Emby best practices and suggestions?

[ecrispy 38]

I'm planning to reinstall my media server, redo my libraries etc and want to make sure I do it right. Esp for large libraries.

hardware

• server hw - currently old Dell pc, bought a sff mini pc with much lower power draw for new server.
• server sw - prev Windows, new one will be Linux (Debian 12) in docker
• client hw - FireStick4k, and unused Odroid N2
• client sw - from what I've read Kodi has best skins and decode, so want to use Emby next gen. Also the N2 will only work with CoreElec and thus has no Emby client
• I have an Onkyo AVR with 7.1 HT system thats pretty decent

questions :-

1. hevc - I want to convert a lot of my media to x265, either transcode or redownload, to save space. Is there a way to handle aac 5.1/he-aac 7.1, or should they be converted to ac3?
2. this thread isn't to discuss quality as it depends on source, but given my clients, they should all play hevc 10 bit. I know transcode h264->hevc will lose quality but a lot of people have said difference isn't noticeable even with nice tv/audio? has anyone else done this?
3. naming - do people usually add [imdbid] [tmmdb] [quality] [codec] etc to the file name? I've never done this but there may be some benefits?
4. metadata - I've read that nfo/artwork/metadata should NOT be stored next to media for better performance? i.e. the metadata dir can be separate on ssd, and thus media drives don't need to be accessed to browse. I've also stored next to media as its the most compatible. is there a way to do both and keep in sync?
5. video seek previews - planning to enable .bif generation for all media this time. is it supported on all clients? I'm guessing it won't work with Kodi though?

any other suggestions/ideas?

 

Edited August 12, 2023 by ecrispy

[rbjtech 5284]

6 hours ago, ecrispy said:

1. hevc - I want to convert a lot of my media to x265, either transcode or redownload, to save space. Is there a way to handle aac 5.1/he-aac 7.1, or should they be converted to ac3?

AC3 is more portable than aac5.1 and definately more portable than aac7.1 - why not use eac3 5.1 ?

6 hours ago, ecrispy said:

1. this thread isn't to discuss quality as it depends on source, but given my clients, they should all play hevc 10 bit. I know transcode h264->hevc will lose quality but a lot of people have said difference isn't noticeable even with nice tv/audio? has anyone else done this?

Head over to doom9 - a lifetime of discussion on there.  But in summary - if YOU are happy with the results, then it does not realy matter what is 'best' ..

6 hours ago, ecrispy said:

1. naming - do people usually add [imdbid] [tmmdb] [quality] [codec] etc to the file name? I've never done this but there may be some benefits?

Yes - try and do this.  Using Media Managers such as Sonarr/Radarr can automatically rename files for you to include all the codecs and provider id in any format you choose.   Emby will use this provider id if available.

6 hours ago, ecrispy said:

1. metadata - I've read that nfo/artwork/metadata should NOT be stored next to media for better performance? i.e. the metadata dir can be separate on ssd, and thus media drives don't need to be accessed to browse. I've also stored next to media as its the most compatible. is there a way to do both and keep in sync?

For even better performance, than yes, it should be on ssd, but the primary reason for that is to avoid HDD's spinning up - not for performance reasons.   I'm sure there are lots of ways to do this, by using DrivePool software, you can just automatically 'place' those file types on an SSD - unsure if this is available in Linux but I'm sure it can be done with symbolic links etc.

6 hours ago, ecrispy said:

1. video seek previews - planning to enable .bif generation for all media this time. is it supported on all clients? I'm guessing it won't work with Kodi though?

Correct, only emby direct clients uses the BIF - but believe that may be changing soon anyway to a more portable format.

6 hours ago, ecrispy said:

any other suggestions/ideas?

Try and stick with core clients if you want an emby native and relatively painless experience.   When you start using 3rd party/plugins - you are potentially opening up yourself to a world of pain.

As a brand new install - I would go emby out the box - with just a small test library - if it meets your needs then stick with it, if not, then try other options.

Edited August 12, 2023 by rbjtech

[ecrispy 38]

9 hours ago, rbjtech said:

but believe that may be changing soon anyway to a more portable format.

any info on this? isn't bif the potable format as its also used by Roku etc?

9 hours ago, rbjtech said:

Try and stick with core clients if you want an emby native and relatively painless experience.   When you start using 3rd party/plugins - you are potentially opening up yourself to a world of pain.

