Pdftools OCR Service release notes

Learn about new updates in the OCR Service, including new features, improvements, and bug fixes.

Version 1.1.2

2 April 2026

Added

• Configurable communication and timeout settings for the manager node, such as worker response timeouts, concurrent-request limits, and job-polling intervals. The following configuration options were added:

  • JobPollingIntervalMs
  • WorkerHttpTimeoutMinutes
  • MaxConcurrentRequests
  • SafetyTimeoutMinutes

  For default values and how to change them, review Manager configuration overview.

• Docker images for Pdftools OCR Service include a Software Bill of Materials (SBOM) and a provenance attestation, so you can verify the contents and build origin of each image.

• Updated the underlying ABBYY FineReader Engine to version 12.7.4, improving recognition stability when the engine encounters invalid characters and the appearance of separators in exported tables.

Fixed

• Before this update, if communication with a worker node failed during background OCR processing, the affected job could remain without a result. With this update, the service reports the failure as a job error with error details. As a result, clients receive a clear error response instead of a job that never completes.
• Before this update, deleting a job during processing could cause its status request to fail with an internal error. With this update, the service detects the missing job and stops polling. As a result, the request returns a clear “not found” response.
• Before this update, processing certain PDF files could cause the OCR engine to crash. With this update, the underlying issue was resolved. As a result, Pdftools OCR Service processes affected documents reliably.
• Updated third-party dependencies to resolve known security vulnerabilities.

Version 1.1.1

8 March 2026

Fixed

• Before this update, an internal licensing operation could fail silently in certain Licensing Gateway Service (LGS) configurations. With this update, the issue is resolved and licensing works as expected in all supported setups.

Version 1.1.0

6 March 2026

Added

• Docker container support for Pdftools OCR Service. With this release, the Manager and Worker nodes can be deployed as Docker containers. For more details, review OCR Service in Docker.
• Added support for multiple OCR languages: Arabic, Chinese, Hebrew, Japanese, Korean, and Thai.
  Additional language support

  Before you use Arabic, Chinese, Hebrew, Japanese, Korean, or Thai in OCR analysis, contact us through the Contact page for exact pricing information. OCR analysis of these languages has additional costs.

• Added predefined profiles: DataExtraction, DocumentConversion_Normal, HighCompressedImageOnlyPdf, BusinessCardsProcessing, and MachineReadableZone. For details on each profile, review Predefined profiles.
• Added INI file sections for custom profiles:
  • [DocumentStructureDetectionParams] for configuring headline, footnote, caption, column, list, and table of contents detection.
  • [SortingBlocksParams] for controlling how Pdftools OCR Service combines recognized text blocks into groups.
  • [TextLayerInjectionParams] for configuring searchable PDF creation with text layer injection.
• Persistent connections with automatic retry and reconnect logic for interservice communication between processes within a Worker node. Worker service processes maintain long-lived connections instead of establishing a new connection for each request, reducing connection overhead.

Changed

• Removed predefined profiles:
  • DocumentConversion_Speed
  • BookArchiving_Accuracy
  • BookArchiving_Speed
• All predefined OCR profiles use updated parameter names and settings. If you use custom profiles that reference or extend predefined profiles, review and update your configurations after upgrading. For details, review Predefined profiles and the following section Parameter changes by INI file section.

Parameter changes by INI file section

The following sections describe changed, moved, and removed parameters in each INI file section.

[PrepareImageMode]

The following parameters moved to other INI file sections or have new replacements:

Removed [PrepareImageMode] parameters	Replacement
Rotation	Use RotationType in [ImageProcessingParams]
InvertImage, MirrorImage	Moved to [ImageProcessingParams]
CorrectSkew, CorrectSkewMode	Use CorrectSkewMode in [PagePreprocessingParams]
AutoOverwriteResolution, OverwriteResolution, XResolutionToOverwrite, YResolutionToOverwrite	Use OverwriteResolutionMode and ResolutionToOverwrite in [PagePreprocessingParams]
DiscardColorImage, UseFastBinarization	Moved to [PagePreprocessingParams]
EnhanceLocalContrast	Use BackgroundWhitening in [PagePreprocessingParams]
PhotoProcessingMode	Use ImageSourceType in [PrepareImageMode] and DetectImageType in [PagePreprocessingParams]
ImageCompression	Use CompressImageMode in [PrepareImageMode]
BackgroundFillingColor, CreatePreview, PreviewHeight, PreviewWidth	Removed with no replacement

New [PrepareImageMode] parameters:

• CompressImageMode
• DocumentType
• ImageSourceType
• RasterizeFreeText

[PagePreprocessingParams]

Removed [PagePreprocessingParams] parameters	Replacement
CorrectOrientation (Boolean)	CorrectOrientationMode (enum: COM_Auto, COM_Clockwise, COM_CounterClockwise, COM_UpsideDown, COM_No)
CorrectShadowsAndHighlights	Removed with no replacement
CorrectSkew + CorrectSkewMode (6-value flag enum)	Single CorrectSkewMode with new enum values: CSM_Auto, CSM_Fast, CSM_Off
GeometryCorrectionMode	CorrectGeometry (ThreeStatePropertyValueEnum)
ResolutionCorrectionMode	OverwriteResolutionMode and ResolutionToOverwrite

