Flitto Image Translation - All within a single API for global expansion

Flitto's one-stop AI image translation service delivers fully localized images for global expansion while preserving original text formatting (size, style, and thickness).

Our ML-powered technology applies context-aware translations (e.g., cosmetics, fashion, brand etc.) through an LLM-based proprietary AI model.

• Supported image formats and dimensions

• JPEG | JPG | PNG | WEBP

• width/height each no more than 5,000px

• file size no more than 20MB

• Supported languages

• ko: Korean

• en: English

• ja: Japanese

• zh-CN: Chinese(Simplified)

• zh-TW: Chinese(Traditional)

We provide the translated result in 2 formats below to serve different needs:

1. Inpainting image & bounding box textual data:

This format can be used as a blueprint for rendering HTML by including data such as bounding-box coordinates, CSS/CanvasRenderingContext2D attributes that adheres to Web Standards. Simple, effortless integration with existing client frameworks like React/Angular/Vue/Svelte as well as Vanilla HTML/JS can be utilized to customize user-interaction with the images as well as boost SEO optimization.

2. Fully rendered translated images:

We also provide fully rendered images that preserves original textual attributes, dimension and content types.

This product can be seamlessly integrated with your prior AWS infrastructure, or you can quickly provision the required services automatically through the provided quick-launch template. Launching our product will require you to provide certain permissions for specific services outlined below:

1. S3 Bucket (can be newly provisioned or you can select your existing bucket at launch) :

Vendor can read / write items from a certain path (flitto-image-tr/*). Public ACL is not required for this resource (nor is it recommended)

2. SQS (can be newly provisioned or you can selected your existing queue at launch):

Vendor can publish messages to this queue.

Subscription lifecycle and API key rotation:

1. Subscribing to this product will redirect you to our page for your registration step. Email and contact details must be provided to complete this step.

2. On completing the first step above, you will be directed back to the AWS marketplace to complete your launch. Use the quick-launch template we have provided to automatically provision required services.

3. If your CloudFormation stack creation from step 2 is successful, the stack output will provide you with your API key. This key can immediately be used for requesting image translation.

4. If you need to rotate your API key, unsubscribe and repeat the subscription process from step 1. Your previous key will be immediately inactivated when you unsubscribe and a new key will be issued on subscription.

Image translation request lifecycle:

1. Authenticate your requests to our image translation endpoint with your API token provided at product launch.

2. If your authentication & provided image file passes validation, an new image translation job in a "processing" state will be created and the API will respond with a identifier for this job (job_id) that can be used to track the progress of each job.

3. On completion, the job is marked with the "complete" status. Also the translation result (rendered imaged & HTML blueprint JSON data) will be uploaded to your custom S3 bucket, and a success message(information about the job & translation file s3 path) will be published to your SQS.

4. While the job is in a processing state, you can abort the job using the job_id provided at request. This will update the job's status to "aborted". This will publish a failed message to the SQS queue.

5. The vendor can reject some jobs that violate our content policy or inputs not feasible for translation. This will update the job's status to "rejected" and publish a failed message to the SQS queue.