What codecs/container do you save as your primary?

[HawkXP71 99]

When you RIP from a DVD/BluRay using MakeMKV do you then transcode it to another container? What about the video codecs/audio codecs?  Do you transcode to H265 lossless? compress it a bit?  What about audio?

What about if you get a h264, or some other codec in video? do you convert it?

Do you only do this "when" its created? or have you gone through and updated older codecs to this one for media you have had for years?

For me, I use matroska containers exclusively, I transcode everything to H265 using a lossless transcode, and I leave the audio alone if its eac3 or aac.  If its one of the weirder audio codecs ill see if Its direct streamed to my webbrowser as well as my TV.  Shield does everything, so no point in testing  but my kids who watch remotely dont have shields.

TIA. just wondering what people use

 

[Luke 37068]

I think mkv with h264 is still the best universal option, for now at least. It probably won't be too long before that changes though.

[HawkXP71 99]

17 minutes ago, Luke said:

I think mkv with h264 is still the best universal option, for now at least. It probably won't be too long before that changes though.

Which audio codec? 

[ebr 14913]

9 hours ago, HawkXP71 said:

Which audio codec? 

AAC 2.0 would be the most compatible but I think pretty much everything supports AC3 (DD) now too.

[sfatula 185]

I'm transcoding them. I encode to H.265, certainly not lossless. I use MP4 container, and, audio depends on the source audio. Either DD 5.1 (ac3) or EAC3 for 7 channel. 

[HawkXP71 99]

22 hours ago, sfatula said:

I'm transcoding them. I encode to H.265, certainly not lossless. I use MP4 container, and, audio depends on the source audio. Either DD 5.1 (ac3) or EAC3 for 7 channel. 

The reason I use lossless, is its a transcode from something that has already lost the data.  So Im trying to maximize whatever data I had

[tedfroop 33]

Also ripping to h265 with Handbrake.  H265 and pass through and a single audio stream - whatever highest quality is DD or DTS for video.  I do use MKV files to rip from disk occasionally or so I can join multi Disk movies with  MKVToolnix.

Audio is all FLAC lossless.  Use DBPoweramp as you can rip to multiple libraries at the same time (So I could rip to alac for my ipod and Flac) plus you can add things like Replay gain all at the same time.  Plus it uses accurate rip so you can tell your rip is accurate and tagging support is superb.

[Jammi3D 8]

I use mkv container for all my files. Have a few FFMPEG scripts setup so just a matter of changing CRF values to suite the file that is being encoded.

SD is encoded to 540p H264 with AAC audio @ 192kbps (2.0) (DVB 1080p source)

All UK drama's, Favourite Movies & Docs are encoded to 1080p H265 with AC3 audio (passthrough for DVB captures otherwise 384kbps for Movies)

For the devices that connect to my server, basic firesticks ect all direct play with no issues.
 

[Deathsquirrel 741]

I used to reencode things to save space but it was way, WAY too much effort to get decent results consistently.  Drives are cheap so I switched to full rips.

I did make some effort to reencode VC1 to H264 though, just to cut down on transcoding.  If I keep the bit rate identical to the source the output is consistently fine.

[sfatula 185]

16 minutes ago, Deathsquirrel said:

I used to reencode things to save space but it was way, WAY too much effort to get decent results consistently.  Drives are cheap so I switched to full rips.

I did make some effort to reencode VC1 to H264 though, just to cut down on transcoding.  If I keep the bit rate identical to the source the output is consistently fine.

If you automate things, can be as simple as move file to a folder, come back in morning, encoded. So, 10 seconds of time or so. Saving space is only one reason as you have noted. 

[tedfroop 33]

MP4, M4V, wrapper for H265 encoded video, with the best quality available audio stream pass through.  I recently upgraded my TV and it allows pretty much any audio stream to pass through.

I have MakeMKV, Handbrake, and MKVToolnix.  MakeMKV I generally use for "cranky" disks or for multi - disk movies, (rip, join with MKVToolnix)

[HawkXP71 99]

On 3/31/2023 at 9:21 PM, Deathsquirrel said:

I used to reencode things to save space but it was way, WAY too much effort to get decent results consistently.  Drives are cheap so I switched to full rips.

I did make some effort to reencode VC1 to H264 though, just to cut down on transcoding.  If I keep the bit rate identical to the source the output is consistently fine.

