JUMBF

The JPEG Universal Metadata Box Format (JUMBF) is an international standard that defines a universal container format for embedding any type of metadata in box-based JPEG file formats.^[1] It is standardized as Part 5 of the ISO/IEC 19566 JPEG Systems series (ISO/IEC 19566-5), maintained by the ISO/IEC Joint Technical Committee JTC 1, Subcommittee 29.^[1][2]

The standard defines the syntax of the JUMBF box and the mechanism to assign specific content types, including XML, JSON, CBOR, embedded files, codestreams, and UUID-typed boxes.^[3] It also defines the syntax to reference or request embedded metadata content within or outside the image.^[1] JUMBF has been adopted as the underlying metadata container by several important standards and industry initiatives, most notably the Coalition for Content Provenance and Authenticity (C2PA) Content Credentials framework and the JPEG 360 standard for omnidirectional images.^[4][5]

JUMBF was first published as ISO/IEC 19566-5:2019 in 2019 as part of the JPEG Systems family of standards, which provides file formats, transport mechanisms and metadata frameworks for JPEG-coded image data.^[2][6] The standard was subsequently revised and superseded by ISO/IEC 19566-5:2023, which expanded support to include CBOR and embedded file content types, in addition to the XML, JSON, codestream and UUID types covered in the 2019 edition.^[1][2] A 2021 amendment to the 2019 edition (ISO/IEC 19566-5:2019/Amd 1:2021) added support for embedding mixed codestreams, but was subsequently withdrawn when the 2023 edition was published.^[7]

In March 2025, an amendment to the 2023 edition was published (ISO/IEC 19566-5:2023/Amd 1:2025), introducing JUMBF box compression and support for standalone JUMBF files that can exist independently of a host image file.^[8] A further revision (ISO/IEC AWI 19566-5) is under development as a replacement for the 2023 edition.^[3]

JUMBF is built on the ISO Base Media File Format box structure, providing compatibility with the JPEG XT box format (ISO/IEC 18477-3) and the JPEG File Interchange Format (JFIF, ISO/IEC 10918-5).^[5] The fundamental building block of JUMBF is the superbox, which groups a description box and one or more content boxes into a single logical unit.^[5][9]

Each JUMBF superbox consists of the following components:^[5][9]

• A superbox header, comprising an LBox field (the total byte length of the superbox) and a TBox field whose value is the four-character code jumb (hexadecimal 0x6A756D62).
• A JUMBF Description Box (type jumd, 0x6A756D64), which always appears first and specifies the UUID identifying the content type, a one-byte toggle field indicating the presence of optional label, ID and signature fields, and an optional human-readable label string.
• One or more JUMBF Content Boxes, whose internal structure and semantics are determined by the UUID declared in the description box.

JUMBF boxes can be nested: the content boxes of a superbox may themselves be complete JUMBF superboxes, enabling hierarchical metadata structures.^[10] Padding boxes are also permitted within a JUMBF superbox to allow space reservation or alignment.^[10]

The standard defines normative content types for common data formats. As of the 2023 edition, these include:^[1][3]

The standard specifies a URI-based referencing and requesting mechanism (defined in Annex C of ISO/IEC 19566-5) that allows JUMBF boxes to be referenced from within the same file or from external locations using self#jumbf URIs.^[1][11] This mechanism enables cross-referencing between different metadata boxes inside a complex JUMBF hierarchy or linking to external metadata resources.^[1]

For embedding JUMBF boxes in traditional JPEG-1 images (as defined in ITU-T T.81 / ISO/IEC 10918-1), the standard uses the JPEG XT box format (ISO/IEC 18477-3), which is carried in the APP11 marker segment of the JPEG bitstream.^[10][9][11] Because a single JPEG-1 marker segment is limited to 65,535 bytes, large JUMBF payloads may be split across multiple contiguous APP11 segments.^[11]

The most prominent application of JUMBF is as the binary container format for C2PA (Coalition for Content Provenance and Authenticity) Content Credentials, the open technical standard for establishing the provenance and authenticity of digital media.^[12] C2PA was founded in February 2021 by Adobe, Arm, BBC, Intel, Microsoft, and Truepic.^[13]

In the C2PA framework, a manifest store is encoded as a JUMBF superbox and embedded in the media file. This manifest store contains one or more C2PA Assertion boxes, each of which is itself a JUMBF superbox carrying assertion data in CBOR, JSON, embedded file or UUID content types.^[11] A C2PA Claim box and optional Credential Store are also defined as JUMBF superboxes with specific labels and UUIDs.^[11] The entire structure is cryptographically signed, and any modification to the asset invalidates the manifest, enabling tamper detection.^[12]

C2PA uses the JUMBF referencing mechanism (ISO/IEC 19566-5, Annex C) to cross-reference assertions within the same manifest store using self#jumbf URIs.^[11] JUMBF box compression (introduced in the 2025 amendment) is also used in C2PA to reduce manifest size in compressed manifests.^[11]

It enables "the cryptographic storage of information such as time of creation, author, processing steps and authentication data in image and video files."^[14] Platforms such as LinkedIn already display C2PA provenance data carried in JUMBF containers, and companies including Meta and Google have announced plans to support the standard.^[14] Cloudflare has integrated JUMBF-based Content Credentials preservation into its image transformation pipeline.^[15]

ISO/IEC 19566-6 (JPEG 360), the international standard for 360-degree omnidirectional images, uses JUMBF as its metadata embedding mechanism. The JPEG 360 standard defines a dedicated JPEG 360 Content Type JUMBF superbox and specifies the structure and syntax of an XML box for 360-degree image metadata.^[5] A 2021 peer-reviewed study published in Applied Sciences (MDPI) demonstrated a practical tool for embedding JPEG 360 metadata into standard JPEG files using JUMBF, highlighting the role of the standard in enabling interoperability between 360-degree camera hardware and virtual reality services.^[5]

A Java reference implementation of ISO/IEC 19566-5, the mipams-jpeg-systems library (version jumbf-2.0), is available as open source and implements the complete JUMBF data model.^[16] The C2PA open-source SDK (c2pa-rs, written in Rust) also implements JUMBF embedding and parsing as part of its content credentials toolchain.^[13]

• Exif
• XMP (metadata)
• IPTC Information Interchange Model
• Coalition for Content Provenance and Authenticity
• JPEG
• ISO base media file format

• ISO/IEC 19566-5:2023 on ISO.org
• JPEG Systems overview on JPEG.org
• C2PA Technical Specification 2.2
• Java reference implementation of JUMBF on GitHub