I'd expect it to be more like when you get blocky artifacts, it breaks those blocks down into noisier and less uniform shapes so they appear more natural and less digital. Like how an old VHS tape with image degradation still looks natural whereas a realplayer stream doesn't.
Yep, that's right. The artifact blocks will be broken up by the noise making them far less visible to the naked eye. I know in theory it sounds like it's replacing one problem with another, but the concept is a good idea. Many artists use a similar technique to create less visible lines.
Quote:
Originally Posted by oakie
if you have the bandwidth to send randomized noise that attempts to "motion-blur the eff out of compression artifacts," why not just encode the original file with less compression, thus more effectively utilizing the bandwidth?
The article is talking about hardware on the client machine. The hardware will buffer the video content and apply the algorithm after analysing each frame. This does not effect the source file at all.
Quote:
Originally Posted by kerryn
I think this is more a client technology in the player, rather than sending the noise in the stream.
Big difference between a ubiquity argument and a quality one. Yes, agreed, Flash video is everywhere. But no, the quality isn't very good in an absolute sense and it's absolutely awful if you factor in quality for consumed processing power.
I use the ClickToFlash Safari plugin.
It loads any web page that includes Flash (this one for example)and replaces each Flash player window with a gray box of the same size. You can click on the box and load Flash or (if h264 is available) play the content with QuickTime (plugin or app).
Pages load 2-3 times faster, there is no irritating "Flam" (Flash Spam), no Flash exposures (security, memory leaks, CPU hog).
As to quality, if h.264 is available, it looks noticeably better when played with QuickTime than Flash.
Fuzzy Blocking Artifacts Reduction Algorithm Based on Human Visual System
Wei-Bin Zhao Zhi-Heng Zhou
Guangdong Police Officers Coll., Guangzhou;
This paper appears in: Machine Learning and Cybernetics, 2007 International Conference on
Publication Date: 19-22 Aug. 2007
Volume: 3, On page(s): 1626-1630
Location: Hong Kong,
ISBN: 978-1-4244-0973-0
INSPEC Accession Number: 9797175
Digital Object Identifier: 10.1109/ICMLC.2007.4370406
Current Version Published: 2007-10-29
Quote:
Abstract
Traditional blocking artifacts reduction algorithm often causes image edges losing. To address this problem, a fuzzy edge-sensitivity blocking artifacts reduction algorithm is proposed. Based on some characteristics of human visual system, this algorithm applies adaptive filter to current pixel by integrating fuzzy logic technique. The experimental results show that this algorithm can obtain better results than traditional ones.
Quote:
A Simple Algorithm for the Reduction of Blocking Artifacts in Images and Its Implementation (1998) [6 citations — 1 self]
by Roberto Castagno , Stefano Marsi , Gianni Ramponi
Quote:
Abstract:
In this paper a simple but effective operator for the reduction of blocking artifacts is presented, together with a simple hardware implementation. The method is based on the Rational Filter approach: the operator is expressed as a ratio between a linear and a polynomial function of the input data. Such filters proved to outperform other conventional methods in other applications, such as noise smoothing [1], thanks to their capability of adapting gradually to the local image characteristics. The filter is capable of biasing its behaviour in order to achieve good performance both in uniform areas, where linear smoothing is needed, and in textured zones, where nonlinear and directional filtering is required. A detector of activity is embedded in the expression of the operator itself so that the biasing of the behaviour of the filter is smooth and not based on fixed thresholds. A solution for the hardware implementation of the scheme is presented in detail. Despite the simplifications im...
P.S. Ummm... I seem to have guessed right what it may be... They now have OpenCL and GC to build software filters being fast enough...
Apple has investigated a method to make imperfections in compressed video files less visible to the human eye, a new patent application revealed this week shows.
What about cats' eyes? Cats like watching YT videos of other cats.
This article immediately made me think, "Aren't they describing some sort of proprietary dithering process?"
There are a number of licensed dither algorithms in the digital audio world. Aren't there any in the video world?
The differences in audio dithering algorithms are subtle, if even distinguishable by the untrained ear. Surely dithering algorithms exist for video. Could this be that much of a breakthrough? Sounds like incremental improvement at best (at first read).
Increased bandwidth is the real answer to quality video, IMO. Period.
I haven't read the patent, only the AI article, but it sounds like it's adaptive dithering. In audio, you're essentially working in one dimension, so you add noise over the entire audio file. In video, you have 3 dimensions (x, y, time) so you can choose to apply this dithering to certain places at certain times.
Also, what I wrote earlier about FLV not being up there with more advanced codecs, was me trying to write in a quick way that Flash simply isn't the answer, without just saying "Yeah, but Flash is rubbish."
I just hope that one day, in the far future, maybe there will be a 1,024-core mac that can play a youtube video over a broadband link without stuttering.
