All Activity

Sometimes the playback will stop for 10-20 seconds or it just freezes completely. This morning about 8:40 am CST it froze three times in different places while playing GMA. We started watching from the beginning and while it was still recording GMA. I upgraded the QNAP server software a few days ago to version 4.9.5.0, this made no change. embyserver.txt embyserver-63924508802.txt

Hi, I was checking out your CSS theme (thanks for putting that together, by the way) and noticed a small typo: On line 735 it currently says: background-color: RGB(from var(---hpRow1bgcolor) r g b / 100%)!important; It should just be --hpRow1bgcolor (two dashes). Everything else seems to work fine.

Hey everyone, I'm using the Emby Android app on my Sony Google TV. Lately, whenever I watch Dolby Vision content, there is constant micro-stuttering/judder. If I open the Apple TV app to watch the exact same content (4K DV at 23.976 fps), playback is buttery smooth. It seems like a bug has crept in. Everything runs perfectly smooth in Emby with HDR10 or standard SDR, this issue only happens with Dolby Vision. Has anyone else run into this or found a fix? Emby for Android 3.5.55

same problem here on my nvidia shield pro

Hi, is there any update regarding the issue? It seems that the issue still exists even in the latest and newest version of the server.

Checked 4.10.0.40, the stable build, and the tail behaves exactly as it did in the betas. I also updated my own production server to 4.10 and it stops in the same place there. Same setup as before — fresh container, same synthetic sample, same probes. Each row is 4.10.0.30 -> 4.10.0.31 -> 4.10.0.40: - identical final segments, 6-second segments: 2 -> 2 -> 2 - identical final segments, 3-second segments: 3 -> 3 -> 3 - md5 of those segments: f40fe109... -> the same -> the same - last fMP4 segment, fresh sessions: no response -> no response -> no response (8 of 8 ) - "FfRunException" in the server log: 0 -> 0 -> 0 The image id is new and there are nine builds between .31 and .40, so plenty was built — just not this. And nothing regressed either: the .30 fix still holds, FfRunException is still zero. The mechanism is the one from #comment-1537253: ffmpeg is restarted per segment and seeked to the nominal segment boundary with the video copied, so when a file's last keyframe falls before the penultimate segment boundary with none after it, every later run rewinds to that same keyframe and emits the same frames. The recipe for a file that reproduces it is in that post and takes about two minutes to build. Mentioning it because until now everything in this thread was beta testing, and this build is the one people are running. Still happy to test anything you want against this setup.

Its becoming clear that you guys dont want to help me by asking stupid questions like "what website" when im on the website that emby provides. Just give me information to how to start from scratch and ill be on my way, cause clearly your all must be new to this

I see the same thing. Using Docker - emby/embyserver:latest

Some confusion due to cross posting, I am responding but at the same time you are posting so I will update. "The gracenote schedule shows the airings for Charlie and Jerry on BBC R1 for 8th to 17th September as being repeats of the 7th September aired episode. It changes for the airing of the 21st September when it is showing as different episode. Obviously this is wrong and I will refer it to Gracenote." Correct. None of these shows are repeats. They are all live. "Both did not work as the air date is 9 Sep 2026" Charlie an Jerry should not be a repeat and the air date is 9 Sep 2026 DJ Day Day Mix again is not a repeat and the air date is 9 Sep 2026. This is probably related to point 1. "But this does not match the channel numbers - for example BBC R1 is 181 on this lineup and not 700." It's the channel number that is assigned by the tuner. If you check here: https://en.wikipedia.org/wiki/List_of_DTT_channels_in_the_United_Kingdom You will see that BBC radio 1 is 700. I am not sure why it is set to 181 in the lineup. I have not changed the channel numbers. BBC Radio 1 has always been 700.

Users table view overlay persists after page navigation. Running on QNAP, Emby Server 4.10.0.40 dashboard_user_table.mp4

I am seeing this as well in both the TestFlight iOS and web apps. Prior to server version 4.10.0.40, I had only ever had the issue on the TestFlight version of the iOS app. I don't see any specific pattern to cause the issue, it just seems to happen about 75% of the time I go to the dashboard.

