<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<title>high.definition.x264.standards.revision.3.1.addendum.1.nfo</title>
<style type="text/css">
@font-face {
font-family: nfo;
font-style: normal;
font-weight: normal;
src: url(nfo.eot);
}
.nfo {
padding: 12px;
font-family: nfo, courier new;
font-size: 11px;
line-height: 1em;
}
</style>
</head>
<body>
<pre class="nfo">��
High.Definition.x264.Standards.Revision.3.1.Addendum.1-HDX��
��
As there lately have been several pres with the wrong number of reference frames��
we thought we'd explain the reasoning behind the rule set.��
��
On various x264 pages on the internet the following can be read about reference frames:��
��
"Selects the maximum number of reference frames that can be used. Referenced frames��
are frames that refer to other frames (eg. if both frames are similar). Having a high ��
referenced frame will improve quality but slow up encoding. For typical content, ��
a reference frame of 3 to 5 is recommended. For content with a lot of repetition ��
(eg. animation), a reference frame of 8 to 10 can be used."��
��
So, higher reference frames means higher quality, why do we then enforce max 4 ��
reference frames on high resolution video? It has to do with hardware players. ��
The popcorn hour, twix, wd etc. all support Level 4.1 (L4.1) of the ITU-T h264 ��
specification. All graphic cards that have DXVA also support L4.1. So let's see ��
what L4.1 says about reference frames.��
��
In table 'A-1 Level limits' on page 283 (pdf 305) of the ITU-T specification it ��
says that MaxDPB for L4.1 is 12288 KiB. MaxDPB is the Maximum Decoded Picture Buffer, ��
which is the largest size allowed of the decoded picture buffer when decoding a video.��
By supporting L4.1, the hardware players must have at least 12 MiB of buffer for ��
storing the decoded pictures. This means that a video that requires a buffer of ��
13 MiB is not guaranteed to work on one of these players.��
��
As 16 * 16 pixels macroblocks are used, all resolutions needs to be mod16, for ease of��
reading, the maths to make them mod16 is not included below.��
��
The DPB in KiB is calculated as follows:��
��
DPB = vertical resolution * horizontal resolution * 1.5 * reference frames / 1024��
��
If we transform this formula to get the reference frames instead we get:��
��
ref = 12288 * 1024 / (vertical resolution * horizontal resolution * 1.5)��
��
We of course can't use partial frames for referencing and thus the reference frames ��
should be rounded down to the closest integer. We can also transform this to get the ��
maximum vertical resolution for a specific reference frames value, here we need to��
round the vertical resolution down to mod16:��
��
vertical res = 12288 * 1024 / (horizontal resolution * 1.5 * reference frames)��
��
With the above formula we can conclude that the highest vertical resolution that we ��
can have ref 5 on and still be L4.1 compliant is 864 pixels.��
��
873.813333 = 12288 * 1024 / ( 1920 * 1.5 * 5 )��
864 = floor( 873.813333 / 16 ) * 16��
��
The 1.5 in the calculations above is the YV12 colourspace, it needs 12 bits to store ��
1 pixel. In other words, 1.5 bytes per pixel.��
��
So, to conclude this, the reason we put ref 4 as max for movies with vertical resolution��
greater than 864 in rules is not because we want to be able to encode releases faster. ��
It's because we want releases to be L4.1 compliant and thus possible to play on the ��
popcorn hour, twix and other hardware players. And we require at least ref 5 on all ��
videos where it's possible while still respecting L4.1, this to ensure high quality.��
��
ITU-T specification:��
http://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-REC-H.264-200711-I!!PDF-E&type=items��
��
12 bits per pixel for YV12:��
http://msdn.microsoft.com/en-us/library/aa904813.aspx#yuvformats_420formats_12bitsperpixel</pre>
</body>
</html>