Encoding Video for Blu-Ray using H264/AVC

[shon3i]

Here is part of specification that apply when encoding with H264/AVC codec, all parameters showed here are not guessing, it's based on original Blu-Ray specification, which are not available in public.

NOTE: This is discussion about settings that are for creating 100% compliant Blu-Ray structures, not for encoding from Blu-Ray sources to matroska/mp4 or other media, if you encode to those skip this nightmare

1. GENERAL CONSTRAINTS FOR MPEG-4 H264/AVC

1.1 ALLOWED RESOLUTIONS/FRAMERATES

tom.niko.users.sbb.rs/resolutions.png


1.2 ALLOWED LEVELS, PROFILES, REFERENCE FRAMES, VBV

1.2.1 Primary Video Rules

tom.niko.users.sbb.rs/gopsrefs2.png


1.2.2 Secondary Video Rules

NOTE: In case of 1080 and 720 resolutions same rules apply as Primary video, while 576 and 480 resolutions have stricter parameters showed in table:

tom.niko.users.sbb.rs/gopsrefs3.png



1.3 SIMPLE ASPECT RATIO

tom.niko.users.sbb.rs/sar.png



1.4 COLOR CHARACTERISTICS

tom.niko.users.sbb.rs/color.png



1.5 OTHER REQUIRED PARAMETERS

8-bit (4:2:0) Colorspace is only allowed
Main and High profiles are allowed
Maximum 3 B-Frames is allowed
NAL-HRD Type 2 capable encoder is needed


2. ENCODE BLU-RAY VIDEO USING x264

2.1 MANDATORY PARAMETERS

Every encode for Blu-Ray must contain this parameters, without it will not meet Blu-Ray specification.

--bluray-compat

Enforce x264 to create BD compliant stream, that will reduce x264 settings to BD compatible: bframe<=3, ref<=4 for 1080, ref<=6 for 720/576/480, bpyramid<=strict, weightp<=1, aud=1, nalhrd=vbr

--level

Blu-Ray level, see the table for the appropriate level

--keyint

Maximum number displayed pictures in GOP. If you set Maximum bitrate less or equal to 15mbps you can use 2 second long GOP otherwise, 1 second GOP must be used.

--vbv-maxrate

Maximum allowed bitrate. Depend of destination media, for normal BD Media, maximum settings is allowed (40000), while for DVD Media (eg BD-5 or BD-9) there is no maximum per Blu-Ray specification, but is strongly recommended that use less or equal to 15000, since DVD reading is slower.

--vbv-bufsize

Maximum allowed buffer, buffer size is in relation with maxrate. Since STD Buffer Delay for Blu-Ray allows maximum 1 second. This mean to divide buffer with maxrate and see if result is less or equal to 1, if not, buffer is need to be reduced to lower value that meet 1 second STD. For BD media, maximum settings is allowed (30000), but if not meet STD Buffer Delay, then need to be reduced to lower value. For DVD media there is no maximum in Blu-Ray specification, but is strongly recommended that use less or equal to 15000, and again need to meet STD delay of 1 second.

For example:

--vbv-bufsize 30000 --vbv-maxrate 40000 is allowed
--vbv-bufsize 30000 --vbv-maxrate 30000 is allowed
--vbv-bufsize 15000 --vbv-maxrate 30000 is allowed
--vbv-bufsize 30000 --vbv-maxrate 15000 is not allowed (delay is greater than 1 second)

Basicly vbv-bufsize should newer been greater than vbv-maxrate, that is whole point.

--sar

Simple Aspect Ratio, must present in stream, see the table for the appropriate settings

--slices

Only when encoding with --level 4.1, slices need to be set to 4 or greater, otherwise you can skip this setting or set it to 1.

-o xxx.264

Output file must be in raw 264 elementary stream (extension .264) otherwise settings will not applied correctly. DO NOT USE MKV OR MP4.

2.2 OPTIONAL PARAMETERS

This parameters is optional and not need for every encode.

--colorprim
--transfer
--colormatrix

Specify color primaries, transfer characteristics and color matrix. This is optional, and not need to be present, see the table for the appropriate settings.

--pic-struct

Need only when fake interlaced is used, and it's default on with interlaced and pulldown. It's fine with progressive aslo, and some BD verification tools request to be present in stream.

--open-gop

Use non-IDR pictures instead classic IDR picture, this will tend to resolve pulsing picture problem that is usually came with short GOP's.

--fake-interlaced

Since 25p or 30p video is not allowed for Primary Video, encoding progressive video that meet Blu-Ray specification is only possible with this option. If use this option you aslo need to specify --pic-struct

--qpfile

Need for chapters.

2.2 EXAMPLES

Progressive

1080p
720p
576p
480p

Interlaced

1080i
576i
480i

[b66pak]

thanks a lot...
_

[Guest]

Please advise when it is complete so we can make it a stickie. Thank you.

[AnonCrow]

Little nitpick : since you're using images and you're hosting them yourself, might want to optimize them for smaller size - pics 1 and 2 went from 50 KiB and 62 KiB to 13 and 16 KiB, respectively , just by decreasing bit-depth and using maximum PNG compression. This without even resorting to PNGOUT or Pngcrush.

[ajp_anton]

What does "Max bitrate BD, DVD and with or without 2 sec GOP" mean?

And to do 25fps progressive, you have to either double the framerate to 50fps, or encode it as 25fps interlaced?

[creamyhorror]

Thanks for doing this. I think a specific x264 section with recommended commandlines and workflow notes would be essential.

