DMO Architecture
[The feature associated with this page, DirectShow, is a legacy feature. It has been superseded by MediaPlayer, IMFMediaEngine, and Audio/Video Capture in Media Foundation. Those features have been optimized for Windows 10 and Windows 11. Microsoft strongly recommends that new code use MediaPlayer, IMFMediaEngine and Audio/Video Capture in Media Foundation instead of DirectShow, when possible. Microsoft suggests that existing code that uses the legacy APIs be rewritten to use the new APIs if possible.]
This section describes the overall architecture of a DMO.
Streams
A DMO is an object that takes m inputs and produces n outputs. The inputs and outputs are called streams. Every DMO has at least one stream. Streams are not objects; they are simply referenced on the DMO by index number. The number of streams is fixed at design time.
Media Types
All data is typed using a media type, which defines how to interpret the contents of the data. For example, 320 x 240 24-bit RGB video is one type; 44.1-kilohertz (kHz) 16-bit stereo PCM audio is another type. Media types are described using the DMO_MEDIA_TYPE structure. Before the client can process any data, it must set the media type for each stream on the DMO.
Typically, a stream can accept a range of media types. Some DMOs support a wider range of types than others. The DMO interfaces define methods for the client to discover the supported types. For example, one DMO might support RGB video at any bit depth, while another might support only 24-bit RGB. Also, a DMO might be limited to certain combinations of inputs and outputs. For example, if the input type is 16-bit video, the output stream might require the same bit depth. The client can enumerate each stream's preferred types and then test specific combinations.
Buffers
In the default DMO model, the client allocates separate input buffers and output buffers. It fills the input buffers with data and delivers them to the DMO, and the DMO writes new data into the output buffers.
Optionally, a DMO can support "in-place" processing. With in-place processing, the DMO writes the output directly into the input buffer, over the original data. In-place processing eliminates the need for separate buffers. On the other hand, it alters the original data, which may not be acceptable for some applications.
The default (non-in-place) buffering model is supported through the IMediaObject interface. All DMOs must implement this interface. If a DMO supports in-place processing, it also exposes the IMediaObjectInPlace interface. The client is responsible for allocating all buffers, both input and output.
Related topics