« Previous | Main | Next »

Wimbledon HD HTTP Streaming Trial

Post categories:

Andy Armstrong Andy Armstrong | 09:30 UK time, Wednesday, 22 June 2011

Staff watching television in an office

Andy Armstrong's colleagues testing HTTP HD streaming video

I'm Andy Armstrong, technical architect for Programmes and On Demand here at the BBC. We investigate new technologies for online video and audio delivery looking for ways to enhance existing services, provide new services and streamline the way we handle media content on the internet.

Important sporting events like Wimbledon are great opportunities for us to test new technologies. This year we're testing a number of things including high definition HTTP adaptive bit rate video streaming.

Currently we use multiple approaches to bring video to your computers and other connected devices. The majority of you are using iPlayer in your web browser. That uses RTMP - a protocol developed by Macromedia (now Adobe) specifically for streaming audio and video over the web.

Unfortunately RTMP isn't available on all of the devices we support so we have to use other protocols too - principally HTTP and HTTPS progressive download - the same mechanism that is used for web pages and images.

We're always on the lookout for ways to support a wider range of computers and connected devices. If we can find a way to do that and at the same time serve more of them with a common protocol we will be able to provide a better service and make our infrastructure simpler and more efficient.

One technology that may make that possible is adaptive bit rate HTTP streaming. It works by splitting a video into short chunks (typically between 4 and 10 seconds long). Each chunk is available in a range of bit rates and hence qualities. For each chunk the player decides which bit rate to request based on its estimate of how much bandwidth your internet connection has.

We already use ABR streaming for some sports coverage with a limited range of available bit rates. For the Wimbledon test we will be using Adobe HTTP Dynamic Streaming for the first time and will be supporting bit rates of 416k, 624k, 944k, 1408k, 2112k and 3168k and a maximum resolution of 1280x720 (720p HD).

Until the player has had a chance to assess how much bandwidth it has available it will tend to choose lower bit rates. That means that the first few seconds of the video may be lower quality. Then, if you have enough bandwidth, the player will gradually select higher bit rates until it either detects that it is using as much bandwidth as it can safely sustain or it reaches the highest available quality. To get the highest bit rate the player needs to be displaying full screen video; the embedded version will limit the maximum bit rate to 1408k.

The player maintains a small buffer of downloaded chunks so that it can keep playing without interruption as the available bandwidth varies. At times it may detect that the overall available bandwidth has reduced and switch to a lower bit rate to compensate.

All this should happen smoothly and unobtrusively. Buffering times should be greatly reduced. The fact that the player adapts to the available bandwidth is particularly helpful for live events such as Wimbledon; it allows the player to display the highest possible quality video without lagging too far behind the live action as it struggles to buffer downloaded video.

However - as you might have guessed - if we had all this working properly now we'd be using it for the main web coverage of Wimbledon rather than testing it. This is an extremely promising technology but we don't feel that we know enough about it yet to do that. We know, for example, that the method the player uses to estimate the available bandwidth will need to be adjusted and we need to see how our servers handle this new kind of traffic.

So we certainly can't promise that it will be perfect. In fact we'd be rather surprised if it was. We have been testing it with a small group of BBC staff since February; now we've reached the point at which we need your help to find out how it works for a larger, more representative audience.

row of diamonds showing increasing definition with increasing bandwidth

For the test we have placed a small icon (see above) in the upper left corner of the video that shows the current bit rate that you are viewing (1 to 5, HD). It will be interesting to hear from you how quickly the player settles to a sustainable quality - that can vary depending on the characteristics of your connection. If you haven't seen it already take a look at the test page.

We are extremely grateful for your help. We expect to run more tests of HTTP streaming over the coming months and it will be particularly valuable for us to have feedback and suggestions from those of you who have been in it from the start.

Andy Armstrong is the Technical Architect for Programmes and On Demand.


Page 1 of 3

  • Comment number 1.

    Umm, the test page says 'trial over for Day two'. It's day three now and I was keen to try it out...

  • Comment number 2.

    Falken: thanks. You're quite right - that text was left over from yesterday. We've changed it now.

    Thank you.

  • Comment number 3.