I have written an app (Media Manager) that will scan your folder, and give a list of media that needs transcoding (depending on your settings) then run ffmpeg to do the transcoder.

The default is H265 + adding a 5.1 AAC as the default audio channel

 

[sfatula 185]

Yeah, tools I may use on Video includes Handbrake, ffmpeg, mkvtoolnix, Subtitle Editor, Audacity, TS Muxer, MediaInfo, makemkv, bd3d2mk3d, bdsup2sub++, dolby media encoder, adobe audition, vobsub2srt. Just depends what I am doing (like saving my 3d blurays to Emby take a number of these), for most things, likely just an encode. I decided long ago any idea of "losing quality" was all in my head, could never tell side by side. 

I used to use flac, but, wanted instead to use Apple standard formats as that's what I use. Thus, nothing seems to ever transcode (whether in player or via emby). But I don't have any 4k stuff yet (maybe 1). 

Edited April 4, 2023 by sfatula

[Deathsquirrel 741]

On 3/31/2023 at 9:39 PM, sfatula said:

If you automate things, can be as simple as move file to a folder, come back in morning, encoded. So, 10 seconds of time or so. Saving space is only one reason as you have noted. 

You can't automate things with quality output at a guaranteed space savings.  Source material is too varied.

Just not worth the effort IMO.

[sfatula 185]

2 hours ago, Deathsquirrel said:

You can't automate things with quality output at a guaranteed space savings.  Source material is too varied.

Just not worth the effort IMO.

Not very hard actually. I have done exactly that. But if it doesn't work for you, can't argue with that either! Space savings is the easiest part.

[HawkXP71 99]

On 4/14/2023 at 6:05 PM, Deathsquirrel said:

You can't automate things with quality output at a guaranteed space savings.  Source material is too varied.

Just not worth the effort IMO.

While I certainly agree, how is it any different than the "default transcoding" options being used by emby?

If your client needs transcoding, either audio or video, emby is using a CRF of 23 with an "auto" preset which I'm guessing is based on the bit rate needed on the client (if transcoding is necessary due to bandwidth limitations).

My thoughts on this, were not for disk space savings, but GPU resource savings trying to remove transcoding as a need. 

My goal is to write a script to analyze WHY transcoding is occurring (see 

 ) for a my questions on that front.

Then trying to make sure the media on my server covers most of the reasons before hand.

For instance, in analyzing my data, the #1 reason for transcoding, was the primary audio was a 7.1 format and needed to be transcoded to a 5.1 for browser viewing.

I also like to keep the media meta info tags accurate   so MKV containers are a must (changing metainfo is trivial/fast with MKV, for MP4 it takes a ton of time)

So my current process

• All containers converted to MKV
• Using a CRF of 21 transcode the video to h265
• Make sure the "first/default" audio track is 5.1 or stereo
  • Including the addition of a new track if necessary via downmixing
• Use CRF MUX (merge) all SRT/SUB+IDX files into a single file

Since ive started doing this, I have seen much fewer transcoding requests.   

If I notice the video quality isnt where I like, Ill re-transcode it (I keep the old version around until I am happy...)

 

 

Edited April 19, 2023 by HawkXP71

[tedfroop 33]

5 hours ago, HawkXP71 said:

Since ive started doing this, I have seen much fewer transcoding requests.   

A lot of that depends on your playback device does it not?

Since its so easy I usually burn in subtitles on a second copy of titles I want subtitles for, which I think totals about two files.

Just started trying out webm (Av1) to rip some files.  Did a bunch of Looney Tunes cartoons.  Had a few quality issues on a few files but for the most part very encouraging results.

[HawkXP71 99]

33 minutes ago, tedfroop said:

A lot of that depends on your playback device does it not?

Since its so easy I usually burn in subtitles on a second copy of titles I want subtitles for, which I think totals about two files.

Just started trying out webm (Av1) to rip some files.  Did a bunch of Looney Tunes cartoons.  Had a few quality issues on a few files but for the most part very encouraging results.

Of course.  If every client was a nvidia shield I wouldn't do anything.

But most are TVs or web clients.  Which have limited decoders.

I don't burn in subtitles, just add them as a strean

[tedfroop 33]

Just for anyone trying Av1, if you use Windows use the WebM wrapper.  Windows has some playback issues if you use other wrappers.

[rbjtech 4266]

I'm on this journey as well.