 

Isn't Emby for Kodi considered a core client? On my Odroid N2 there are no Emby clients other than Kodi so there's no other option. 

[Luke 42083]

Yes and yes. We may eventually move away from the bif format but it won’t be in the short term.

[visproduction 316]

Quote

1. this thread isn't to discuss quality as it depends on source, but given my clients, they should all play hevc 10 bit. I know transcode h264->hevc will lose quality but a lot of people have said difference isn't noticeable even with nice tv/audio? has anyone else done this?

I see the difference, even on a smaller monitor and I don't like it.  Why not increase your storage size and try to get original h.265 for all new content?  Also test if your system can handle any transcoding demands when users cannot play h.265 back?  I don't know what you have for content, but typically new content is hit maybe 30 times more, on average, than older content.  Does it make sense to spend so much effort to convert all libraries for content, that may actually get used very little?  Are you doing it because you like to organize your files and keep them more compact?  What does that decision have to do with what users need?  If you want more space, cycle out content that has not been viewed for 6 months.  You can always bring anything back in a special collection promotion.

A quick check today, it seems that Facebook and Youtube prefer uploads in h.264. I think if you upload h.265 they will convert it anyway and not store it on their servers as h.265.

No post production service would convert from h.264 to h.265, without getting the client to sign a statement that they understand the quality will alter and conversion fees are not refundable.  The best quality h.265 is a first conversion from a master.  Normally, that is the only production path that is offered to a h.265 copy.

It is true that most people will not notice a h.264 conversion to h.265 difference.  But it will be more fatiguing to watch this reconverted media.  Cable and online products are known to loose clients with a quality drop.  People just get tired faster, watching content that has edgy artifacts, even if they cannot tell you that the video looks any different.

These are just my opinions.
Hope it helps.

 

Edited August 14, 2023 by visproduction

[ecrispy 38]

On 8/14/2023 at 9:02 AM, visproduction said:

I see the difference, even on a smaller monitor and I don't like it.  Why not increase your storage size and try to get original h.265 for all new content?  Also test if your system can handle any transcoding demands when users cannot play h.265 back?  I don't know what you have for content, but typically new content is hit maybe 30 times more, on average, than older content.  Does it make sense to spend so much effort to convert all libraries for content, that may actually get used very little?  Are you doing it because you like to organize your files and keep them more compact?  What does that decision have to do with what users need?  If you want more space, cycle out content that has not been viewed for 6 months.  You can always bring anything back in a special collection promotion.

A quick check today, it seems that Facebook and Youtube prefer uploads in h.264. I think if you upload h.265 they will convert it anyway and not store it on their servers as h.265.

No post production service would convert from h.264 to h.265, without getting the client to sign a statement that they understand the quality will alter and conversion fees are not refundable.  The best quality h.265 is a first conversion from a master.  Normally, that is the only production path that is offered to a h.265 copy.

It is true that most people will not notice a h.264 conversion to h.265 difference.  But it will be more fatiguing to watch this reconverted media.  Cable and online products are known to loose clients with a quality drop.  People just get tired faster, watching content that has edgy artifacts, even if they cannot tell you that the video looks any different.

These are just my opinions.
Hope it helps.

 

you make good points. let me ask you - do you see the difference between 720p/1080p/4k at all times? I find that its very dependent on the source material (as well as bitrate of course). I've tested this with my own media as well as streaming services. and for a lot of tv shows/moves the 1080p is just as good as the 4k unless you pause on certain scenes.

if you go on the unraid forums and tdarr forums, there are a LOT of people with much more experience and brains than me who do this for their media collection. A lot of them also use gpu for encoding, it is understood that cpu gives much better quality but with an Nvidia gpu you get 10x faster encodes.

I am not really interested in uploading to youtube/fb. Youtube will use vp9 which is basically as efficient as hevc but has never become popular even thought its just as good.

I have a lot of hard disks and can always buy more. but I cannot afford to duplicate my media and another consideration is not needing all the disks connected. I don't have 'users' other than family and they are all given the FireSticks which have no issue with hevc, and its actually less  data to stream.