    You seem to sugest in your post that Flash is not needed yet when I went to your test page it said flash was needed. Can these streams be used by players other than flash and what is the url. Many mobiles do not support flash as so this could be important to quite alot of people. I look forward to trying this out.

    I recently discovered that your national radio progams are available on itunes using high quality AAC encoding but without flash. I can now get these streams on my internet radio. The odd thing is there has been no anouncement from the BBC so many internet listeners don't know about this. Some web sites such as Tunein Radio are now using these streams.

  • Comment number 4.

    Apologies if I gave that impression Trevor. Yes, for the trial Flash is required.

    In general our aim is to use common infrastructure to handle as many devices as possible - both Flash and non-Flash enabled. However for the trial we're concentrating on Flash based streaming.

  • Comment number 5.

    Oh, came here for the misleading title, disappointed in the contents. I have a HTTP tool, it's called Firefox. Still waiting for BBC to trial streaming WebM, do understand theres a small collection of people that cannot run Flash on their platforms.

  • Comment number 6.

    Sorry for the disappointment John. We're well aware that there are platforms that don't support Flash and platforms that don't support h264.

    Although this trial is limited to Flash/h264 capable devices it's part of a wider effort to build a common streaming platform that will serve the widest possible range of devices - both Flash and non-Flash.

  • Comment number 7.

    My main concern is the BBC's reliance on Flash as it's cross platform streaming solution. For example Adobe recently dropped AIR support for Linux meaning the iPlayer desktop is now available on one less platform. What would happen if they dropped Flash support for Linux? Even now it's not uncommon for major security holes in Flash to linger for weeks on non-Windows platforms.

    I do hope the BBC does have some plans in place on what to do if it was forced into a post-flash streaming world. Hopefully if that does happen the solution can be built on openly implementable standards.

  • Comment number 8.

    Alex: yes, that's a concern for us too.

    HTTP adaptive bit rate streaming is conceptually platform neutral - in the sense that it's a technique that can be used on many different platforms albeit with slightly different implementations. The trial is Flash-specific because it's widely supported and that gives us a good opportunity to tweak the algorithm that the player uses to switch between available bit rates. It also lets us observe how well our content delivery networks handle HTTP dynamic streaming traffic.

    Much of what we learn - and much of the infrastructure we're building to support the trial - will be applicable to non-Flash devices too.

  • Comment number 9.

    And now for some feedback:

    While running in the browser the little icon goes up to 4. If I full screen the plugin the image freezes although I can still here audio. Hitting Esc to drop back to normal view I can see the icon at HD before it drops back down to 4.

    Trying again I got playing fullscreen, to 5 and then it froze (presumably when HD kicked in). Esc again returns me the normal view.

    Right clicking on the fullscreen does show it has selected the 3168k stream.

    Adobe Flash
    Chromium 14.0.797.0 (Developer Build 89613 Linux) Ubuntu 10.04

  • Comment number 10.

    For some reason, it shows tennis match from 1-4 but when it gets to HD it switches to some upscaled cached clip of BBC News. Hmm...

  • Comment number 11.

    Is there any chance of tweaking the numbers a bit? As things stand the transparency makes the numbers quite hard to read. IMHO an opaque background and a black number would be much clearer.

  • Comment number 12.

    Alex: I've just tried Flash, Chromium 12.0.742.91, Ubuntu 11.04. That combination is fine on my hardware.

    Could you try toggling the hardware acceleration on the Flash player please? You can get to that setting by right clicking on the player then Settings => Display.


  • Comment number 13.

    I have used the trial yesterday and today and so far it has been excellent. It takes about 15-20 secs for it to get up to HD but when it does the quality of the image really is superb. Previously I would avoid watching online as the quality wasn't great but if this trial is successful it would be an impressive leap forward for online viewing. Having a varying quality sounds like a very sensible approach to ensure everyone gets the best experience possible. Having said all this I am aware that I have a very fast internet connection and I imagine other users probably don't benefit as greatly as I do!

  • Comment number 14.

    andyb: I think that the bit of BBC News you're seeing is from the start of today's coverage - although clearly you shouldn't be seeing it now. I'm going to assume that's a bug with our player and raise it with the appropriate people.

    Roughly what time did you start watching?

