How to Use Image to Text - Complete OCR Guide

Welcome to the comprehensive guide for using our Image to Text feature. This powerful OCR (Optical Character Recognition) tool allows you to extract text from images without translation, perfect for transcription, data entry, and text digitization tasks.

Overview

The Image to Text feature is designed for users who need to extract readable text from images quickly and accurately. Unlike the translation features, this tool focuses purely on text extraction powered by Google's advanced Gemini AI models, supporting multiple languages and offering flexible model choices based on your quality and budget needs.

Key Features at a Glance

• Advanced OCR Technology: Powered by Google Gemini AI models for high accuracy
• Multi-Language Support: Extract text in any language supported by Gemini models
• Three Model Tiers: Choose between quality and cost with Flash Lite, Flash, and Pro models
• Flexible Credit Pricing: 1-3 credits per image based on model selection
• Multiple Upload Methods: Drag-and-drop or file browser upload
• Instant Results: Fast text extraction with detailed statistics
• Copy to Clipboard: Quick one-click copying of extracted text
• Processing Metadata: Detailed information about OCR confidence and performance

---

Understanding Model Options

Before extracting text, it's important to understand the three available Gemini models and when to use each:

Model Comparison

When to Choose Each Model

Flash Lite (1 credit):

• Typed text on clean backgrounds
• Receipts and simple forms
• Street signs and basic labels
• When cost efficiency is priority
• Quick text extraction for non-critical tasks

Flash (2 credits) - Recommended Default:

• Mixed content (text + images)
• Documents with formatting
• Product labels and packaging
• Most general OCR tasks
• Balance between quality and cost

Pro (3 credits):

• Handwritten text
• Complex multi-column layouts
• Low-quality or blurry images
• Dense text with small fonts
• Critical accuracy requirements

---

Step-by-Step Usage Guide

Step 1: Configure OCR Settings

The first step card allows you to configure language and model preferences.

1.1 Select Output Language

Icon: Languages icon

Purpose: Choose the language for the extracted text output

How it works:

1. Click the "Output Language" dropdown
2. Browse available languages
3. Select your preferred language

Important Notes:

• Output language automatically syncs with your website locale setting
• You can manually override this selection
• The output language determines how the text is formatted and presented
• This does NOT translate the text - it only affects text formatting

Available Output Languages (16 languages):

• Arabic (ar)
• German (de)
• English (en)
• Spanish (es)
• French (fr)
• Italian (it)
• Japanese (ja)
• Korean (ko)
• Polish (pl)
• Portuguese (pt)
• Russian (ru)
• Chinese (zh)
• And more...

1.2 Select OCR Model

Icon: Sparkles icon with model information

Purpose: Choose the AI model that best fits your needs

How it works:

1. Review the three model options displayed as cards
2. Each card shows:
   • Model Name: Gemini 2.5 Flash Lite / Flash / Pro
   • Credit Cost: 1, 2, or 3 credits per image
   • Speed Indicator: Fast, Faster, Fastest
   • Quality Indicator: Good, Better, Best
3. Click the card to select your preferred model
4. Your selection is highlighted with a blue border
5. The credit cost is clearly displayed

Model Selection Tips:

• Start with Flash: The middle tier (2 credits) works for most use cases
• Upgrade to Pro: When accuracy is critical or image quality is poor
• Downgrade to Flash Lite: For simple, clear text on clean backgrounds
• Test First: Try Flash Lite on one image to see if quality meets your needs

Credit Cost Display:

• Shows credit cost (e.g., "2 credits per image") below the model card
• Updates in real-time when you change model selection
• Helps you budget credit usage before processing

Model Preference Saving:

• Your model selection is saved to browser local storage
• Next time you visit, your preferred model is pre-selected
• Output language always syncs with website locale (not saved)

---

Step 2: Upload Image

The second step card provides flexible upload options for your image.

