
Zoran ZR36506 USBV ision – Datasheet
Zoran Proprietary and Confidential
8-10
Video Input Interface
The ZR36506 compressor compresses specific frames using its “Intra” mode and all others using
its “Inter” mode. Intra mode does not require any previous frame, while Inter mode is always
based on a reconstructed previous frame. “Intra” frames provide the algorithm with relative
resistance to error propagation between frames but consume more bits per pixel than “Inter”
frames. Error propagation within the frame itself (from higher lines to lower lines) is eliminated
by dividing the frame into many horizontal strips.
Y UV 4:2:2 I nterleaved Format
In this format, the ZR36506 transfers to the host computer 2 bytes for every pixel (16 bits/pixel).
The Y-component is available for every pixel, but the U and V components are each available
for every second pixel (Y0, U0, Y1, V2, Y2, U2, Y3...).
Y UV 4:2:0 Planar Format
In this format, the ZR36506 transfers to the host computer 3 bytes for every 2 pixels (12
bits/pixel). The Y-component is available for every pixel, but the U and V components are only
available for every second pixel in even lines.
In this mode, the host computer receives a frame preformatted in the planar mode; this can save
CPU time in the host computer in most video conferencing applications. The Y and U/V
components are packed by the ZR36506 in 64-byte packets, and have the following structure:
Packet Number
Contents
1
Pixels 0-63 of the Y-Image.
2
Pixels 64-127 of the Y-Image.
3
Pixels 0-63 of the U-Image (or V-Image).
The U-Image and V-Image are half-size of
the Y-Image in both horizontal and vertical
dimensions. The ZR36506 produces a line of
U pixels followed by a line of V pixels, and
then U pixels again in a toggling manner.
These pixels are always packed in this
modulo-3 packet. The host computer -
knowing the frame-size - can separate
between U and V components.
4
Pixels 128-191 of the Y-Image.
5
Pixels 192-255 of the Y-Image.
6
Pixels 64-127 of the U-Image (or V-Image).