  • Comment number 15.

    I don't have any of the issues stated above, when I switch to full screen I get beautiful 720p HD stream (thanks to my 50M connection)

    but when I minimise the screen, sometimes the feed reverts back to the 'Coverage will commence shortly' screen that was broadcast before the tennis went live on the stream. I'm running XP Home 32bit SP3 and firefox 4.0.1

    Otherwise, thanks for this wonderful HD stream I've been waiting for so long to enjoy!

  • Comment number 16.

    Watching the V Williams match now. Going to full screen I'm getting a 4 then 5 then HD in just a few seconds. Seems to be working just great for me.

  • Comment number 17.

    When will I be able to watch BBC programmes in HD on the iPlayer on the PS3?

  • Comment number 18.

    @Andy: Unfortunately right click->settings leave Flash on the Webcam tab and I can't move from it or exit the settings dialog.

    My machine is running with Intel graphics and 3D acceleration enabled. However I don't believe flash takes any advantage of it. I thought it only could use nVidias proprietary hooks?

  • Comment number 19.

    I'm running this from my office (in the) UK. But we have a German IP address so I need to run through a proxy, so my setup is complicated. I've just been watching for about 10 minutes. What is interesting is that I get up to 5 successfully but then the video stutters and after a 1 or 2 second delay I drop back to 3. Then after a while get to 5 and the stutter happens, and the circle repeats.

    First of all this is a *huge* improvement on the fixed bit rate scheme, where if you run out of bandwidth, the video completely stops, you wait for the buffering.

    My question and I guess one of the things you're trying to find out is at what point would/do you set a bitrate ceiling and then how often do you test that ceiling by moving to the next rate up? or does it work differently?

  • Comment number 20.

    Alex: we think we've managed to reproduce the problem on a Windows box running Flash 10.3 - we seem to get exactly the same symptoms you describe. We'll investigate further and report back.

  • Comment number 21.

    How many frames per seconds are you encoding on the HD stream?

    I ask because I noticed a lot of stutter in the video when the camera pans from side to side during matches. There wasn't any motion blur and the fine detail in the net was preserved across frames... but there is a distinct stutter in the picture during whole-screen movement.

    I have a very fast video card and a Core i5 processor in my computer. I'm running Win7 with Flash 10.3.181 and I'm on a high bandwidth university internet connection.

  • Comment number 22.

    In revision to my earlier post, I momentarily had the BBC news intermission issue, on the very same system (XP Home 32bit SP3 and firefox 4.0.1), although a ctrl+F5 refresh solved the issue.


  • Comment number 23.

    John: I'm not sure what are plans are in relation to PS3 - and to be clear we're still at a fairly early stage with this. I suspect that we're not in a position to make that kind of commitment yet but I'll see what I can find out.

    Chris: I assume that our current bit rate switching algorithm is getting confused by your proxy setup. In general the bit rate ceiling will be dynamic - the player constantly estimates the available bandwidth and should adapt up and down as it varies.

    We're running with fairly simple-minded switching rules at the moment - and the cyclical switching behaviour you're seeing is something we've occasionally seen in testing too. It's useful to us to know that it can happen in the wild as well.

    Thanks for feedback.

  • Comment number 24.

    Works really well for me and I'm not even on a blazing fast broadband connection. Goes through the numbers and up to HD in full screen in a matter of seconds and holds it at that quality very well. Just hope that this is something the BBC will consider to implement in other live sports broadcasts too.

    Running latest Chrome and Adobe Flash

  • Comment number 25.

    David Bailey: we're encoding at 25 frames per second. We've seen the kind of stuttering you describe but it's certainly not something that we consider to be acceptable.

    Do you happen to know what the video card is in the computer you're using?

    It would be interesting to hear what happens if you enable / disable hardware acceleration for the Flash Player (right click on player, Settings => Display).


  • Comment number 26.

    Andy Street: yes, we were hoping for cobalt blue skies to make the numbers stand out better...

    You're absolutely right that they're not clear enough. I'm not sure how easily we can adjust them - almost certainly not for today but I'll look into it.

  • Comment number 27.

    andyb, DaveShaw: the little fragment of old video is interesting. We think it's caused by you player getting an old copy of the 'bootstrap' file which describes which video fragments are available. We're not yet sure how that's happening but you've given us a really good lead, thanks.

  • Comment number 28.

    Simply stunning quality. Got all the way to HD in full screen mode on a 2560x1440 external monitor. Couldn't get it to work (at all) in Safari [No iPlayer window -- not even jpeg placeholder], but Chrome worked.

    MacBook Pro OSX 10.6.7 / Flash 10.3.181 / Chrome 12.0.742 / ISP on cable.

  • Comment number 29.

    Neil MacLennan: Lovely to hear that the quality is good for you.

    I'm playing it in Safari now; it should work for you. I wonder if you could give it another try once we get to a lull in the tennis action?


  • Comment number 30.

    My connection only gets me to level 5.

    Still infinitely better than iPlayer.

  • Comment number 31.

    Andy Armstrong: My video card is an ATI Radeon HD 4600 with 1GB of GDDR2 video memory. I tested with hardware acceleration on and off and the stutter was present in both, although not as noticeable with the acceleration on.

  • Comment number 32.

    I'm also seeing the "video freezes, audio continues" behaviour when I go full-screen. This is on Ubuntu 10.10, Firefox 4.0.1, Flash on a Lenovo x201 (Intel graphics) with a 1920x1080 external monitor.
    Changing the flash hardware acceleration settings makes no difference.

  • Comment number 33.

    I've been watching this on a 17" laptop, 3gb ram,in IE8 and have to say I'm impressed.Its flawless,the clarity is superb, sound is great. I'm mega impressed, lets have more.

  • Comment number 34.

    Sorry,meant to add, re symbol in top left, goes thru to 5 almost immediately then shows HD. I've had nothing less than 5 throughout, mostly HD.

  • Comment number 35.

    That's it working on Safari now. I use ClickToFlash, but normally iPlayer shows me a static image, *then* the Flash Player when I click through. Having uninstalled it, Safari works fine.
    Other notes: (1)After right clicking and changing hardware acceleration settings a few times, I lose the roll-over on the embedded video.
    (2) After watching in full-screen mode and going back to the embedded video, I'm still getting the HD stream ("HD" diamond). It flickered down to 5 for a moment or two then uprated back to HD. I just thought that happened in Full Screen mode?

  • Comment number 36.

    The quality is great, and on my 20Mbit ADSL connection, it flies up to HD very quickly.

    However, your Flash player is using the 'zoom' method of scaling when hitting fullscreen (compared to the normal Wimbledon player). Thus, when playing on monitors higher than 720p, the video can appear pixelated.

  • Comment number 37.

    Worked fine on my Windows 7 pc actually it was better than BBC One HD on my TV which has mosquitos around the graphic. The horizontal resolution of BBC One Hd is only 1440 so not very different from 1280.

    I tried it on my HTC Desire running android an it did not run very well at all. Half screen seem to be ok but full screen it kept on stopping.

    I will have another go tomorrow.

  • Comment number 38.

    Neil MacLennan: Yes, it sounds as if there's some potential for the player to get confused about which state it's in there. I'll ask the media player team for their thoughts.

    Richard: the full screen / no video problem is reminding the media player team of a problem they've seen previously; hopefully that means that a fix will be forthcoming soon - but I mustn't make promises on their behalf.

    David Bailey: thanks. I'm slightly envious of your computer and agree that there shouldn't be a performance problem. I've been playing the feed all day on a very old Pentium 4 box; if that can handle it I concur that your computer definitely should... Have you noticed any similar stuttering when you play video from other sources or is it just the HD trial?

    Trevor Harris: better than BBC One HD? We should probably keep quiet about that :)

    I had some success on a Nexus S - when the network was willing. It plays full screen quite happily. Does your HTC normally play full screen video?

    And a sincere thank you from all of us for the excellent, precise, insightful feedback you've given us today. I hope that the issues you ran into didn't spoil your enjoyment of the tennis too much.

    We're already working on the problems that have been reported. We may not necessarily be able to release fixes for them during the trial but we're all fired up by your positive feedback and helpful reports.

    Thanks and see you tomorrow.

  • Comment number 39.

