View Full Version : What is Deblocking and B-Frame
Alex_080
6th October 2009, 17:28
Can any one tell what is Deblocking and what value i use for normal video encoding mean -2 or -1:confused:
And tell me about B-frame setting .....what happen if i set b-frame to maximum 16 or minimum 3 please tell me:confused:
thewebchat
7th October 2009, 02:10
You have to determine the values for --deblock (I am assuming we are discussing x264) by visual analysis of the output. It may be optimal to use lower values to preserve more detail (e.g. -1:-1), but the loop filter is already adaptive by QP, so higher bitrate videos automatically get deblocked less. For b-frames, higher maximum consecutive b-frames increases coding efficiency and therefore quality. If you use --b-adapt 1 (default), then there is no reason to use anything but --b-frames 16 because the speed penalty is almost 0 for this. If you use --b-adapt 2 (better than --b-adapt 1), then the speed decrease as you increase --bframes increases linearly and something like 3-6 is a reasonable value.
Some more background on this: H.264 specifies a in-loop deblocking filter that processes decoded frames before they are displayed and used for motion compensation. This means that the encoder can spend less bits on blocks that it knows the loop filter will smooth out, allowing for some compression gain. This post (http://forum.doom9.org/showpost.php?p=810932&postcount=1) by MP4 Guy explains what the numbers mean. B-frames are a type of predictive frame that uses information from two different frames to predict the current frame. This page (http://bmrc.berkeley.edu/research/mpeg/mpeg_overview.html) explains how I/P/B frames work in MPEG-like formats.
Alex_080
7th October 2009, 07:59
You have to determine the values for --deblock (I am assuming we are discussing x264) by visual analysis of the output. It may be optimal to use lower values to preserve more detail (e.g. -1:-1), but the loop filter is already adaptive by QP, so higher bitrate videos automatically get deblocked less. For b-frames, higher maximum consecutive b-frames increases coding efficiency and therefore quality. If you use --b-adapt 1 (default), then there is no reason to use anything but --b-frames 16 because the speed penalty is almost 0 for this. If you use --b-adapt 2 (better than --b-adapt 1), then the speed decrease as you increase --bframes increases linearly and something like 3-6 is a reasonable value.
Some more background on this: H.264 specifies a in-loop deblocking filter that processes decoded frames before they are displayed and used for motion compensation. This means that the encoder can spend less bits on blocks that it knows the loop filter will smooth out, allowing for some compression gain. This post (http://forum.doom9.org/showpost.php?p=810932&postcount=1) by MP4 Guy explains what the numbers mean. B-frames are a type of predictive frame that uses information from two different frames to predict the current frame. This page (http://bmrc.berkeley.edu/research/mpeg/mpeg_overview.html) explains how I/P/B frames work in MPEG-like formats.
thanks very much
:thanks:
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.