That has nothing to do with the processor but bandwidth speeds.
Comments
I'd expect it to be more like when you get blocky artifacts, it breaks those blocks down into noisier and less uniform shapes so they appear more natural and less digital. Like how an old VHS tape with image degradation still looks natural whereas a realplayer stream doesn't.
Yep, that's right. The artifact blocks will be broken up by the noise making them far less visible to the naked eye. I know in theory it sounds like it's replacing one problem with another, but the concept is a good idea. Many artists use a similar technique to create less visible lines.
if you have the bandwidth to send randomized noise that attempts to "motion-blur the eff out of compression artifacts," why not just encode the original file with less compression, thus more effectively utilizing the bandwidth?
The article is talking about hardware on the client machine. The hardware will buffer the video content and apply the algorithm after analysing each frame. This does not effect the source file at all.
I think this is more a client technology in the player, rather than sending the noise in the stream.
Exactly.
Big difference between a ubiquity argument and a quality one. Yes, agreed, Flash video is everywhere. But no, the quality isn't very good in an absolute sense and it's absolutely awful if you factor in quality for consumed processing power.
I use the ClickToFlash Safari plugin.
It loads any web page that includes Flash (this one for example)and replaces each Flash player window with a gray box of the same size. You can click on the box and load Flash or (if h264 is available) play the content with QuickTime (plugin or app).
Pages load 2-3 times faster, there is no irritating "Flam" (Flash Spam), no Flash exposures (security, memory leaks, CPU hog).
As to quality, if h.264 is available, it looks noticeably better when played with QuickTime than Flash.
I love it!
*
Fuzzy Blocking Artifacts Reduction Algorithm Based on Human Visual System
Wei-Bin Zhao Zhi-Heng Zhou
Guangdong Police Officers Coll., Guangzhou;
This paper appears in: Machine Learning and Cybernetics, 2007 International Conference on
Publication Date: 19-22 Aug. 2007
Volume: 3, On page(s): 1626-1630
Location: Hong Kong,
ISBN: 978-1-4244-0973-0
INSPEC Accession Number: 9797175
Digital Object Identifier: 10.1109/ICMLC.2007.4370406
Current Version Published: 2007-10-29
Abstract
Traditional blocking artifacts reduction algorithm often causes image edges losing. To address this problem, a fuzzy edge-sensitivity blocking artifacts reduction algorithm is proposed. Based on some characteristics of human visual system, this algorithm applies adaptive filter to current pixel by integrating fuzzy logic technique. The experimental results show that this algorithm can obtain better results than traditional ones.
A Simple Algorithm for the Reduction of Blocking Artifacts in Images and Its Implementation (1998) [6 citations — 1 self]
http://ipl.univ.trieste.it/ipl/publications/itce98
by Roberto Castagno , Stefano Marsi , Gianni Ramponi
Abstract:
In this paper a simple but effective operator for the reduction of blocking artifacts is presented, together with a simple hardware implementation. The method is based on the Rational Filter approach: the operator is expressed as a ratio between a linear and a polynomial function of the input data. Such filters proved to outperform other conventional methods in other applications, such as noise smoothing [1], thanks to their capability of adapting gradually to the local image characteristics. The filter is capable of biasing its behaviour in order to achieve good performance both in uniform areas, where linear smoothing is needed, and in textured zones, where nonlinear and directional filtering is required. A detector of activity is embedded in the expression of the operator itself so that the biasing of the behaviour of the filter is smooth and not based on fixed thresholds. A solution for the hardware implementation of the scheme is presented in detail. Despite the simplifications im...
P.S. Ummm... I seem to have guessed right what it may be... They now have OpenCL and GC to build software filters being fast enough...
Yeah, because hey, what's better than patenting an algorithm?
Don't know?
Why hell, to announce to the world the friggin' algorithm!
Sorry, what do you mean?
Apple has investigated a method to make imperfections in compressed video files less visible to the human eye, a new patent application revealed this week shows.
What about cats' eyes? Cats like watching YT videos of other cats.
Apple needs an algorithm to mask awkward writing.
There are a number of licensed dither algorithms in the digital audio world. Aren't there any in the video world?
The differences in audio dithering algorithms are subtle, if even distinguishable by the untrained ear. Surely dithering algorithms exist for video. Could this be that much of a breakthrough? Sounds like incremental improvement at best (at first read).
Increased bandwidth is the real answer to quality video, IMO. Period.
gc
Also, what I wrote earlier about FLV not being up there with more advanced codecs, was me trying to write in a quick way that Flash simply isn't the answer, without just saying "Yeah, but Flash is rubbish."
I just hope that one day, in the far future, maybe there will be a 1,024-core mac that can play a youtube video over a broadband link without stuttering.
That has nothing to do with the processor but bandwidth speeds.