    Craig Mason: sorry - missed your message. I'm seeing pretty smooth upscaling at resolutions greater than 720p. I wonder if it depends on your graphics card / driver... Having said that I'm not a Flash expert. I'll pass your comment to the people who are and let you know what they have to say, thanks.

  • Comment number 40.

    johndrinkwater: WebM is really just a container for VP8 and Vorbis. Unfortunately there are few viable platform options for broadacasters who need a large scale mechanism to protect and deliver video. Flash HTTP streaming is the best on offer today, and is widely supported. Offering WebM today is the same as any progressive download, so not much trialing needed.

    If you follow the industry you will know that there is a huge appetite for an "every device" video delivery platform. It must be able to cope with the volumes of users currently accessing live and on-demand video, offer content encryption and key distribution, total platform comparability (Windows, Mac, Linux, set top box, Android, WM7, iOS ....) have back-end tools for capture, encoding and editing, be license free..... It does not exist, and I have not seen anything come as close as the Flash platform.

    When the BBC require people to have Flash to access this type of stream there are always complaints. When the BBC required them to upgrade to a colour TV did any of them complain as loudly?

  • Comment number 41.

    Michael K: you've captured the problem succinctly thanks.

    We do absolutely understand the frustration people experience when they have a device which is technically capable of playing video but which we don't support. It's completely natural to ask "why not?". It even frustrates me and I already know why not.

    Unfortunately Michael's summation of the situation is accurate. We strive to support as many connected devices as possible but we don't have the benefit that the pioneers of television broadcasting had of a single universally adopted standard (at least on a national level).

    As an aside: I can't find any information about whether there were complaints during the upgrade to colour but, assuming there weren't, I imagine that would have been because the designers of the colour system were able to pull off the remarkable feat of designing a new system that was largely forwards /and/ backwards compatible with the old one - so B&W TV's worked with the colour signal and colour TV's worked with a black and white signal.

    We don't control the standards that exist in digital media - but we certainly express our opinion when there's an opportunity to effect changes that will allow us to provide a better service.

  • Comment number 42.

    @Michael: The BBC don't require you to upgrade to a colour TV. You can still watch on a B&W one if you wish.

    However, they DO require you to install the slow and bloated piece of software known as Flash if you wish to watch their stuff online, and since there are many ways around the encryption anyway, as users of get_iplayer very well know, I'm pretty sure they only use Flash to keep the managers happy. Which is rather irritating. I mean, using no encryption at all and having everyone happy would be better than having broken encryption and lots of frustrated Linux users, especially those with slower boxes which Flash seems to hate (I've recently been having issues getting iPlayer to run smoothly in fullscreen on a relatively old Pentium 4 and my netbook with an Atom - judging by Andy's comments above, this works fine on Windows).

  • Comment number 43.

    Muzer0: we have happy managers? I'm not sure I like the sound of that...

    To be serious: the reason we support Flash is that it's more widely available than any other single technology. Our remit explicitly - and rightly - requires us to be as inclusive as possible and Flash allows us to do that. It is, in that sense, purely a practical decision. And to the extent that it lets us deliver on our commitment to the licence payers, yes, it makes the managers happy too.

    We also recognise that it's inappropriate for us to force a technology choices on the public which is why we make services available for people who do not wish or are not able to use Flash. But, as Michael's summary explained it's a complex area and we just don't have the resources to support every device.

    As I mentioned in the post one of our goals in exploring the use of HTTP dynamic streaming is to be able to be both more inclusive and more efficient - to serve a wider range of devices with a simplified infrastructure.

    The use of encryption and the level of that encryption that we're required to use is often set by the rights owners. We're not doing it just to be awkward and it often complicates things for us too - for example it can be harder to diagnose problems and perform ad-hoc experiments.

    I'm sure that it sometimes appears that our non-support for a particular platform is gratuitous or dismissive - but I can assure you that that's not the case. We really do want to be on every device that's capable of playing video and make our programmes as widely and as easily available as possible.

  • Comment number 44.

    Muzer0: funnily enough the P4 box is running Ubuntu 11.04 / Flash 10.3 and it only has 1Gb RAM. I was quite surprised that it did as well as it did.

  • Comment number 45.

    amazing quality - brilliant - keep up the good work!

