Media processing

Create a processing task

A processing task is a step within a processing flow. You can configure the output of a processing task to be an entity rendition. To do this, enable the Store output feature for a task and create rendition links.

To create a processing task:

On the menu bar, click Manage .
On the Manage page, click Media processing.
On the Media processing page, select a processing set, then click the flow where you want to add a task.
On the visual canvas, on the node of the flow you want to add the task to, click .
In the right-hand pane, select a task type.
On the Parameters tab, fill out the task parameters.
Click Save task.

Task parameters

You can use the following task parameters.

Store file

The Store file task stores a copy of the incoming source file or an upstream output directly into file storage as a specific rendition or entity attachment, without modifying its content or format.

Parameter	Description
Name	Name of the task.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.

Convert image

The Convert image task transforms image assets by resizing, cropping, changing color profiles, adjusting DPI, or converting file formats to produce web-friendly previews, thumbnails, and custom download renditions.

Note

You cannot upscale or enlarge images. This is to minimize unnecessary processing and to ensure there is not a reduction in image quality, such as blurriness or stretching.

Parameter	Description
Name	Name of the task. This must match the output name on the Outputs tab.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
Resize option	Method used to resize images. This parameter is mandatory.
Target extension	File extension of resulting output.
Width (px)	Width of the output image. Parameter not available if Resize option is None.
Height (px)	Height of the output image. Parameter not available if Resize option is None.
Resize algorithm	Algorithm used for cropping. The available algorithms are: Four corners (default) - examines the color of the pixel in each corner of the image. If all corner pixels are the same color, the image is identified as a logo. Histogram - builds a histogram of all the pixel colors in the image. Based on criteria such as sharp peaks, dominant peaks, or color omission, the image might be identified as a logo. Smart - uses the entropy cropping of the NetVips library. It tries to automatically find the point-of-interest of the image. Center - crops the image according to the Height and Width parameters, and uses the middle of the image as its focus point. Note If an image is identified as a logo, the contents of the image are reduced to 75% of their original size, while the overall size of the image remains the same. This results in an empty border around the original image, which is then filled with the color detected at all four corners of the original.
Color profile	Color profile applied on the output image. This parameter is mandatory.
Color space	Color space applied on the output image. This parameter is mandatory and only available if Color profile is None.
Density	DPI used for the output image. This parameter is mandatory.
Multi-page	Determines whether renditions should be created for all pages of the file, or only the first.
Clipping path	Name of the clipping path you want to crop the image to.
Embedded preview priorities	An ordered, comma-separated list of previews. The processing agent checks for availability of these previews in the RAW file metadata. If a preview is available, the agent applies default image conversion to it to generate the target image. If none of the listed previews are available, the agent applies default image conversion to the original image instead.

The following is an example of RAW file metadata extracted with the Exiftool command exiftool -preview:all image:


exiftool -preview:all RAW_CANON_EOS_5DMARK3.CR2
Preview Image          : (Binary data 3192436 bytes, use -b option to extract)
Thumbnail Image         : (Binary data 18733 bytes, use -b option to extract)

To add these previews to the Embedded preview priorities field, remove the spaces in the names. For example write PreviewImage or ThumbnailImage.

Convert video

The Convert video task encodes and transcodes video files into standard playback formats (such as MP4) and generates preview frames, allowing you to configure quality bitrates, dimensions, and segment durations.

Parameter	Description
Name	Name of the task.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
Seek	The time point in a video from which the rendition starts. For example, with a value of 5, the output file starts at the fifth second of original file.
Length	Duration of the output file in seconds.
Number of preview frames	Maximum number of frames extracted from the video for preview creation.
Target extension	File extension of resulting output.
Width (px)	Width of the output image.
Height (px)	Height of the output image.
Bitrate (% from original)	Defines the quality of the rendition based on the original file.

Behavior of the Bitrate (% from original) parameter:

The value of this parameter can range from 1 to 100. If an invalid value is used, this parameter will default to 30.
The default value, based on YouTube's 360p recommended minimum (1 Mbps videostream, 96khz audiostream) serves as minimum quality parameters. If the calculated target bitrate is lower, the default bitrate is used.
To prevent upscaling, if the calculated target bitrate is higher than the original, that original bitrate is applied instead.

Convert document

The Convert document task converts office documents and text-based files into standard PDF formats and renditions, with optional full-text extraction to make document content searchable across Content Hub.

Note

You cannot convert documents to images directly. You must first convert the document to a PDF and then convert the PDF to an image.

Parameter	Description
Name	Name of the task.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
Multi-page	Determines whether renditions should be created for all pages of the file, or only the first.
Low resolution	Defines whether the output file uses compression methods to reduce the file size.
Extract content	Determines whether text content is extracted from supported files and added to the search index, making document content searchable. Enable this option only if document content needs to be indexed and searchable, as content extraction can increase the amount of indexed data. There is no file size or page limit for PDF or Word document extraction. Extracted content, however, must not exceed 5 million characters or 800-1000 pages of standard text (depending on the document format and structure). Note As content approaches the defined limits, performance might be impacted because search indexing processes the full text. This can result in longer indexing times and slower search queries, depending on the environment and usage. The impact on storage from extracted content, however, is minimal.

