TranslateImage

Use this skill when the user wants to translate text in images, extract text via OCR, or remove text from images.

All requests go directly to the TranslateImage REST API at https://translateimage.io using curl.

Setup

Set your API key (get one at https://translateimage.io/dashboard):

CODEBLOCK0

All endpoints require:

Authorization: Bearer $TRANSLATEIMAGE_API_KEY

Image Input

All tools accept images as multipart file uploads. Handle the input type like this:

CODEBLOCK2

Only fetch URLs the user explicitly provides. Do not fetch URLs from untrusted sources.

Tools

Translate Image

Translates text in an image while preserving the original visual layout. Returns the translated image as a base64-encoded data URL.

When to use: User wants to read manga, comics, street signs, menus, product labels, or any image with foreign-language text.

Endpoint: INLINECODE1

Form fields:

• - image (file, required) — The image to translate (JPEG, PNG, WebP, GIF — max 10MB)
• INLINECODE3 (JSON string, required) — Translation options:

Example:
CODEBLOCK3

Response (JSON):
CODEBLOCK4

Save the translated image:

RESULT=$(curl -s -X POST https://translateimage.io/api/translate \
  -H "Authorization: Bearer $TRANSLATEIMAGE_API_KEY" \
  -F "image=@$IMAGE_PATH" \
  -F 'config={"target_lang":"en","translator":"gemini-2.5-flash"}')

# Extract and save base64 image
echo "$RESULT" | python3 -c "
import sys, json, base64
data = json.load(sys.stdin)
img = data['resultImage'].split(',', 1)[1]
with open('/tmp/translated.png', 'wb') as f:
    f.write(base64.b64decode(img))
print('Saved to /tmp/translated.png')
"

Extract Text (OCR)

Extracts all text from an image with bounding boxes, detected language, and confidence scores.

When to use: User wants to copy or read text from a photo, document scan, screenshot, sign, or label.

Endpoint: INLINECODE31

Form fields:

• - image (file, required) — The image to process

Example:
CODEBLOCK6

Response (JSON):

{
  "text": "All extracted text joined by newlines",
  "language": "ja",
  "regions": [
    {
      "bounds": { "x": 10, "y": 20, "width": 200, "height": 40 },
      "languages": { "ja": "detected text in this region" },
      "probability": 0.97
    }
  ]
}

Remove Text

Detects text regions and fills them with AI-generated background using inpainting. Returns a clean image.

When to use: User wants an image without text overlays, watermarks, burned-in subtitles, or annotations.

Endpoint: INLINECODE33

Form fields:

• - image (file, required) — The image to process

Example:
CODEBLOCK8

Response (JSON):

{
  "cleanedImage": "data:image/png;base64,..."
}

Image to Text (AI OCR + Translation)

Uses Gemini AI for high-quality text extraction. Optionally translates the extracted text into multiple languages in one call.

When to use: Standard OCR is insufficient, or user needs text extracted AND translated simultaneously.

Endpoint: INLINECODE35

Form fields:

• - image (file, required) — The image to process
• INLINECODE37 (JSON string, optional) — INLINECODE38

Example — extract only:
CODEBLOCK10

Example — extract and translate:
CODEBLOCK11

Response (JSON):
CODEBLOCK12

API Scopes

Each endpoint requires a specific scope on your API key:

Configure scopes when creating your API key at https://translateimage.io/dashboard.

Error Handling

CODEBLOCK13

Common errors:

Important Considerations

• - Always confirm the target language with the user before translating
• For manga/comics use WildWords or BadComic fonts for an authentic look
• For Chinese content use MaShanZheng; for Korean use INLINECODE50
• Images over 5MB may take longer — inform the user
• Inpainting works best on simple backgrounds; complex textures may show artifacts
• INLINECODE51 is the recommended default translator — fast and high quality
• Clean up temp files after processing: INLINECODE52