IPSR-VIDEO Altera, IPSR-VIDEO Datasheet - Page 123

• Manufacturer Part Number: IPSR-VIDEO
• Description: RENEWAL Of IPS-VIDEO
• Manufacturer: Altera
• Series: IP Suitesr

Specifications of IPSR-VIDEO

• Software Application: IP CORE, SUITES
• Supported Families: Arria GX, Cyclone II, HardCopy II, Stratix II
• Features: Common Avalon Streaming (Avalon-St) Interface And Avalon-St Video Protocol
• Core Architecture: FPGA
• Core Sub-architecture: Arria, Cyclone, Stratix
• Rohs Compliant: NA
• Lead Free Status / RoHS Status: na

Chapter 5: Functional Descriptions
Deinterlacer
January 2011 Altera Corporation
Frame Buffering
1
1
The motion-adaptive algorithm requires the buffering of two frames of data before it
can produce any output. The Deinterlacer always consumes the three first fields it
receives at start up and after a change of resolution without producing any output.
The weave and motion-adaptive algorithms cannot handle fields of different sizes (for
example, 244 lines for F0 and 243 lines for F1). Both implementations discard input
fields and do not produce an output frame until they receive a sufficient number of
consecutive fields with matching sizes.
Pass-Through Mode for Progressive Frames
In its default configuration, the Deinterlacer discards progressive frames. Change this
behavior if you want a datapath compatible with both progressive and interlaced
inputs and where run-time switching between the two types of input is allowed.
When the Deinterlacer lets progressive frames pass through, the deinterlacing
algorithm in use (bob, weave or motion-adaptive) propagates progressive frames
unchanged. The function maintains the double or triple-buffering function while
propagating progressive frames.
Enabling the propagation of progressive frames impacts memory usage in all the
parameterizations of the bob algorithm that use buffering.
The Deinterlacer MegaCore function also allows frame buffering in external RAM,
which you can configure at compile time. When using either of the two bob algorithm
subtypes, you can select no buffering, double-buffering, or triple-buffering. The
weave and motion-adaptive algorithms require some external frame buffering, and in
those cases only select double-buffering or triple-buffering.
When you chose no buffering, input pixels flow into the Deinterlacer through its
input port and, after some delay, calculated output pixels flow out through the output
port. When you select double-buffering, external RAM uses two frame buffers. Input
pixels flow through the input port and into one buffer while pixels are read from the
other buffer, processed and output.
When both the input and output sides have finished processing a frame, the buffers
swap roles so that the frame that the output can use the frame that you have just
input. You can overwrite the frame that the function uses to create the output with a
fresh input.
The motion-adaptive algorithm uses four fields to build a progressive output frame
and the output side has to read pixels from two frame buffers rather than one.
Consequently, the motion-adaptive algorithm actually uses three frame buffers in
external RAM when you select double-buffering. When the input and output sides
finish processing a frame, the output side exchanges its buffer containing the oldest
frame, frame n-2, with the frame it receives at the input side, frame n. It keeps frame n-
1 for one extra iteration because it uses it with frame n to produce the next output.
When triple-buffering is in use, external RAM usually uses three frame buffers. The
function uses four frame buffers when you select the motion-adaptive algorithm. At
any time, one buffer is in use by the input and one (two for the motion adaptive case)
is (are) in use by the output in the same way as the double-buffering case. The last
frame buffer is spare.
Video and Image Processing Suite User Guide
5–43