[visproduction 316]

Ecrispy,

Ah, I love discussing details on quality.  There are a lot of factors.  How far you sit from the screen, size of the screen, what pixel detail can average eyesight see at that distance have a direct effect on the user experience.  Small screen and you sit way back, you won't see much difference between 1080P and 4K.  This cut off is more important than people think.  On an average screen size the ability to resolve pixels on 4K on a 65" monitor can drop below 20/20 vision at any distance past 8 feet.   This measurement applies only to still images.  Once there is a moving video, people can still see difference further back.  That is much more difficult to measure.  There is accepted math (photography: circles of confusion) to figure this out for average eyesight.  I have a whole page about such things on my site with demos.  There are interactive tests to see where your visible acuity lies.  There area lso THX specs screen sizes for home theater amd other graphs to show what size TV at home matches where you like to sit in a movie theater.  Lots of fun stuff:   https://www.producerelease.com/blu-ray/brres.htm  Try out the tests and see what your distance is to see a 4K difference.  (I stop seeing a 4K vs. 1080P pixel difference on a still image at around 9 feet away from a 65" TV.)

Anyway, if you are sitting within the range to be able to see 4K on a particular screen size, then yes, I can see the difference... but 4K media streams bandwidths are different.  Currently, HD blu-ray has the best bandwidth to playback for the best available quality and even then this varies from title to title.  4K streaming is another animal.  Netflix, Prime and whomever else that does it, try to get an amazing media version to stream in 4K.  They might jump through hoops to encode it directly from a master with high quality and a longer time to encode, so they can sell theirr top media on their service in 4K to look amazing.  The bandwidth is probably 1/2 or maybe even 1/4 of HD blue-ray, so some quality has got to give. 

Blu-ray is still using h.264 and I think Netflix and Youtube are also doing that plus some AVI and are not yet using h.265 because people with browsers can be stopped by h.265.  The way that h.265 encodes has many improvements which can result in better contrast, details, edge sharpness and compression to get a nice looking playback are all better than h.264.  However, h.264 has, in my opinion has more natural edges on everything in the frame.  The difference is similar to putting on yellow sunglasses which takes out any blue haze of daylight.  When you look out with yellow sunglasses edges are sharper because there is no blue color defracting the edges.  Yes, it looks like your eyesight just improved, but it still is not natural.

4K with a little lower bandwidth for streaming can also result in color shade clumping.  You might not have the bandwidth to do a gradual color shift in a sunset, so you can get bands of color like stripes on a zebra.  Also streaming 4K can start to have motion pixelization in heavy motion scenes, fog, explosion, panning, fast moving objects and so on.  Sometimes a 1080P copy of the same scene has proportionally more bandwidth per pixel group and motion artifacts look better.  This happens because whoever made the streaming playback files, may have set a bandwidth for 1080P, but not made it at least 4 times larger for the 4K version.  In these cases 1080P can look nicer for some issues.  Automatic encoding usually is not set to give 4K 4 times the size of 1080P.  A lot of the quality issues will happen because of the encoding choices and that the master used is already compressed, so the encoded copy is compression ontop of compression.

In film and TV post, new versions of content is only pulled from masters, so the result never gets the compression on top of compression problem.  So, when you ask about, can you see the difference, the answer really depends on what you are using to playback for your tests.  Of course, most people will not be this critical, but if the quality is not there, viewers will get more tired, quicker from viewing and the same principle applies to audio.

I hope this is interesting and not just too much to read.
 

Edited August 19, 2023 by visproduction

[ecrispy 38]

11 hours ago, visproduction said:

Ecrispy,

Ah, I love discussing details on quality.  There are a lot of factors.  How far you sit from the screen, size of the screen, what pixel detail can average eyesight see at that distance have a direct effect on the user experience.  Small screen and you sit way back, you won't see much difference between 1080P and 4K.  This cut off is more important than people think.  On an average screen size the ability to resolve pixels on 4K on a 65" monitor can drop below 20/20 vision at any distance past 8 feet.   This measurement applies only to still images.  Once there is a moving video, people can still see difference further back.  That is much more difficult to measure.  There is accepted math (photography: circles of confusion) to figure this out for average eyesight.  I have a whole page about such things on my site with demos.  There are interactive tests to see where your visible acuity lies.  There area lso THX specs screen sizes for home theater amd other graphs to show what size TV at home matches where you like to sit in a movie theater.  Lots of fun stuff:   https://www.producerelease.com/blu-ray/brres.htm  Try out the tests and see what your distance is to see a 4K difference.  (I stop seeing a 4K vs. 1080P pixel difference on a still image at around 9 feet away from a 65" TV.)

