interMedia Exchange Tool

In the Oracle8i era, Oracle interMedia was a bundled feature set that helped developers store and work with unstructured and “rich” content inside the database: text documents, images, audio, and video. Over time, Oracle split these capabilities into separate product areas. The text search portion matured into Oracle Text, while the in-database multimedia processing portion was renamed Oracle Multimedia, later deprecated, and ultimately desupported in Oracle Database 19c.

What interMedia was in Oracle8i–10g

Oracle interMedia originally covered multiple “content” domains under one umbrella. In practice, this included:

• interMedia Text (iMT) for full-text indexing and document search (wildcards, fuzzy matches, proximity, ranking).
• Multimedia object types (for example, ORDImage, ORDVideo, ORDAudio) intended to store media data and related attributes in object-relational structures.
• Client/server workflows that let a Java client retrieve media from the database, modify it on the desktop, and write it back.
• Visual Information Retrieval, a legacy feature that attempted image similarity querying (historical only).

Evolution to Oracle Multimedia (Oracle 11g–12c)

Beginning in Oracle Database 11g Release 1, Oracle renamed “Oracle interMedia” to Oracle Multimedia. The scope and APIs were broadly similar—the change was primarily branding and product packaging.

Deprecation in 18c and desupport in 19c

Starting in Oracle Database 18c, Oracle Multimedia was deprecated. In Oracle Database 19c, it was desupported and the implementation was removed. In many upgraded environments, legacy objects may still appear in the data dictionary, but attempts to use the multimedia processing APIs can fail because the underlying implementation is no longer present.

What plays an analogous role in 19c, 23c, and 23ai

There is no single, built-in “drop-in replacement” for the full interMedia feature set in modern Oracle releases. Instead, the capabilities are handled by a combination of features and architectural patterns:

Legacy interMedia capability	Modern Oracle approach (19c / 23c / 23ai)
Full-text indexing and search	Oracle Text (CTXSYS indexes, CONTAINS queries, ranking, linguistics)
In-database image/audio/video processing (resize/convert/thumbnails)	No built-in equivalent; move processing to the application tier or external services
Media storage inside the database	SecureFiles BLOBs when DB storage is required; or store media externally and keep references + metadata in tables
Media metadata and flexible attributes	Relational columns + JSON (and duality-view modeling patterns in newer releases)
Content-based discovery	Keyword search (Oracle Text) and, when appropriate, semantic search patterns (embeddings + vector search)

Migration strategy from Oracle Multimedia / interMedia

If you maintain a legacy schema that used ORDImage/ORDVideo/ORDAudio or related packages, a practical migration plan usually looks like this:

1. Inventory usage: identify tables, types, and stored procedures that reference ORDSYS multimedia types and methods.
2. Separate concerns:
   • Storage: keep binary content in SecureFiles BLOBs (or move to object storage and store a URI).
   • Processing: perform transforms (resize, rotate, transcode) outside the database using libraries/services.
   • Metadata: store attributes (width/height/duration/codec/tags) in relational columns or a JSON document.
3. Replace in-database transforms: if legacy PL/SQL performed media processing, migrate that logic to application code and update procedures to persist results back to BLOBs or external storage.
4. Modernize search: keep document and text search in Oracle with Oracle Text; use structured JSON indexing for JSON-path queries.

The text portion of interMedia did not disappear—it evolved. What Oracle8i documentation calls interMedia Text (iMT) maps most directly to Oracle Text in current releases. Oracle Text provides full-text indexing and querying over large documents and document-like content.

Era	Text capability naming
Oracle 8i	interMedia Text (iMT) — full-text indexing and content queries inside the database
9i–10g	interMedia Text continues; Oracle’s text indexing stack becomes more standardized around CTXSYS
11g–23ai	Oracle Text — CONTEXT/CTXCAT/CTXRULE index families, CONTAINS queries, linguistic features, ranking

A safe, portable pattern is to keep JSON as JSON (for structure and path queries) and to create a derived text column (virtual or materialized) for full-text search with Oracle Text. This avoids assumptions about direct indexing of native JSON storage.

Example: JSON column + virtual CLOB for Oracle Text indexing

CREATE TABLE documents (
  doc_id   NUMBER PRIMARY KEY,
  content  JSON
);

-- Create a derived text projection for full-text indexing
ALTER TABLE documents ADD (
  content_text CLOB
    GENERATED ALWAYS AS (JSON_SERIALIZE(content RETURNING CLOB))
    VIRTUAL
);

-- Create an Oracle Text index on the derived text column
CREATE INDEX documents_content_text_idx
  ON documents(content_text)
  INDEXTYPE IS CTXSYS.CONTEXT;

-- Full-text query across the JSON document text
SELECT doc_id
FROM documents
WHERE CONTAINS(content_text, 'deep learning') > 0;

Use JSON indexing (for example, JSON search indexes) for targeted key/path queries, and Oracle Text for keyword-based discovery across the full document.

Historically, interMedia offered a database-centered workflow for multimedia. A Java client could retrieve a media object, edit it locally, and return the updated object to the database. This model is useful to understand the original intent of interMedia, but it is not a supported design baseline for Oracle 19c+ multimedia processing.

Visual Information Retrieval was historically positioned as a companion to interMedia for image similarity searching. In modern architectures, comparable outcomes (image similarity, classification, face/object detection) are typically implemented with external computer-vision services or libraries rather than in-database multimedia object types. Oracle’s in-database evolution focuses on text search (Oracle Text) and, for semantic use cases, embedding-based search patterns.

Relevance ranking: Relevancy ranking orders a result set so that records most likely to match a user’s intent appear near the top, reducing the time required to find useful items.