Enuclia Semiconductor, a start-up company working on HDTV electronics, used the Display Industry Investment Conference in New York City yesterday, to unveil its vision of the future of video processing for HDTVs. While short on specifics, company CEO and President Bruce Berkoff did reveal the vision his company is working on, which they call "Pipeline".

Chris Chinnock
Sr. Analyst and Sr. Editor
of Insight Media

The "Holy Grail" of the DTV electronics industry is a single-chip solution. But this is not necessarily the lowest-cost solution. Chip-set solutions may actually be more cost-effective, and many vendors provide such solutions, packing various combinations of functional processing block into their chips. Berkoff said current DTV processing architectures have a lot of weaknesses.

At each stage in the process, typical architectures must verify the incoming signal format, perform trade-off analyses on the process tasks and then do the functional processing. This is repeated at the next functional block. As a result, errors in an early stage of the processing can be multiplied as signals propagate through the electronics, particularly where there is redundancy in processing cores and memories and there are too many liner steps in the architecture.

In addition, the way the functional integration is done in chip sets is not efficient. "Most vendors just bundle modules of functional processing to turn two of three chips into one chip," said Berkoff, "But this is not an efficient way to do the integration and it can actually slow the time to market for the TV customer. These architectures have a long way to go."

Price compression in TVs is well documented and it filters down top the electronics providers as well. To meet the demands for lower prices and better image quality, designers need to take a clean-slate approach to looking at the architectures for responding to HDTV video processing needs. It just so happens that this is exactly what Enuclia has done. According to Berkhoff, there are some key video processing elements needed to optimize image quality. These include:

• Detail
• Motion
• Noise
• Reconstruction
• Scaling
• Custom

The problem is, each video frame may require a different combination of the above elements. Current solutions do this in a linear, module-to-module fashion. In the Enuclia approach, as best as we can gage, the idea is to do this more in a parallel processing approach. Once the processing elements for a scene are done, the frame can be processed in the most effect manner using a common set of processing gates. The next scene may need different element, so it follows a different path through the processing channel, but reuses the same fundamental processing core. This save space and cost.

Plus, it will more readily allow TV integrators to tailor the look and feel of the image to provide differentiation from the same processing chip set. This vision sounds good, especially delivered by Berkoff, but we will need to delve a lot deeper to see if there are real advantages to this approach. The company says it plans to reveal more soon, so stand by.