[opendtv] Re: Video compression artifacts and MPEG noise reduction

  • From: Craig Birkmaier <craig@xxxxxxxxx>
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Mon, 16 Jun 2008 09:42:04 -0400

At 5:35 PM -0400 6/14/08, Albert Manfredi wrote:

No, it's not just that. Read again their web site. Let me quote the parts you missed:

---------Begin-------------------
http://www.algolith.com/products/index.html

Solutions for Broadcast & Professional Markets

Algolith offers a series of products, called Algogear?, which are designed to significantly improve the way broadcast, post production, cable, satellite and IPTV service providers deliver digital video content to their viewers and subscribers.

Here you go trying to change the terms of engagement again.

Yes Algolith offers a range of products for broadcasters that are used across a range of applications. But this is irrelevant to this thread, which YOU started by posting a story about a specific product for the removal of artifacts from content "damaged" by excessive MPEG-2 compression.

My response was concerned with your misunderstanding of the uses and application of this specific product. Apparently you still do not understand that this product has no application for the construction of a broadcast multiplex, which may cause such damage to its content.


So, given that most broadcasters create their own multicasts, rather than simply rebroadcasting a ready-made 19.39 Mb/s stream from a network feed, individual broadcasters could put such a product to good use.

Obviously there are a range of pre-processing products that can help to improve the delivered quality of the streams that make up a digital multiplex. I already discussed the use of noise reduction and closed loop low-pass filtering techniques. To this you can add statistical multiplexing and other stream grooming techniques.

But post-processing artifact reduction techniques cannot be used until the damage is already done.

Exactly, Craig. You said the right words, but then you drew the wrong conclusion, missing what makes this clever.

Sorry Bert, but it is you drawing the wrong conclusion, which is why i have tried to clarify your misunderstandings.


 Most commercial MPEG-2 encoders DO include noise
 reduction and pre-filtering that operates in closed loop with
 the encoder. If the encoder is stressed, the low pass filter
 is used to reduce encoder stress.

And they specifically addressed this, Craig. They are pre-processing in a more clever way than merely low-pass filtering.

NO. The tread was about POST-PROCESSING using clever techniques to remove compression artifacts after the damage has been done - This simply is not possible at a the origin of a broadcast - it can only be applied to the undo the damage to an overly compressed MPEG-2 bitstream, as in a TV receiver.


Why should it be so hard to believe that with higher speed processors, the encoding of MPEG-2 (H.262) can be improved? It seems totally believable that whatever pre-filtering MPEG-2 encoders have been doing can and will be made more clever.

Pre-filtering is a BAND-AID Bert. An attempt to overcome the limitations of the MPEG-2 algorithm. Yes, there are clever ways to build processors and encoders that go well beyond the tools available in the MPEG-2 toolbox to create an MPEG-2 compliant encoded bitstream (note that the h.262 and h.264 only define the characteristics of the encoded bitstream, not the methodology of encoding).

But efforts to build technology around a legacy standard face the reality of diminishing returns, while expansion of the compression toolbox via new standards creates the opportunity to improve quality at both ends of the chain - i.e. the encoder AND the decoder. And it fuels real innovation, rather than an emphasis on putting band-aids on something that is causing the trauma.


I'm not saying that Algolith has the only answer. I'm saying that this is an on-going process, and Algolith has brought it to the forefront, with what might be the most advanced solution as of today.

The most advanced solution to what Bert?

Certainly not pre-encoder processing as you continue to try to claim.

Regards
Craig


----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts: