Q+A CENTER
Latest question:
How do you think the new GigE standards will influence the machine vision industry?
Respond or ask your question now!
Advanced Imaging Magazine
Most Read
Most E-mailed
Print Article
OpenGL 2.0: Uniting Graphics and Imaging
FEATURE
OpenGL 2.0: Uniting Graphics and Imaging
A new class of programmable parallel processors is contained in existing graphics cards and optimized for pixel processing. Can we harness all this power?
by John Kessenich, Barthold Lichtenbelt, Randi Rost & Neil Trevett
June 2003
Click Here to view the GL slang shader code.
The imaging and graphics industries have been unusually disconnected considering both are based on processing pixels. The imaging community is often input-focused-analyzing pixel data with dedicated hardware for frame-rate throughputs-whereas the graphics community is more output-focused and has developed sophisticated processing pipelines to translate scene descriptions into increasingly realistic imagery.
Because
of these different objectives, the hardware and software infrastructures
of the graphics and imaging industries have grown up separately, with
different APIs, hardware architectures and participating companies. There
have been some notable exceptions, however. For example, the OpenGL standard
has contained imaging functionality since its initial release in 1992,
making it unique among 3D graphics APIs. The imaging functionality provided
by version 1.4 includes scaling and biasing color values, convolutions
and high-quality color space conversion. The texturing capabilities of
3D hardware can also be used to scale, translate, rotate and warp imagery,
but the majority of OpenGL graphics boards sold today have been in the
traditional 3D markets of CAD and digital content creation, not imaging
applications. That is about to change.
A FUNDAMENTAL REVOLUTION
Traditional 3D accelerator pipelines were hardwired, consisting of a series
of fixed function blocks, each performing a step in generating 3D imagery.
New generations of graphics devices, increasingly called Visual Processors
or VPUs, are instead constructed from programmable processor arrays, sometimes
integrating hundreds of floating point processors onto a single device,
which can be flexibly programmed for any task, not just 3D graphics.