Anyway, if you are sitting within the range to be able to see 4K on a particular screen size, then yes, I can see the difference... but 4K media streams bandwidths are different.  Currently, HD blu-ray has the best bandwidth to playback for the best available quality and even then this varies from title to title.  4K streaming is another animal.  Netflix, Prime and whomever else that does it, try to get an amazing media version to stream in 4K.  They might jump through hoops to encode it directly from a master with high quality and a longer time to encode, so they can sell theirr top media on their service in 4K to look amazing.  The bandwidth is probably 1/2 or maybe even 1/4 of HD blue-ray, so some quality has got to give. 

Blu-ray is still using h.264 and I think Netflix and Youtube are also doing that plus some AVI and are not yet using h.265 because people with browsers can be stopped by h.265.  The way that h.265 encodes has many improvements which can result in better contrast, details, edge sharpness and compression to get a nice looking playback are all better than h.264.  However, h.264 has, in my opinion has more natural edges on everything in the frame.  The difference is similar to putting on yellow sunglasses which takes out any blue haze of daylight.  When you look out with yellow sunglasses edges are sharper because there is no blue color defracting the edges.  Yes, it looks like your eyesight just improved, but it still is not natural.

4K with a little lower bandwidth for streaming can also result in color shade clumping.  You might not have the bandwidth to do a gradual color shift in a sunset, so you can get bands of color like stripes on a zebra.  Also streaming 4K can start to have motion pixelization in heavy motion scenes, fog, explosion, panning, fast moving objects and so on.  Sometimes a 1080P copy of the same scene has proportionally more bandwidth per pixel group and motion artifacts look better.  This happens because whoever made the streaming playback files, may have set a bandwidth for 1080P, but not made it at least 4 times larger for the 4K version.  In these cases 1080P can look nicer for some issues.  Automatic encoding usually is not set to give 4K 4 times the size of 1080P.  A lot of the quality issues will happen because of the encoding choices and that the master used is already compressed, so the encoded copy is compression ontop of compression.

In film and TV post, new versions of content is only pulled from masters, so the result never gets the compression on top of compression problem.  So, when you ask about, can you see the difference, the answer really depends on what you are using to playback for your tests.  Of course, most people will not be this critical, but if the quality is not there, viewers will get more tired, quicker from viewing and the same principle applies to audio.

I hope this is interesting and not just too much to read.
 

haha, I love these types of discussions too. I've read the usual stuff about distances/resolution and I have a 65" tv and sit about 6-7ft from it, so theoretically I should be able to see the differences. And yes, I of course can. But its not night and day. Its most certainly not sd vs hd back when BR came out.

Its the same with Atmos which IMO is the biggest scam (i.e. hype) in audio. Going from 2.0 to AC3 5.1 was massive. HD audio is a much smaller gain - I've attended gatherings put together with members of avsforum.com, who are all nutcases with $$$$ amount spent on insane audio/video setups, and these people cannot tell difference between a TrueHD and 640Kbps DD5.1 track in 90% of cases. The difference in Atmos is even less. There was someone who used to do a test of Atmos soundtracks by disconnecting every speaker except overheard. and the amount of sound was miniscule.

so back to video - the quality of the encode, the crf factor etc, matter much more. I have neither the time equipment or expertise to tune these but others do, and if you download their encodes you cannot tell any difference. I've tested my hevc encodes, things like black backgrounds with variation (i.e. not fully black) is so problematic, you see pixellation, banding etc. Motion seems to be handled better in hevc.

I do believe h264 is slightly better visually for most cases except for certain high quality encodes. e.g. very slow preset crf 10-15 is going to take years but the result is basically identical.

But here's the thing - if I can get 30-50% reduction in size for 5% loss in visual quality I will never notice till pointed out, then its a compromise I will take any day. Its not really about more storage, its consolidating and simplifying things.

the online landscape is forever shifting, when you upload a video internally it gets reencoded to maybe 10-15 different resolutions/bitrates as well as adaptive hls streams. There's a reason Google picked VP9. Remember h264 and then h265/hevc when they came out - cpu pegged at 100%, playback was limited and people used to buy dedicated hw cards and setup Madvr on $$$$ htpc's for playback. Those days are long gone. Been there done that. Now its impossible to buy a streaming stick that won't do hw hevc decode. AV1 is coming as well and in 2-3 years you will have $25 players with hw av1 decode.

we are probably approaching the perceptual limits of av compression. But I think we'll see a couple more iterations of codecs. And lossless audio/video is going to be a thing of the past.