2.1 Upload Methods

Method A: Drag and Drop (Recommended)

Icon: Upload icon

The fastest way to upload an image:

1. Select an Image:

   • Open your file explorer
   • Find the image file you want to extract text from
   • Drag the file into the upload area

2. Drop Zone Behavior:

   • The upload area highlights when you drag a file over it
   • Drop the file to add it
   • File is instantly validated and preview appears

Tip: The drag-and-drop area is large and easy to target.

Method B: Click to Browse

Icon: Upload icon

Traditional file selection method:

1. Click anywhere in the upload area
2. A file browser opens
3. Navigate to your image file
4. Click "Open" to upload

Upload Constraints

• Maximum File Size: 7MB per image
• File Formats: PNG, JPG, JPEG, WebP, HEIC, HEIF
• Single Image: One image at a time (no batch processing)

Error Handling:

• Unsupported formats are rejected with clear error messages
• Oversized files display size information and maximum allowed size
• File validation happens before upload

Format Notes:

• HEIC/HEIF Support: iOS photos are supported
• WebP Recommended: Modern format for best compression
• JPG/PNG: Universal compatibility

---

Step 3: Review Upload Preview

After uploading, you'll see a preview of your image with file information.

3.1 Preview Card

Displays:

1. Image Thumbnail: Visual preview of the uploaded image
2. Filename: Name of the uploaded file
3. File Size: Size in MB (e.g., "3.45 MB")
4. Format Badge: File type indicator

3.2 Replace Image

Icon: X (close) button

How to use:

1. Click the X button in the top-right corner of the preview
2. The image is removed
3. Upload area reappears
4. Upload a different image if needed

Use Case: Replace image if you uploaded the wrong file or want to try a different image.

---

Step 4: Start Text Extraction

Once your image is uploaded and settings configured, extract the text.

4.1 Initiate OCR Processing

Button: "Start OCR" with Sparkles icon

How to use:

1. Ensure you have:

   • Selected output language
   • Chosen OCR model (Flash, Flash Lite, or Pro)
   • Uploaded an image

2. Review credit cost displayed below the button:

   • Displays the credit amount based on selected model
   • Example: "2 credits per image" for Flash model

3. Click "Start OCR" button

4. Processing Flow:

   Stage 1 - Uploading Image (2-5 seconds)

   • Your image is securely uploaded to cloud storage
   • Progress indicator shows upload status
   • Button text changes to "Uploading..." with loader animation

   Stage 2 - Processing with AI (5-15 seconds)

   • Gemini AI model analyzes the image
   • Text regions are detected and extracted
   • Layout and formatting are preserved
   • Button text shows "Processing..." with loader animation

Processing Time:

• Flash Lite: 5-8 seconds (fastest)
• Flash: 8-12 seconds (moderate)
• Pro: 10-15 seconds (most thorough)

Credit Usage: Each OCR operation costs 1-3 credits based on model:

• Flash Lite: 1 credit
• Flash: 2 credits
• Pro: 3 credits

What Happens During Processing:

1. Image is uploaded to secure R2 storage
2. Gemini AI model receives the image
3. OCR analysis detects text regions
4. Text is extracted with confidence scores
5. Results are formatted and returned
6. You're redirected to the results view

---

Step 5: View Extracted Text Results

After processing completes, you'll see a comprehensive results view with the extracted text and detailed statistics.

5.1 Extracted Text Display

Main Content Area:

Card Elements:

1. Extracted Text Box:

   • White background with clear typography
   • Monospace font for accurate text representation
   • Scroll enabled for long text content
   • Preserves line breaks and spacing
   • Read-only text area

2. Text Information:

   • Character Count: Total number of characters extracted
   • Word Count: Total number of words (space-separated)
   • Line Count: Total number of lines

Text Formatting:

• Original line breaks are preserved
• Spacing and indentation maintained where possible
• Multi-column layouts converted to linear text
• Text direction respected (LTR/RTL)

5.2 Processing Statistics

Icon: Clock icon, Star icon, Badge icon

Displayed Metrics:

