Transcript DIF

DV over IP
Team 18
Björn Thorsén Hallberg
Eva Codina Sanuy
Fredrik Blomqvist
Johan Fredriksson
Tobias Karlsson
May 2000
Agenda
•
•
•
•
•
Overview
Technologies used
Business ideas
Video
Questions
DV over IP
Main goal: develop a
Remote Lecture System using DV
on Linux Platforms
RTP
CIP
DIF DIF DIF DIF
DIF DIF DIF
DV Server
FireWire
IP network
UDP socket
RTP
DIF DIF DIF DIF
DIF
DIF
DIF
DV Client
RTP
CIP
DIF DIF DIF DIF
DIF DIF DIF
DV Server
FireWire
IP network
UDP socket
RTP
DIF DIF DIF DIF
DIF
DIF
DIF
DV Client
Digital Video (DV)
• provides high-quality video data
• individual frame compression
• supports NTSC and PAL video formats
Digital Interface Format (DIF)
(120.000 bytes)
DV Frame
(12.000 bytes)
DIF sequence 0
Header
Section
DIF block 0
DIF block 1
DIF block 2
DIF sequence 1
Subcode
Section
VAUX
Section
DIF sequence n-1
Audio & Video
Section
NTSC: n=10
PAL : n=12
DIF block 149
FireWire (IEEE 1394)
• High Perfomance Serial Bus
• Characteristics:
– Low cost
– High-bandwidth
• Topology:
– Peer-to-peer protocol
– Automatic bus configuration
FireWire (IEEE 1394) cont.
• Benefits:
–
–
–
–
–
–
Guaranteed bandwidth
Hot plugging
Scalability
Flexibility
Universal interconnection
Cable power
Common Isochronous Packet (CIP)
• FireWire supports isochronous transfer
isochronous = equal time
• isochronous data transfer
– ensures constant data rate
– time-dependent
• 6 DIF blocks in each CIP packet
GNU/Linux IEEE1394 Subsystem
The driver hierarchy:
Real-time Transport Protocol (RTP)
• provides end-to-end delivery services for
data with real-time characteristics
• services:
– timestamping
– sequence numbering
– payload type identification
• “sublayer” of the transport layer
Server application – DV Server
Main program:
There are two processes working with shared
memory:
– Read CIP packets from the FireWire
– Send the DIF blocks encapsulated in RTP packets
through the UDP socket to the IP network.
Client application – DView
• Functionality: decode and display DV data
• Input: series of frames
frame = full video image + 1/30 sec. sound + aux. data
• Algorithm (3 steps)
1. Decode DIF blocks
2. Run iDCT (inverse-Discrete-CosineTransformation)  luminance & chrominance
3. Convert YCrCb to RGB and draw the data on
the screen
Client application – DView
Options to improve the perfomance:
1. Downscaling: reduces the displayed size of the
video
2. Reduced rendering precision
3. Grayscaling: removes the need of:
- decode and render the chrominance
- convert from YCrCb to RGB
KnowDV
• The objective of the company:
– provide real-time video transfers
• How will we succeed?
–
–
–
–
unique and useful products
user-friendly interface
consultant services
multi-platform solutions
DV Lecturing System
• The product: DV Lecturing System
– Cheap
– High quality video
– Multi-platform
• Target groups:
– Companies
– Universities
– Private users
DV Lecturing System
• Competitors:
– No similar product for Linux platforms yet
– Applications with less bandwidth
requirement
– Conferencing systems
DV Lecturing System
• Market segmentation:
Questions ??