  • Comment number 46.

    All this user's posts have been removed.Why?

  • Comment number 47.

    Thanks Jeremy!

    emailheaven: because it's a test of new technology we're only able to operate it during regular office hours - we need to monitor it manually. Sorry about that. We'd have liked to have run it in the evenings to find out how well it worked on people's home broadband but it just wasn't possible.

  • Comment number 48.

    I really hope you make iPlayer available for use with HTML5 using webm or h264 that way slower pc's can playback the video stream smoothly and with low cpu usage. Even on my i5 laptop flash videos give me jerks now and again and flash is poorly coded.

    Can't you have the bbc's own shows and uk production companies sign a contract to let the bbc use html5 which doesn't support copy protection yet? UK tv shows are already on torrents with minutes/hours after airing so the whole copy protection is a poor argument.

    Or you could allow users who have silverlight installed to use that as the default player. Silverlight has been perfectly smooth for me in 720p since version 1 in 2007 i believe. Itv player used that but not enough users had silverlight installed so they switched to flash, but now i believe silverlight is installed on 60% of pc's as of around 1yr ago so creating a silverlight version would be of huge value to us.

    If the trials go well will 720p be allowed for all shows airing on bbc1hd/bbchd? I hope you allow a bitrate of 1200 soon for streaming 640x360 or whatever you are currently using, 800 is severely lacking in detail, especially for tennis and formula 1.

  • Comment number 49.

    @Andy: Thanks for the reply. I definitely support what you're doing here - I was just having a general rant ;)

    Yeah, my media PC running MythTV has a 3GHz P4, Mythbuntu 11.04, 1GiB RAM, and a horrible, horrible ATi GPU. It runs iPlayer passably well in original size but goes very jerky for about 2 seconds about every 5 seconds when in fullscreen mode, so obviously video scaling is a bit messed up in Flash (no such problems when the video is downloaded and played in a decent media player like mplayer or the MythTV one). I have a feeling it'd be better with an NVIDIA GPU, but it's PCI/AGP only so it's difficult to find newer ones.

  • Comment number 50.

    hajj_3: We're very interested in both Silverlight and HTML 5. As the architecture of our player evolves it should become easier for us to support a range of technologies using more shared components than would currently be possible. It's never a good idea to commit to specific technologies that we may support though: consider that two years ago the iPad was but a rumour - now it's a major platform.

    I'm afraid I don't know enough about the rights part of the organisation to give you a sensible answer on the contract question - but I'll see if I can track down a response for you.

    I also can't be too specific about our future plans in relation to resolutions and bit rates - but we're always looking to push the quality up and we recognise the shortcomings of the current encodings for live action.

    Muzer0: Thanks for the support. It's greatly appreciated. And yes, when the GPU offload actually works it makes a huge difference - as you'd hope.

  • Comment number 51.

    Do you have any plans to use MPEG DASH? I know it's not an approved international standard yet but it's likely to get much wider industry support than the proprietary Apple system. From a server point of view, it would "just" need a different manifest to be provided.

  • Comment number 52.

    Why9: We're not using Apple's system (HLS) for this trial (we're using Adobe HDS) but as far as I know HLS is on the IETF standards track.

    DASH is certainly interesting and we'll monitor its progress. At this stage we're taking a more abstracted view of HTTP streaming. The various approaches to HTTP streaming are conceptually similar and it may be that in the medium term we have to support more than one protocol.

  • Comment number 53.

    Andy Armstrong: I've not seen this stuttering on HD streams (YouTube or Brightcove HD videos) before, but I've had regular incidents where BBC News videos play for the first 2 seconds, then pause before playing through completely. I'm not sure if it's a bandwidth issue, but it was on our university's high speed internet connection.

    It would be interesting to see if using QuickTime for playback behaves differently to Flash playback. Is it possible to pull the same video fragments from the same server but send them to different software clients? I'm assuming of course that you're encoding it as H.264?

    BTW, I'm a web developer and the computer I'm using belongs to my employers. I smiled sweetly and asked for it really nicely! I'm sure Auntie Beeb would similarly oblige if you put forward a case that you needed a good one for "testing purposes".

  • Comment number 54.

