Comments on: Mode-Dependent Directional Transform (MDDT) in JM/KTA http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html Witness the development of H.265 Mon, 26 Jul 2010 04:00:42 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: hohai http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2474 hohai Wed, 21 Apr 2010 01:03:03 +0000 http://www.h265.net/?p=255#comment-2474 please,can you tell me how can we get basis images of MDDT .I have see the mddct i have found that there is something different .I want to know the principle of the algorithm. please,can you tell me how can we get basis images of MDDT .I have see the mddct i have found that there is something different .I want to know the principle of the algorithm.

]]> By: Yu Liu http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2432 Yu Liu Tue, 03 Nov 2009 13:34:08 +0000 http://www.h265.net/?p=255#comment-2432 Yes, if you just want to evaluate the subjective quality of reconstructed video, you have to use the KTA decoder to parse the KTA bitstream. Yes, if you just want to evaluate the subjective quality of reconstructed video, you have to use the KTA decoder to parse the KTA bitstream.

]]> By: Zhou Jin http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2431 Zhou Jin Tue, 03 Nov 2009 13:30:48 +0000 http://www.h265.net/?p=255#comment-2431 >>The KTA software includes a lot of additional coding tools that are not compatible with H.264 syntax, so >>you can’t use H.264 video player to view the KTA bitstream. Thank you. So basically if I want to evaluate the stream then have to do it after decoding? >>The KTA software includes a lot of additional coding tools that are not compatible with H.264 syntax, so >>you can’t use H.264 video player to view the KTA bitstream.
Thank you. So basically if I want to evaluate the stream then have to do it after decoding?

]]> By: Yu Liu http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2430 Yu Liu Tue, 03 Nov 2009 13:25:11 +0000 http://www.h265.net/?p=255#comment-2430 The KTA software includes a lot of additional coding tools that are not compatible with H.264 syntax, so you can't use H.264 video player to view the KTA bitstream. If you want the KTA software to produce a H.264 compatible bitstream, you should disable the KTA MACRO definitions in the "defines.h", i.e., "ADAPTIVE_FILTER" to "SIMPLIFIED_RDPIC_DECISION". The KTA software includes a lot of additional coding tools that are not compatible with H.264 syntax, so you can’t use H.264 video player to view the KTA bitstream. If you want the KTA software to produce a H.264 compatible bitstream, you should disable the KTA MACRO definitions in the “defines.h”, i.e., “ADAPTIVE_FILTER” to “SIMPLIFIED_RDPIC_DECISION”.

]]> By: Zhou Jin http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2429 Zhou Jin Tue, 03 Nov 2009 06:54:50 +0000 http://www.h265.net/?p=255#comment-2429 Please, Did anyone ever tested the KTA software and tried to view the .264 file it produced? I mean, I tried several versions but none of them produced a .264 file that can be viewed by players such like StreamEYE. Or did it need some specific players or something? Please, Did anyone ever tested the KTA software and tried to view the .264 file it produced? I mean, I tried several versions but none of them produced a .264 file that can be viewed by players such like StreamEYE. Or did it need some specific players or something?

]]> By: Yu Liu http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2408 Yu Liu Wed, 07 Oct 2009 03:36:35 +0000 http://www.h265.net/?p=255#comment-2408 The sequences for VCEG common test conditions are available at ftp://vceg-seq.ateme.net/reference/, but the ftp server is a password protected site. You are advised to contact the Rapporteur (Gary Sullivan from Microsoft? I guess) to get the password according to VCEG-AJ10r1. The sequences for VCEG common test conditions are available at ftp://vceg-seq.ateme.net/reference/, but the ftp server is a password protected site. You are advised to contact the Rapporteur (Gary Sullivan from Microsoft? I guess) to get the password according to VCEG-AJ10r1.

]]> By: Fatih http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2407 Fatih Tue, 06 Oct 2009 20:22:20 +0000 http://www.h265.net/?p=255#comment-2407 Can anyone tell me where I can download the HD test sequences used here such as BigShips City Crew Night ShuttleStart etc.. ? Can anyone tell me where I can download the HD test sequences used here such as BigShips City Crew Night ShuttleStart etc.. ?

]]> By: Sandip Ray http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2405 Sandip Ray Mon, 05 Oct 2009 09:11:59 +0000 http://www.h265.net/?p=255#comment-2405 Why are they not using closed-form sinusoidal or non-sinusoidal expression for directional transforms rather than using a training set for finding basis vectors ? If it is a closed form expression, I think we can easiliy implement fast algorithms using periodicity, symmetry and anti-symmetry properties. Why are they not using closed-form sinusoidal or non-sinusoidal expression for directional transforms rather than using a training set for finding basis vectors ? If it is a closed form expression, I think we can easiliy implement fast algorithms using periodicity, symmetry and anti-symmetry properties.

]]> By: Jie Dong http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2404 Jie Dong Mon, 05 Oct 2009 08:13:23 +0000 http://www.h265.net/?p=255#comment-2404 >>Is it eigen decomposition of the the average data or average of eigen decomposition of each and every data? I am not sure. I guess it should be the former, because the autocorrelation matrix, of which the eigenvectors construct the KLT matrix, shows the average property of a random field, e.g., a video source. >>Is the non-orthogonality arises because of rounding to integer ? I think so. Eigenvectors are always orthogonal to each other, which is not guaranteed after scaling and rounding. >>Is it eigen decomposition of the the average data or average of eigen decomposition of each and every data?
I am not sure. I guess it should be the former, because the autocorrelation matrix, of which the eigenvectors construct the KLT matrix, shows the average property of a random field, e.g., a video source.

>>Is the non-orthogonality arises because of rounding to integer ?
I think so. Eigenvectors are always orthogonal to each other, which is not guaranteed after scaling and rounding.

]]> By: Sandip Ray http://www.h265.net/2009/09/mode-dependent-directional-transform-mddt-in-jmkta.html/comment-page-1#comment-2403 Sandip Ray Mon, 05 Oct 2009 05:58:33 +0000 http://www.h265.net/?p=255#comment-2403 It is mentioned "The transform matrices are derived based on a large set of video sequences, which are all intra-coded." Is it eigen decomposition of the the average data or average of eigen decomposition of each and every data ? Is the non-orthogonality arises because of rounding to integer ? It is mentioned “The transform matrices are derived based on a large set of video sequences, which are all intra-coded.” Is it eigen decomposition of the the average data or average of eigen decomposition of each and every data ? Is the non-orthogonality arises because of rounding to integer ?

]]>