1. Processing Time:

   • Icon: Clock
   • Shows how long OCR took (e.g., "3.2s")
   • Faster times for simpler images

2. Model Used:

   • Icon: Sparkles
   • Displays which Gemini model processed the image
   • Example: "Gemini 2.5 Flash"

3. Output Language:

   • Icon: Languages
   • Shows the selected output language
   • Example: "English" or "简体中文"

4. Credits Used:

   • Icon: Star
   • Displays credit cost for this operation
   • Example: "2 credits" for Flash model

Additional Metadata (if available):

• OCR confidence score
• Number of text regions detected
• Image dimensions processed

---

Step 6: Take Action on Results

6.1 Copy Extracted Text

Button: "Copy Text" with Copy icon

How to use:

1. Click the "Copy Text" button
2. All extracted text is copied to your clipboard
3. Button text changes to "Copied!" with checkmark icon
4. Returns to "Copy Text" after 2 seconds
5. Paste the text anywhere you need (Ctrl+V or Cmd+V)

Use Cases:

• Paste into documents (Word, Google Docs, Notion)
• Import into spreadsheets for data entry
• Transfer to translation tools
• Save to text files
• Send via email or messaging apps

Keyboard Shortcut: You can also manually select text and copy with Ctrl+C (Windows) or Cmd+C (Mac)

6.2 Download Original Image

Button: "Download Image" with Download icon

How to use:

1. Click the "Download Image" button
2. Original uploaded image is downloaded
3. Filename format: ocr-source-[timestamp].webp

Use Cases:

• Archive the source image
• Compare with extracted text
• Share original with team members
• Keep record of processed images

File Format: WebP (high quality, small file size)

6.3 Extract New Image

Button: "Extract New" with RefreshCw icon

How to use:

1. Click the "Extract New" button
2. You're returned to the upload form
3. Your settings (model and language) are preserved
4. Upload area is reset
5. Start a new OCR operation

Use Cases:

• Process multiple images sequentially
• Try different images with same settings
• Compare OCR results across images
• Batch processing workflow (one at a time)

---

Advanced Features

Image Quality Optimization

For best OCR results, consider these image quality factors:

Resolution and Size

• Minimum: 1000 pixels wide
• Recommended: 1500+ pixels wide
• Maximum File Size: 7MB
• Optimal Format: PNG for text, JPG for photos

Lighting and Contrast

• Good Lighting: Ensure text is clearly visible
• High Contrast: Dark text on light background (or vice versa)
• Avoid Glare: No reflections or bright spots on text
• Even Illumination: Consistent lighting across the image

Text Clarity

• Sharp Focus: Avoid blurry or out-of-focus images
• Straight Alignment: Align text horizontally when possible
• No Distortion: Minimize skew, tilt, or perspective distortion
• Clean Background: Remove clutter around text areas

Supported Text Types

Works Best With

• Typed Text: Printed documents, books, screens
• Product Labels: Packaging, tags, stickers
• Street Signs: Traffic signs, store signs, banners
• Receipts: Purchase receipts, invoices, forms
• Screenshots: Software interfaces, web pages
• Scanned Documents: PDFs, office documents

Challenging Scenarios

• Handwritten Text: Use Pro model for best results
• Artistic Fonts: Decorative or stylized fonts may have lower accuracy
• Small Text: Magnify or use higher resolution images
• Overlapping Text: Separate text regions work better
• Vertical Text: Some languages handle vertical better than others

Language Support

The OCR engine supports text extraction in 100+ languages, including:

Latin Scripts:

• English, Spanish, French, German, Italian, Portuguese
• Dutch, Polish, Turkish, Czech, Romanian
• And many more European languages

Asian Languages:

• Chinese (Simplified and Traditional)
• Japanese (Kanji, Hiragana, Katakana)
• Korean (Hangul)
• Thai, Vietnamese, Indonesian

Other Scripts:

• Arabic, Hebrew (RTL languages)
• Russian, Ukrainian, Greek (Cyrillic and Greek scripts)
• Hindi, Bengali, Tamil (Indic scripts)

Multi-Language Detection:

• Automatically detects multiple languages in one image
• Extracts text from mixed-language content
• Preserves language boundaries

---

Workflow Examples

Example 1: Receipt Data Entry (Flash Lite Model)

Scenario: Extract text from a supermarket receipt for expense tracking

Steps:

1. Configuration:

   • Output Language: English
   • Model: Gemini 2.5 Flash Lite (1 credit)
   • Reason: Receipt has clear typed text

2. Upload:

   • Take photo of receipt with smartphone
   • Transfer to computer or upload directly
   • File size: ~2MB (well under 7MB limit)

3. Processing:

   • Click "Start OCR"
   • Processing time: ~6 seconds
   • Flash Lite handles clear printed text perfectly

4. Results:

   • Extracted text shows:
     • Store name and address
     • Purchase date and time
     • Itemized list with prices
     • Payment method
     • Total amount

5. Action:

   • Click "Copy Text"
   • Paste into expense tracking spreadsheet
   • Verify amounts match receipt
   • File receipt image for records

Outcome: Accurate text extraction for 1 credit, perfect for routine data entry.

---

Example 2: Book Page Transcription (Flash Model)

Scenario: Digitize a page from a printed book for note-taking

Steps:

1. Configuration:

   • Output Language: English
   • Model: Gemini 2.5 Flash (2 credits)
   • Reason: Book has dense text with varied formatting

2. Upload:

   • Scan or photograph book page
   • Ensure good lighting and focus
   • File size: ~4MB

3. Processing:

   • Click "Start OCR"
   • Processing time: ~10 seconds
   • Flash model handles multi-column layout

4. Results:

   • Extracted text includes:
     • Main body paragraphs
     • Headers and subheaders
     • Footnotes and page numbers
     • Character count: ~2,500
     • Word count: ~450

5. Action:

   • Click "Copy Text"
   • Paste into note-taking app (Notion, Evernote)
   • Format and organize notes
   • Add personal annotations

Outcome: High-quality transcription preserving text flow for 2 credits.

---

Example 3: Handwritten Notes (Pro Model)

Scenario: Extract text from handwritten meeting notes for sharing with team

Steps:

1. Configuration:

   • Output Language: English
   • Model: Gemini 2.5 Pro (3 credits)
   • Reason: Handwriting requires highest accuracy

2. Upload:

   • Photograph handwritten notes
   • Use good lighting and stable camera
   • Ensure handwriting is legible
   • File size: ~3MB

3. Processing:

   • Click "Start OCR"
   • Processing time: ~14 seconds
   • Pro model's advanced recognition handles handwriting

4. Results:

   • Extracted text captures:
     • Meeting agenda items
     • Action items and owners
     • Key discussion points
     • Dates and deadlines
   • Some errors expected with unclear handwriting

5. Action:

   • Click "Copy Text"
   • Paste into shared document
   • Manually review and correct any misrecognitions
   • Share with team members

Outcome: Best possible handwriting recognition for 3 credits, with manual review recommended.

---

Example 4: Multi-Language Document (Flash Model)

Scenario: Extract mixed English-Chinese text from a bilingual product manual

Steps:

1. Configuration:

   • Output Language: English (or Chinese, doesn't affect extraction)
   • Model: Gemini 2.5 Flash (2 credits)
   • Reason: Mixed languages require robust model

2. Upload:

   • Scan manual page with both languages
   • File size: ~5MB

3. Processing:

   • Click "Start OCR"
   • Processing time: ~11 seconds
   • Gemini automatically detects both languages

4. Results:

   • Extracted text includes:
     • English instructions
     • Chinese translations
     • Product specifications
     • Warning labels in both languages
   • Languages preserved in original layout order

5. Action:

   • Copy text for translation verification
   • Compare extracted text with original
   • Use for localization quality assurance

Outcome: Accurate multi-language extraction for 2 credits.