    ""We'd have liked to have run it in the evenings to find out how well it worked on people's home broadband but it just wasn't possible""

    Further to my earlier posts,I was at home. I'm on 15mb broadband using the laptop wirelessly getting a consistent 13.5mb. The HD symbol went to 5 immediately then stayed on HD for the duration. I had no issues such as flicker, erractic stream or anything else. Mega impressive have to say.

  • Comment number 55.

    David Bailey: Thanks for the advice about getting a new computer. I'll ask nicely :)

    As I mentioned above we're definitely interested in using the same media fragments with multiple http streaming protocols and we've been testing various such combinations internally. For this trial however we need to concentrate on a single approach.

    But we will be running more trials in the future :)

    Testing_Times: thanks for that - glad to hear it's working well for you.

  • Comment number 56.

    Andy Armstrong: I'll grab screenshots for comparison today. Can I post links on this board?

  • Comment number 57.

    Craig: Thanks for that. I believe you can post links, yes.

  • Comment number 58.

    Andy Street: we've made the bit rate indicator numbers a bit clearer for today - thanks for suggesting it.

    And we're live again. We plan to stay with the BBC One HD feed for most of the afternoon.

  • Comment number 59.

    Craig: no need for screenshots now, thanks - we've managed to reproduce the scaling problem here. Thanks again.

  • Comment number 60.

    Ok I have an HTC desire running androide 2.2 with 800x500 pixel screen.

    This works very well at full screen with Iplayer app or with browser.

    With the test page I start playing in not full screen mode fine. It does take a few seconds to sort it self out and settles at speed 4. If I go to full screen while it is playing I get a still image and a message "Press back key to exit full screen mode". It gets stuck in this mode.

    If I go to full screen with the player not playing and then start it works ok and settles down to speed 4 One small point I get white bars top and bottom because the screen is not 16x9. I usually get black bars.

    Please can you tell me what the resolutions are for each of the bitrates used?. Do you have an Android phone for testing?

  • Comment number 61.

    why are the buttons on iplayer blurry, can't we have crisp maximise button, volume icon, "live" text crisp when in fullscreen. This problem is present in regular iplayer aswell as the HD test.

  • Comment number 62.

    My screen gets to HD fairly quickly. It doesn't always stay there, but degrades gracefully to 4 before building up up HD again. Doesn't happen a lot. I have a VM 10M broadband connection and usually can't stream HD channels, so this is an improvement. I get picture stutter sometimes though.

    I will try again on my netbook running Windows 7, it seemed to work as well or better yesterday.

  • Comment number 63.

    Was ill yesterday (still am really but less so) so missed the excitement then. Anyway:

    Testing on Linux (Ubuntu 11.04 64-bit) with Chrome 12.0.742.100, nVidia geForce 210 with nvidia drivers. Flash hardware accel is on. Internet connection via virgin cable.

    Probably a flash issue: When playing in window on secondary monitor, then clicking to go full-screen, it goes fullscreen on primary monitor.

    Also, when playing fullscreen on one screen, then clicking on a window in another screen (specifically, to start typing this comment, in another window of the same browser), it falls back out of fullscreen.

    Sits quite happily on "4" while playing in window.

    When fullscreen, it steps up to HD within a few seconds and stays there, but it does seem to be dropping frames and showing quite visible pixellation artifacts and occasional tearing. In that respect, playing in-window at "4" is more pleasing. NB: I regularly use the iPlayer add-on for XBMC on my TV, which seems to stream h.264 720p quite acceptably without these problems - but doesn't use Flash, so I suspect your choice of Flash as a testbed may be introducing its own problems wrt Linux especially, as others have noticed.

    Unable to turn off flash hardware accel to try that as flash player settings crashes.

    2nd machine; MacBook Pro with nVidia 9400, Chrome 12.0.742.100, hardware accel in flash on. Same network. In window, stepped up to 4 quite quickly but seemed to have some buffering issues for a while. After a few minutes seems to have settled down.

    Fullscreen playback basically the same as on Linux: Steps up to HD quickly but it's skipping frames to keep up and pixellating on motion (espec. panning across crowd). Not seeing tearing though.