• Type change: CorrectInvertedImage changed from Boolean (default false) to ThreeStatePropertyValueEnum (default TSPV_Auto)
• New [PagePreprocessingParams] parameters:
  • StraightenLinesMode
  • BackgroundWhitening
  • CropImage
  • DetectImageType
  • DiscardColorImage
  • UseFastBinarization
  • SplitType
  • OverwriteResolutionMode
  • ResolutionToOverwrite

[PageAnalysisParams]

Removed [PageAnalysisParams] parameters:

• EnableTextExtractionMode: replaced with AnalysisMode (set to PAM_TextExtraction).
• AggressiveTableDetection: Removed with no replacement
• DetectMultipleBusinessCards: Removed with no replacement
• ProhibitModelAnalysis: Removed with no replacement

New [PageAnalysisParams] parameters:

• AnalysisMode
• SpeedQualityMode
• DetectHandwritten
• DetectCheckmarks
• DetectStamps
• PaperSizeDetectionMode
• DetectTextOnPictures (moved from [ObjectsExtractionParams])
• CollectPdfExportData

[RecognizerParams]

Removed [RecognizerParams] parameters:

• BalancedMode (Boolean): Replaced with Mode (set to RM_Normal)
• FastMode (Boolean): Replaced with Mode (set to RM_Fast)
• WritingStyle (all WS_* values): Removed with no replacement

Enum rename:

• TT_Handprinted renamed to TT_Handwritten in TextTypeEnum.

New [RecognizerParams] parameters:

• Mode
• DetectTextTypesIndependently
• ProhibitSmallCaps

[BarcodeParams]

Change	Details
Removed [BarcodeParams] parameters	MinRatioToTextHeight, EnableAdvancedExtractionMode, IsCode39WithoutAsterisk
New parameter	EnableBarcodesCheck
New BarcodeTypeEnum values	BT_AutodetectWithoutPostal, BT_Australia4State, BT_JapanPost, BT_KIX, BT_RoyalMail4State
Removed BarcodeSupplementTypeEnum value	BS_Unknown
New CodePageEnum values for added languages	CP_Hebrew, CP_Hebrew_ISO, CP_Hebrew_Macintosh, CP_Hebrew_MSDOS, CP_Japan_Mac, CP_Japan_SJIS, CP_Korean, CP_Korean_Johab, CP_Korean_Mac, CP_Thai, CP_Thai_Macintosh

[FontFormattingDetectionParams]

• Moved parameter: DetectDropCaps moved to [PageAnalysisParams].
• Default change: DetectSpacing default changed from true to false.

[ObjectsExtractionParams]

• Moved parameter: DetectTextOnPictures moved to [PageAnalysisParams].
• New [ObjectsExtractionParams] parameter: SourceContentReuseMode.

[SynthesisParamsForPage]

New [SynthesisParamsForPage] parameters:

• DetectDocumentLinks
• DetectHyperlinks
• SynthesizeBusinessCards
• AllowGrayTextColor

Removed [OrientationDetectionParams]

• The [OrientationDetectionParams] INI file section was removed. Use CorrectOrientationMode in [PagePreprocessingParams] instead.

Fixed

• Before this update, Worker service processes established a new connection to the main Worker service for each request. In rare cases, a race condition during connection setup caused request failures. With this update, persistent connections with automatic reconnect logic resolve this issue.

Version 1.0.2

10 November 2025

Fixed

• Fixed SDK corruption issues with ABBYY FineReader Engine that could cause the Pdftools OCR Service to become unstable or crash during processing. As of this update, the service automatically detects corruption and recreates the OCR engine when needed.
• Fixed failures when processing large files by increasing the default maximum request body size from 30 MB to 100 MB. As a result, processing of large TIFF files and multi-page documents no longer fails due to file size limitations.
• Improved error handling and logging for failed OCR jobs, with better exception tracking and proper error code reporting.

Changed

• The maximum request body size is now configurable using the MaxRequestBodySizeBytes setting in the manager node appsettings.json file, allowing administrators to adjust the limit based on their specific requirements. The configured default is 100 MB (specifically, it is defined as 104857600 bytes).

Version 1.0.1

15 September 2025

Fixed

• Fixed PostgreSQL database migration compatibility issues to ensure correct schema updates when upgrading from previous versions.
• Improved OCR validation and file creation processes to enhance reliability and error handling during document processing.

Version 1.0.0

30 July 2025

Added

• First release of the Pdftools OCR Service introduces a new OCR engine that is fully compatible with current 3-Heights® OCR Service workflows and configurations. It acts as a drop-in replacement for existing installations, requiring no changes to your integration or processing pipelines. Compared to the previous OCR configurations (the ABBYY FineReader using your license key and the legacy 3-Heights® OCR Service), the new Pdftools OCR Service also provides:
  • Streamlined installation and license key configuration.
  • Scalability options to support higher throughput.

Changed

• Compared to the 3-Heights® OCR Service, this release of the Pdftools OCR Service doesn’t include support for the BusinessCardsProcessing predefined OCR profile.