I have just tested playback of a TV recording episode on 4.10.0.40 and the recording does not have a season folder. I stopped playback part way and it did show in Continue Watching after this So will need to have a look an example with log provided for when you do that and then afterwars go to the Home page and refresh and look at continue watching - are there a lot of items in Continue Watching ? Please give details of the episode you playback for the test and also give details of the Home Section configuration (User Prefereneces / Home Screen / Contine Watching edit screen - showing library selection etc) Thanks

Each of these needs to be investigated and schedules checked So far I have only looked into the BBC R1 schedules. I will need to look into each of the other examples to understand what has happened Are the airings for New Music for Jackie Sanders all different - I will do some tests But this does not match the channel numbers - for example BBC R1 is 181 on this lineup and not 700. I suggest you stick with either the Freeview lineup for your city or the Local Over The Air Broadcast lineup for your post code.

sorry - your screenshots do show the channel numbers

This has now been referred to Gracenote

The gracenote schedule shows the airings for Charlie and Jerry on BBC R1 for 8th to 17th September as being repeats of the 7th September aired episode. It changes for the airing of the 21st September when it is showing as different episode. Obviously this is wrong and I will refer it to Gracenote The time is only important for shows where they have different time based airings on the same day. It is a separate matter Can you expand on this sentence - Both: Which Both. Did not work: what happened You are mixing channels above - please for each example you give, specify the channel number to make it easier to investigate

@richtwell you prediction was right Beta got bumped up to stable today.

Cross posts jack saunders is NOT correct. It says 2024-05-13. channel 700 (BBC R1) channel 724 (Capital) is correct. "which of the airings recorded with the right filename when you switched to the United Kingdom (extended) National Virtual Lineup" It was my mistake. I thought it had worked. It had not. I missed that the jack saunders was 2024 and not 2026., "I can see why the Jack Saunders would have the same filename - it is because the schedule has the airings as a series/show and not as episodes - so they will all have the same series file name - I would expect them to show as different versions - in the screenshot for this recording it shows two recording files - did you stop and restart the recording on the 8th september? " Yes i stopped and started to test if it added -1 to the filename which it did. "What is the ID of the lineup that you are using - this shows in the Live TV Settings Setup screen - it will be GBR-XXXXXX" United Kingdom [extended] [National Virtual] I should have been clearer. I thought that by changing the guide data provider it had worked. It had not, I just used a different episode for the test. That particular episode worked, the original radio channel did not work. WMC works fine and puts the date in. Both can only use 1 guide data source. What should the series meta data downloaders be for the library? Right now they are all empty. Emby WMC Both did not work as the air date is 9 Sep 2026 <?xml version="1.0" encoding="utf-8" standalone="yes"?> <tvshow> <uniqueid type="gracenote">33871118</uniqueid> <uniqueid type="zap2it">EP06571521</uniqueid> <title>Charlie and Jerry</title> </tvshow> Emby appears to take the air date and make that the episode date. WMC just takes the date and then the time of the program and puts it in the filename. Air date 9 Sep 2026

Ooops!

That is so much better than what I had. Looks very good. But I ultimately decided for a different approach. I'm now using collages which look like this: I do like them, but your version is definitely tempting.

Thank you very much for your detailed help. To make sure nothing goes wrong, could you please tell me exactly which checkboxes to select and which settings to configure so that only my local trailers play and no YouTube trailers play in the two plugins on the server for movie intros and trailers, and what exactly needs to be enabled in Kodi for the Emby plugin? Thanks and best regards