---

Tips for Best Results

Image Preparation

1. Use High Resolution:

   • At least 1000 pixels wide
   • Prefer 1500+ pixels for small text
   • Higher resolution = better accuracy

2. Ensure Good Lighting:

   • Natural light works best
   • Avoid shadows across text
   • No harsh glare or reflections
   • Consistent lighting across entire image

3. Straighten the Image:

   • Keep text horizontal when possible
   • Minimal skew or rotation
   • If using smartphone, use grid alignment

4. Clean Background:

   • Remove clutter when photographing
   • Plain backgrounds work best
   • Avoid busy patterns behind text

5. Sharp Focus:

   • Use camera autofocus
   • Tap to focus on text area (smartphones)
   • Avoid motion blur
   • Use tripod or stable surface

Model Selection Strategy

1. Start with Flash (Default):

   • Good balance of quality and cost
   • Works for 80% of use cases
   • 2 credits per image

2. Downgrade to Flash Lite:

   • When text is very clear and simple
   • Typed text on clean backgrounds
   • When processing many images with limited credits
   • Test one image first to verify quality

3. Upgrade to Pro:

   • When Flash results are unsatisfactory
   • For handwritten text
   • For complex layouts (tables, multi-column)
   • For critical accuracy requirements
   • When image quality is poor

4. Test and Iterate:

   • Try Flash Lite first on one image
   • If quality is insufficient, upgrade to Flash
   • Only use Pro when Flash is not accurate enough
   • This saves credits while maintaining quality

Text Extraction Optimization

1. Review Extracted Text:

   • Always read through extracted text
   • Check for missing sections
   • Verify numbers and dates are accurate
   • Look for misrecognized characters

2. Handle Errors:

   • Common misrecognitions: 0/O, 1/I/l, 5/S
   • Manually correct critical information
   • Note patterns of errors for future image improvements

3. Use Copy Strategically:

   • Copy all text first
   • Paste into editor for corrections
   • Keep original image for reference
   • Compare visually to catch errors

4. Batch Processing Workflow:

   • Process multiple images one by one
   • Use "Extract New" button between images
   • Keep model settings consistent
   • Track which images have been processed

---

Troubleshooting

Common Issues and Solutions

Issue: "No text detected"

Possible Causes:

• Image contains no readable text
• Text is too small or blurry
• Very low image quality
• Handwritten text with Flash Lite

Solution:

1. Verify image actually contains text
2. Check if text is clearly visible to human eye
3. Try higher resolution image
4. Upgrade to Flash or Pro model for challenging text
5. Ensure good lighting and focus

Issue: "File size too large (max 7MB)"

Possible Causes:

• High-resolution photo from camera
• Uncompressed image format
• Very large image dimensions

Solution:

1. Compress image before uploading:
   • Use online tools (TinyPNG, Squoosh)
   • Resize dimensions to reasonable size (1500-2000px wide)
   • Convert to JPG or WebP format
2. Crop unnecessary areas from image
3. Reduce image quality slightly (90% quality is usually sufficient)

Issue: "Unsupported file format"

Possible Causes:

• File is not an image format
• Using PDF, TIFF, or other document formats

Solution:

• Use supported formats only: PNG, JPG, JPEG, WebP, HEIC, HEIF
• Convert unsupported formats using image converters
• Screenshot PDF pages if needed (then process screenshot)

Issue: "Extracted text is garbled or incorrect"

Possible Causes:

• Using Flash Lite for complex text
• Handwritten text with insufficient model tier
• Low image quality (blurry, low contrast)
• Unusual fonts or artistic text
• Incorrect language detection

Solution:

1. Upgrade to Flash or Pro model
2. Use higher quality image:
   • Better lighting
   • Higher resolution
   • Sharper focus
3. For handwritten text, always use Pro model
4. Verify image orientation is correct
5. Crop to focus on main text area

