a huge amount of resource, including resource spent in educating people working with them.

TMS tools are mostly used by project managers in translation/localization industry, as it’s powerful in streamlining the workflow of a complex translation project. But vendors, clients, even developers can get involved in these projects. Many tools provide portals for vendors or clients, but few provides portals for developers. As you can tell from our pilot project, TMS like WorldServer works for processing such non-technical source files, as such localization does not involve developers.

The fact is, however, many products to be localized are websites, APPs, and software. Internationalization engineers need to get strings ready for translation separately. The exclusion of developers may make them unfamiliar with workflows of a localization project, and lead to a more tedious and disjointed cross-team cooperation and communication. Therefore, a developer-friendly TMS offering various APIs for developers to leverage would greatly benefit the seamless localization of a product.

TMS tools today are mostly set up on computers, not mobile devices. As tracking project progress and communication are important to project managers, it would be more convenient if there’s simplified version of TMS APP featuring tracking data, viewing progress, receiving instant notification, sending messages or even making phone calls. This should be helpful for project managers managing a team across time zones.

The world is transforming from mobile first to AI first, TMS tools need to embrace AI. Although some TMS tools have already made it possible for integrating various machine translation APIs, leveraging the power of AI is more than that. For example, an AI-driven TMS can make more accurate analysis and prediction, users will get a clear view of clients’ preferences, vendor’s strengths or weaknesses, thus, greatly improve the decision-making process.

(You can access our project files from here.)

My Blog Post Calendar

May 2018
S M T W T F S
 12345
6789101112
13141516171819
20212223242526
2728293031  

About my program

Tags

array CAT Chinese market CSS currency DTP FizzBuzz function GIS G suite HTML internationalization javascript localization loop Machine Translation Marriott MT MT Customization NMT PhotoShop PR crisis project management San Francisco School commuting software localization TMS Trados Translation Management System Wordfast WorldServer XTRF

© 2026 Ziqi Zhou

Theme by Anders Noren — Up ↑