[shon3i]

@AnonCrow, thanks for tip. i probably will upload later here as attach, but i will optimize as much as possible.

Originally Posted by ajp_anton

What does "Max bitrate BD, DVD and with or without 2 sec GOP" mean?

It mean, if you target is BD media and if you set VBV maxrate <=15000 then you can use --keyint from Max Frames in 2 sec. GOP. Or if you target media is DVD (BD5/BD9) you aslo may not use higher maxrate/buffer than 15000, coz DVD spin speed is lower, that is some error free maximum for DVD. Still if you set maxrate to 15000 you can choose what you want to use 1 second or 2 second GOP, for BD or DVD.

Originally Posted by ajp_anton

And to do 25fps progressive, you have to either double the framerate to 50fps, or encode it as 25fps interlaced?

You need encode it as interlaced, or convert to 720p50.

Originally Posted by creamyhorror

Thanks for doing this. I think a specific x264 section with recommended commandlines and workflow notes would be essential.

To be continued with explanation, examples, and few more tables.

[Snowknight26]

Originally Posted by shon3i

You need encode it as interlaced, or convert to 720p50.

Doesn't actually have to be interlaced, can just be flagged as interlaced yet contain progressive frames.

[kieranrk]

Originally Posted by Snowknight26

Doesn't actually have to be interlaced, can just be flagged as interlaced yet contain progressive frames.

If you had an encoder that can do adaptive PAFF/MBAFF then yes.

Or you could force it manually since the spec says frame_mbs_only shall be 0.

[bob0r]

May i advice, that instead of an image, maybe make a .html page or something silimar.
So in case of link down, or user absense, people can easy update or repost when needed.

[kypec]

Originally Posted by bob0r

May i advice, that instead of an image, maybe make a .html page or something silimar.
So in case of link down, or user absense, people can easy update or repost when needed.

+++ Images are nice but hardly maintainable nor usable (copy/paste) and so on.

[kolak]

I'm preaty sure that 25p is allowed for secondary video stream.


Andrew

[mp3dom]

Infact it's mentioned in the tab
I would like to add that even if it's allowed to have a fullHD secondary video, it's better to prefer SD secondary video because not all the standalone players are able to handle 2 fullhd stream at once.

[shh]

Originally Posted by Snowknight26

Doesn't actually have to be interlaced, can just be flagged as interlaced yet contain progressive frames.

Originally Posted by kieranrk

If you had an encoder that can do "full" PAFF/MBAFF then yes.

Is x264 able to do this already? (= flagging the stream interlaced, but just coding progressive frames)
I've got many progressive PAL-sources (most satellite-streams are p anyway), and wouldn't like to encode that "interlaced".

[kieranrk]

Originally Posted by shh

Is x264 able to do this already? (= flagging the stream interlaced, but just coding progressive frames)
I've got many progressive PAL-sources (most satellite-streams are p anyway), and wouldn't like to encode that "interlaced".

You could modify x264 to write frame_mbs_only flag as zero. You'd have to write pic_struct too but the spec isn't clear whether it wants you to signal the frame as two fields or as one frame.

[ajp_anton]

Originally Posted by shon3i

It mean, if you target is BD media and if you set VBV maxrate <=15000 then you can use --keyint from Max Frames in 2 sec. GOP. Or if you target media is DVD (BD5/BD9) you aslo may not use higher maxrate/buffer than 15000, coz DVD spin speed is lower, that is some error free maximum for DVD. Still if you set maxrate to 15000 you can choose what you want to use 1 second or 2 second GOP, for BD or DVD.

But the "with or without 2 sec GOP" means it is 15000 *always*, whether or not you use 2 sec GOP, and this is false, because if you use 1 sec GOP with BD, it's 40000 and not 15000.
Sure 15000 can be used, but it's not max, which is what this tries to tell you.

IMO it should say something like

Max bitrate
2 sec GOP BD
DVD

[rack04]

Thanks for taking the time to put this together. Here are my comments:

1.) "RESOLITIONS" should be spelled "RESOLUTIONS"
2.) Add a section that includes mandatory items like --aud, --bframes 3, --nal-hrd vbr, etc. I know you mention bframes 3 in the title section of Section 2 but that could easily be overlooked by people who concentrate on big pretty pictures.

[Sharc]

Thanks a lot shon3i for your initiative

Originally Posted by creamyhorror

Thanks for doing this. I think a specific x264 section with recommended commandlines and workflow notes would be essential.

+++ Would indeed be very helpful.
Could it eventually be included in x264 as 'device presets' or similar?

[poisondeathray]

Nice work sho3ni!

Originally Posted by ajp_anton

And to do 25fps progressive, you have to either double the framerate to 50fps, or encode it as 25fps interlaced?

And another option would be to do the slowdown to 24p or 23.976p if you can tolerate the audio shift

[shon3i]

Originally Posted by ajp_anton

But the "with or without 2 sec GOP" means it is 15000 *always*, whether or not you use 2 sec GOP, and this is false, because if you use 1 sec GOP with BD, it's 40000 and not 15000.
Sure 15000 can be used, but it's not max, which is what this tries to tell you.

IMO it should say something like

Max bitrate
2 sec GOP BD
DVD

Not exactly because situation looks like this:

1. 15000 is max for DVD
2. 15000 is max i you want use 2 sec GOP besides you use DVD or BD
3. you can still use 1 sec gop.

so if you don't use 2 sec GOP that mean you will use 1 second instead. I will somehow fix this.

@all, be patient