<div>Hello Gabriel,</div>
<div> yes its not H.264 per se.</div>
<div>But if i encode a 32x32 size MB with 8x8 transform size, then quantize it, Huffman encode it, in one case, and with 32x32 transform size, quantize it and Huffman encode it in second case;</div>
<div>Will the quantization values in both these cases be different as the DCT coefficients obtained after these two differing block size transforms will be different? I believe they should be different to be able to recover same/similar pixels at decoder after Huffman decode, IQ, and IDCT? Am i correct in saying ?</div>
<div> </div>
<div>Thanks,</div>
<div>-Ajit</div>
<div> </div>
<div><br clear="all"><br>Rita Rudner - "Before I met my husband, I'd never fallen in love. I'd stepped in it a few times." <br><br></div>
<div class="gmail_quote">On Tue, Mar 17, 2009 at 4:30 PM, <span dir="ltr"><<a href="mailto:x264-devel-request@videolan.org">x264-devel-request@videolan.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">Send x264-devel mailing list submissions to<br> <a href="mailto:x264-devel@videolan.org">x264-devel@videolan.org</a><br>
<br>To subscribe or unsubscribe via the World Wide Web, visit<br> <a href="http://mailman.videolan.org/listinfo/x264-devel" target="_blank">http://mailman.videolan.org/listinfo/x264-devel</a><br>or, via email, send a message with subject or body 'help' to<br>
<a href="mailto:x264-devel-request@videolan.org">x264-devel-request@videolan.org</a><br><br>You can reach the person managing the list at<br> <a href="mailto:x264-devel-owner@videolan.org">x264-devel-owner@videolan.org</a><br>
<br>When replying, please edit your Subject line so it is more specific<br>than "Re: Contents of x264-devel digest..."<br><br><br>Today's Topics:<br><br> 1. Block DCT query (Ajit Deshpande)<br> 2. Re: Block DCT query (Gabriel Bouvigne)<br>
<br><br>----------------------------------------------------------------------<br><br>Message: 1<br>Date: Tue, 17 Mar 2009 10:59:06 +0530<br>From: Ajit Deshpande <<a href="mailto:ajitsdeshpande@gmail.com">ajitsdeshpande@gmail.com</a>><br>
Subject: [x264-devel] Block DCT query<br>To: <a href="mailto:x264-devel@videolan.org">x264-devel@videolan.org</a><br>Message-ID:<br> <<a href="mailto:c2d46aba0903162229v659d7a03ob31798fbdee1ca75@mail.gmail.com">c2d46aba0903162229v659d7a03ob31798fbdee1ca75@mail.gmail.com</a>><br>
Content-Type: text/plain; charset="iso-8859-1"<br><br>Hello,<br> I am doing some basic Matlab simulations on DCT used in Video<br>Compression. I have following questions arised out of that:<br><br>1.) In a 16x16 MB , if one does a 8x8 size block DCT (4-times) for that MB<br>
in first case and a one shot 16x16 size block DCT in second case, Will the<br>reconstructed block at the decoder will have same/similar values in both<br>cases?<br><br>2.) Which case would there be any benefit in terms of lesser residue, lesser<br>
bits needed for coding etc?<br><br>3.) Will this analysis of block DCT also hold true for other transform like<br>the integer transform used in H.264 codec?<br><br>Regards<br>-Ajit<br><br>Douglas Adams - "I love deadlines. I like the whooshing sound they make as<br>
they fly by."<br>-------------- next part --------------<br>An HTML attachment was scrubbed...<br>URL: <<a href="http://mailman.videolan.org/pipermail/x264-devel/attachments/20090317/7fd3b4b1/attachment-0001.htm" target="_blank">http://mailman.videolan.org/pipermail/x264-devel/attachments/20090317/7fd3b4b1/attachment-0001.htm</a>><br>
<br>------------------------------<br><br>Message: 2<br>Date: Tue, 17 Mar 2009 10:41:01 +0100<br>From: Gabriel Bouvigne <<a href="mailto:gabriel.bouvigne@joost.com">gabriel.bouvigne@joost.com</a>><br>Subject: Re: [x264-devel] Block DCT query<br>
To: Mailing list for x264 developers <<a href="mailto:x264-devel@videolan.org">x264-devel@videolan.org</a>><br>Message-ID: <<a href="mailto:49BF702D.1030102@joost.com">49BF702D.1030102@joost.com</a>><br>Content-Type: text/plain; charset=ISO-8859-1; format=flowed<br>
<br>Ajit Deshpande a ?crit :<br>> Hello,<br>> I am doing some basic Matlab simulations on DCT used in Video<br>> Compression. I have following questions arised out of that:<br><br>This is not really related to x264, is it?<br>
<br>> 1.) In a 16x16 MB , if one does a 8x8 size block DCT (4-times) for that<br>> MB in first case and a one shot 16x16 size block DCT in second case,<br>> Will the reconstructed block at the decoder will have same/similar<br>
> values in both cases?<br><br>An integer dct is lossless, while a floating point dct is usually lossy<br>because of roundings.<br>Now, if you are just using an iDct, there will be no loss, and thus<br>reconstructed will be identical UNLESS you are quantizing the dct<br>
results. (in most perceptual compression schemes, it is not the<br>transform which is lossy but the quantization of the results of the<br>transform)<br><br>Btw, there is no 16x16 transform within h.264.<br><br><br>> 2.) Which case would there be any benefit in terms of lesser residue,<br>
> lesser bits needed for coding etc?<br><br>Sometimes one would be better, sometimes the other one would be better.<br>For "flat" residues, a bigger transform is usually better, while for<br>"transcients" a smaller transform is usually better.<br>
<br>> 3.) Will this analysis of block DCT also hold true for other transform<br>> like the integer transform used in H.264 codec?<br><br>Likely<br><br>--<br>Gabriel<br><br><br>------------------------------<br><br>
_______________________________________________<br>x264-devel mailing list<br><a href="mailto:x264-devel@videolan.org">x264-devel@videolan.org</a><br><a href="http://mailman.videolan.org/listinfo/x264-devel" target="_blank">http://mailman.videolan.org/listinfo/x264-devel</a><br>
<br><br>End of x264-devel Digest, Vol 22, Issue 17<br>******************************************<br></blockquote></div><br>