Hi, I'm experiencing severe and consistently reproducible judder with 1080p50 H.264 content in Emby Windows 2.317.2. System Emby Windows 2.317.2.0 (Microsoft Store) Windows 11 AMD Ryzen 3 3200G Radeon Vega 8 iGPU AMD driver 31.0.21925.1001 LG 4K OLED connected to the PC Display manually set to 50 Hz during testing Emby Server on TrueNAS Test file MKV H.264 High@L4.2 1920x1080 50.000 fps progressive 8-bit YUV420P BT.709 SDR ~3-3.7 Mbps AAC 2.0 Direct Play The same exact file on the same PC and network connection plays smoothly in: MPC-HC Emby Web in the browser Moonfin In Emby Windows, however, playback has continuous visible judder. I have reproduced the problem with both software decoding and D3D11VA. Updating the AMD graphics driver did not change the behavior. I also tested with Windows itself set to 1920x1080 @ 50 Hz, with the same result. The most interesting comparison is: GPU-Next + D3D11VA + All Hardware processing: → visible continuous judder legacy GPU + D3D11VA + All Hardware processing: → smooth 50 fps playback (but darker image) So the system is capable of decoding and presenting the stream smoothly; the problem appears to be specific to the GPU-Next playback/rendering path. During the GPU-Next reproduction, the mpv debug log repeatedly reports: Invalid video timestamp Non monotonically increasing PTS Render queue underrun At one point it also reports: Estimated source FPS: 50.000, display FPS: 25.000 while the D3D11 output itself detects the physical display at 50 Hz. I have attached the complete debug log from this reproduction using the current AMD driver. Since the same file plays smoothly in MPC-HC, Emby Web, Moonfin, and even Emby's legacy GPU output on the same machine, this looks like a GPU-Next/frame-presentation issue rather than a decoding, network or source-performance limitation. Please let me know if you need a short sample of the source file or a reproduction with a specific configuration. Thanks. Emby_Windows_2.317.2_1080p50_judder_debug_log.txt

In the examples that you showed in your screenshots So we have: "Radio 1's New Music Show with Jack Saunders" from channel 700 (BBC R1) "The Capital Evening Show with Jimmy Hill" from channel 724 (Capital) "Pre-Match" from channel 713 (BBC Radio London?) and which of the airings recorded with the right filename when you switched to the United Kingdom (extended) National Virtual Lineup When I compared the schedule and metadata from gracenote for BBC R1 (channel 700 on the Local Over the Air Broadcast lineip and channel 181 in the national lineup) - they were exactly the same - only difference being channel numbers. So can you please tell me which programs and airing times recorded with what you believe is the right filename when you switched to the national I can see why the Jack Saunders would have the same filename - it is because the schedule has the airings as a series/show and not as episodes - so they will all have the same series file name - I would expect them to show as different versions - in the screenshot for this recording it shows two recording files - did you stop and restart the recording on the 8th september? What setting do you have for checking if the airing is already in the library/libraries ? And is the issue that you are not getting them recorded as opposed to be being recorded and appearing as different versions? What is the ID of the lineup that you are using - this shows in the Live TV Settings Setup screen - it will be GBR-XXXXXX

@vdatanet That's actually a much better way of putting it, and I mostly agree with you. You're right that “the server isn't in the media path” is, by itself, an architectural difference rather than automatically an advantage. Whether it becomes an advantage depends on the setup. But you've also just listed several concrete situations where it does become an advantage: avoiding an unnecessary network hop, players whose native file access behaves better, and media structures that benefit from real file access. And that's really all I've been arguing. I never claimed that SMB universally beats HTTP, nor that it somehow produces better picture or sound from identical data. In my particular setup, everything is local, the player and storage are on the same network, Kodi has excellent native SMB support, and I neither need nor want the server to mediate media delivery. So Direct File Access is the architecture that suits that setup better. For a browser, Chromecast, remote client or a client that may require remuxing/transcoding, HTTP obviously has major advantages. I wouldn't suggest otherwise. So yes: two architectures, different trade-offs, and which one is preferable depends on the setup. Which brings us to the slightly ironic conclusion that Direct File Access does indeed have concrete benefits — just not for everybody. And that was my point all along.

Just like in the image, circled in red, it appear when any 3d field is selected. It's fun that field can be replaced or added a "IMAX Enhanced" badge, or, more better, just a single flag yes/no in metadata edit