My core requirement is to convert all h264 over a certain bitrate for it's given resolution to hevc using QSVEncC with ICQ.   Space savings on original h264 Remux's are huge (sometimes 20% of the original size) - and I cannot visually tell the difference on what I consider to be a reasonable AV setup.   

But as has been mentioned above, a blanket 'convert it all with setting x' is totally the wrong way to approach this - you must analyse each file and encode accordingly but even then, it may have been encoded that high because it has action scenes from beginning to end, film grain, huge levels of detail or a very dark movie for example - a quick 'ffprobe' can't tell you those sort of details.

That being said, I'm trying to use 8Mbit/sec for h264 1080p material as my threshold - if it's less than this, then it stays on h264 - as it does not have the original source quality to be converted again.

I looked at AV1, but for my use case, I don't believe it has any significant advantages over hevc - and it has many disadvantages as my client base is only partially able to hardware decode AV1.

Edited April 20, 2023 by rbjtech

[Spaceboy 2493]

from my experience i agree it is not possible to have an automated profile do this. to be honest i have never found any conversion profile that i was happy with the outcome from, i can always see the difference and its not acceptable to me (but remain open to being convinced).

the size of my library and my time available also mean i cannot take a individual analytic approach to conversion either so i just keep throwing hard disk space at it. about half a petabyte and counting

[HawkXP71 99]

10 hours ago, rbjtech said:

I'm on this journey as well.

My core requirement is to convert all h264 over a certain bitrate for it's given resolution to hevc using QSVEncC with ICQ.   Space savings on original h264 Remux's are huge (sometimes 20% of the original size) - and I cannot visually tell the difference on what I consider to be a reasonable AV setup.   

But as has been mentioned above, a blanket 'convert it all with setting x' is totally the wrong way to approach this - you must analyse each file and encode accordingly but even then, it may have been encoded that high because it has action scenes from beginning to end, film grain, huge levels of detail or a very dark movie for example - a quick 'ffprobe' can't tell you those sort of details.

That being said, I'm trying to use 8Mbit/sec for h264 1080p material as my threshold - if it's less than this, then it stays on h264 - as it does not have the original source quality to be converted again.

I looked at AV1, but for my use case, I don't believe it has any significant advantages over hevc - and it has many disadvantages as my client base is only partially able to hardware decode AV1.

Ive been holding off for the next major release of ffmpeg to play with av1, as the non-hardware encoding is just too slow compared to hevc.

 

[HawkXP71 99]

7 hours ago, Spaceboy said:

from my experience i agree it is not possible to have an automated profile do this. to be honest i have never found any conversion profile that i was happy with the outcome from, i can always see the difference and its not acceptable to me (but remain open to being convinced).

the size of my library and my time available also mean i cannot take a individual analytic approach to conversion either so i just keep throwing hard disk space at it. about half a petabyte and counting

Im less worried about harddrive space, then I am transcoding performance.  but 500tb?? dang my wife thinks Im crazy at 70tb

[rbjtech 4266]

11 hours ago, HawkXP71 said:

Ive been holding off for the next major release of ffmpeg to play with av1, as the non-hardware encoding is just too slow compared to hevc.

 

The ffmpeg native encoders are not actually that great - x265, QSVEncC etc are dedicated encoders, so are better suited to the task.   Both x265 and QSVEncC both support AV1 Encoding in hardware if you have the discreet card to do it (ARC etc).   Maybe have a look at FastFlix (python based) - currently my goto GUI for combining these tools - and it gives you the command line, so you can tweak as necessary.  It supports hardware based QSV, NVENC and AMD encoding.

[rbjtech 4266]

12 hours ago, HawkXP71 said:

Ive been holding off for the next major release of ffmpeg to play with av1, as the non-hardware encoding is just too slow compared to hevc.

 

My light bulb moment was doing the analysis that 98% of my 'media' was sitting there 'idle' while consuming power and sitting on relatively expensive disk.   Lets be real here, a 16Tb disk is not what I would call 'inexpensive' .. and when you add them all up + backup's - it's a lot of money..

Storage 'Tiering' is something I'm also looking at - and for 'archive' media - moving it to HDD powered down arrays on older lower capacity disks, leaving the newer high density drives (but much fewer of them) spinning for stuff that is < 1 year old or 'continuing  in Emby' and SSD for the 'current' media (< 3 Months old).  Maybe unrelated to the thread - but in the debate about how the heck we store all this media - I think it's another area that warrants some thought ?