Issue: "Extracted text missing sections"

Possible Causes:

• Text regions outside image boundaries
• Very small text not detected
• Low contrast text (gray on white)
• Text obscured by shadows or glare

Solution:

1. Ensure entire text area is visible in image
2. Increase image resolution
3. Improve lighting for better contrast
4. Try Pro model for better region detection
5. Process sections separately if needed

Issue: "Processing very slow"

Possible Causes:

• Large image file size
• Using Pro model (naturally slower)
• High server load

Solution:

1. Compress image to smaller file size
2. Flash Lite processes faster than Pro
3. Try again during off-peak hours
4. Check your internet connection speed

Issue: "Insufficient credits"

Solution:

• Visit Pricing to purchase more credits
• Subscribe to a plan with monthly credits
• Free tier provides 10 credits/month
• Consider Flash Lite (1 credit) for routine tasks

Issue: "Layout/formatting lost"

Explanation:

• OCR extracts plain text, not formatting
• Multi-column layouts become linear text
• Tables are converted to text sequences
• Visual spacing may not be preserved perfectly

Solution:

• Manually reformat text after extraction
• For complex layouts, consider:
  • Processing sections separately
  • Using dedicated table OCR tools
  • Manually recreating structure

---

Credit System for OCR

How Credits Work

• Flash Lite Model: 1 credit per image
• Flash Model: 2 credits per image
• Pro Model: 3 credits per image

Credit Packages

Free Tier:

• 10 credits/month
• Access to all three models
• Suitable for occasional use

Professional Plans:

• Pro Plan: 800 credits/month
• Enterprise Plan: 2,000 credits/month
• Includes other platform features

One-Time Credit Purchases:

• Available for pay-as-you-go users
• Credits never expire
• Purchase from Pricing page

Credit Efficiency Tips

1. Test with Flash Lite First:

   • Try Flash Lite (1 credit) on one image
   • If quality is good, use it for similar images
   • Saves 50-66% credits compared to other models

2. Match Model to Task:

   • Don't use Pro for simple receipts
   • Don't use Flash Lite for handwriting
   • Right model = better results with fewer retries

3. Optimize Images First:

   • Good image quality means lower model tier needed
   • One Pro attempt (3 cr) vs. Flash (2 cr) + retake = same cost
   • Invest time in image quality to save credits

4. Track Your Usage:

   • Monitor credit balance in header
   • Note which model works for which task
   • Plan credit purchases for your workload

---

Comparison: OCR vs Translation

When to Use Image to Text (OCR)

• Purpose: Extract text only, no translation
• Best for:
  • Data entry and digitization
  • Transcription of printed materials
  • Text archiving
  • Copy-paste workflows
  • Text analysis and processing
  • Creating searchable text from images

When to Use Image Translation

• Purpose: Extract AND translate text
• Best for:
  • Foreign language documents
  • Manga and comics
  • International product labels
  • Multilingual signage
  • Cross-language content adaptation

Feature Comparison

Workflow Integration

Typical Workflows:

1. OCR First, Then Translate:

   • Use Image to Text to extract text
   • Copy extracted text
   • Paste into translation tool for editing
   • Best when you need to edit before translating

2. Direct Translation:

   • Use Image Translator for immediate translated image
   • Best when layout preservation is critical
   • Ideal for manga, signs, marketing materials

---

Getting Help

Documentation

• Single Image Translator: Homepage Guide
• Batch Translator: Batch Translator Guide
• Manga Translator: Manga Translator Guide

Support Channels

• FAQ: Pricing & FAQ
• Contact: Contact Page
• Community: Join our Discord for peer support

---

What's Next?

Now that you've mastered the Image to Text feature, explore our other tools:

• Image Translator: Translate text within images while preserving layout
• Batch Image Translator: Process up to 1000 images simultaneously
• Manga Translator: Specialized tool for manga and comics (100 images)

Start extracting text from images effortlessly and streamline your digitization workflow!