Convert HTML

The Convert HTML task processes raw HTML files to generate standardized text representations and basic previews of HTML-based asset content.

Parameter	Description
Name	Name of the task.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.

The HTML media processor only renders the text without applying layouts on it. It also does not render images referenced in the HTML or use the CSS to style the HTML. Thus, the preview you see on the asset details page might differ from how you see the HTML asset in a web browser.

Create storyboard

The Create storyboard task analyzes video files to extract visual frame sequences and generate storyboard strips and seekbar thumbnail grids for interactive video scrubbing in the media player.

Parameter	Description
Name	Name of the task.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
Target extension	File extension of resulting output.
Seekbar thumbnail - Width	Width of the thumbnail when seeking in a video. This parameter is mandatory.
Seekbar thumbnail - Height	Height of the thumbnail when seeking in a video. This parameter is mandatory.
Seekbar thumbnail - Amount	Amount of frames used for seeking in a video. This parameter is mandatory.
Story thumbnail - Width	Width of the thumbnail for the storyboard rendition. This parameter is mandatory.
Story thumbnail - Height	Height of the thumbnail for the storyboard rendition. This parameter is mandatory.
Story thumbnail - Amount	Amount of frames used for the storyboard rendition. This parameter is mandatory.

Extract clipping paths

The Extract clipping paths task reads embedded vector clipping paths from source images (such as Adobe Photoshop files) so they can be identified and reused for targeted cropping operations.

Parameter	Description
Name	Name of the task.

Extract image embeddings

The Extract image embeddings task processes image assets with machine learning models to generate high-dimensional vector embeddings, enabling AI-assisted visual search and similarity matching across your repository.

Parameter	Description
Name	Name of the task.

This task is automatically added to the Media processing flow when you enable visual search. It generates embeddings renditions that are used by the AI-assisted visual search feature to find visually similar images. This task has no configurable parameters and is required for visual search to work properly.

Extract metadata

The Extract metadata task parses file headers and embedded container data (such as EXIF, IPTC, and XMP) to extract technical and descriptive metadata directly from the source asset.

Parameter	Description
Name	Name of the task.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.

Extract vision data

The Extract vision data task leverages AI computer vision services to automatically analyze images for object detection, text recognition (OCR), visual features, tags, and orientation.

Parameter	Description
Name	Name of the task. Warning We recommend you do not change the name of the task. Use the default vision task name analysis.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
Endpoint	Send a request to either the OCR endpoint for text recognition both to the Analyze endpoint for image features recognition. This parameter is mandatory.
Detect orientation	Defines whether the OCR feature detect the orientation of text.
Minimum confidence	Determines the minimum accuracy level of the analysis (default value: 0.5). Note Changing the minimum confidence level only affects analysis performed after the change; it has no effect on previous analysis. This parameter is mandatory.
Features	Features to extract.
Details	Types of content to analyze.

Extract video indexer data

The Extract video indexer data task analyzes video streams to extract advanced cognitive insights, including facial recognition thumbnails, spoken audio transcripts, topics, and keyframe snapshots.

Parameter	Description
Name	Name of the task.
Content type	ContentType/MimeType for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
Image format	Output format for the storyboard and face renditions.
Thumbnail height	Height for the face thumbnail renditions. This parameter is mandatory.
Thumbnail width	Width for the face thumbnail renditions. This parameter is mandatory.
Key frame height	Height for the key frame renditions. This parameter is mandatory.
Key frame width	Width for the key frame renditions. This parameter is mandatory.

Run external web task

The Run external web task is a media processing task in Sitecore Content Hub that allows you to offload asset processing to an external service or API. When configured in a media processing flow, Content Hub sends the asset source and processing context to an external endpoint (such as an Azure Function or custom web service), which processes the file and reports the results back asynchronously.

The Content Hub request includes the following parameters:

Parameter	Description
Name	Name of the task.
Content type	`ContentType/MimeType` for the output file.
Content disposition	Defines whether the file is an attachment or provided inline. This parameter is mandatory.
URL	URL of the external task.
Headers	JSON containing web request headers.
Parameters	JSON containing web request parameters. These parameters include the following settings: `callback` - the callback URL that the external service calls once processing is complete. `sources` - an array of source files URLs for the asset being processed. `other parameters` - custom key-value pairs or configuration parameters defined in the task matrix.

The request that the external service sends back to the callback URL after processing is complete includes the following parameters:

Parameter	Description
Locations	The URLs of the output files to ingest as renditions.
Value/properties	JSON metadata objects to map to asset or file properties.

If you have suggestions for improving this article, let us know!