    Tried turning flash hardware accel off, was much worse; had to drop down to 1 on fullscreen and still didn't manage full frame rate. Again, this is probably flash being inefficient, as other media players that don't use hardware accel can keep up h.264 playback better than this (EyeTV playing BBC HD from Freesat at its shiny new full resolution for instance, looks perfect).

    Finally, on the mac, the video controls don't work fullscreen, or when coming back down from fullscreen (which can only be done by hitting escape). Not sure why this should be different in the embedded flash in chrome across browsers but never mind...

    NB: In all cases monitor is Dell U2311H 1080p, 60Hz refresh rate.

    In all, I think most of the problems can probably be ascribed to flash itself. I hope it really is just a testbed platform and the service is going to be available on more open protocols allowing for better implementations of the player-side.

  • Comment number 64.

    Trevor: we haven't done any formal testing on Android for this trial but informally we're trying it on the various Android devices people have. I'll certainly pass your observations on to the media player team, thanks.

    The resolutions are:

    1: 512x288 (416k)
    2: 512x288 (624k)
    3: 640x360 (944k)
    4: 640x360 (1408k)
    5: 1024x576 (2112k)
    6: 1280x720 (3168k)

  • Comment number 65.

    also, the text when you hover over those buttons is blurry in fullscreen too.

    There are quite a few pages that don't work in Firefox 5 or 4.01 maybe even 3.6.18 such as this: https://news.bbc.co.uk/sport1/hi/tennis/9520363.stm

    The video won't play, i just see a progress wheel going round, the video plays fine in IE9. I can't even see the screenshot of the tennis player. Here is a screenshot: https://i55.tinypic.com/23su2qg.jpg

    If i open the page in IE9 then open in firefox sometimes it then works.

  • Comment number 66.

    Hi Andy,

    I watched the stream in HD on my home laptop via wifi(it went to HD in about 10 seconds and remained there for duration) I get generally around 9-10 MB /sec and found the picture to be far superior to normal streaming. There was a bit of motion blur especially the ball, which although not a deal-breaker is distracting. That may have been down to limitations of my equipment and/or wifi connection but found it improved greatly when connected via cable. Impressive start.

  • Comment number 67.

    4-7-11; the seconds on each jump up to the next bitrate. then at 22 seconds level 4 stopped/restarted within a second of two and then stuck nicely at 4.

    Then hitting full screen it took 8 seconds to hit level 5 and another 4 to go to 'HD'.

    Although it's a massive improvement when in the embedded player, I can't say I'm that overwhelmed by the 'HD' quality when fullscreened.

    I'm running 1280x1024 which fair enough I'm not expecting BluRay quality but there is a decent amount of blocking on the furthest away player/umpires by the back wall. You can see it very clearly in the this screenshot; https://fusenn.net/img/BBCHDBlock.PNG

  • Comment number 68.

    Tennisfan: thanks for your feedback. Good username too!

    StrangeNoise: it doesn't sound as if it's particularly satisfactory on either of your machines.

    Do you know which version of the Flash player you're using?

    The business with Flash getting confused about which desktop it's on is, I think, a known issue on Linux - but I've flagged it anyway.

    Sorry that it's not working better for you - but the feedback is extremely useful, thanks.

  • Comment number 69.

    Andy Armstrong: Here's a screenshot of the same match using both players.

    Left side is 'normal' quality, right side is HD

    Note how the HD player shows well defined squared/pixellated images, rather than a 'smooth' image. Also note the pixellated control bar, which is not the same as the normal player.

    I understand that smoothing can add to the CPU load, but given that this is HD, surely that would be acceptable. The pixellation becomes a distraction and it's almost preferable to watch the HD either in windowed form, or watch the low-res version smoothed.


  • Comment number 70.

    Running the player within it's own tab; https://www.bbc.co.uk/test/zaphod/emp/emp_files/516516_517751_emp.swf?playlist=https://www.bbc.co.uk/test/zaphod/playlist/wimbledon_bbc1_hd.xml actually gives a nicer experience on HD than using proper fullscreen.

    Wow, comparing these two screenshots;

    The size difference if barely visable between the two actual streams yet the in-browser one is far less pixelated.

  • Comment number 71.

    in chrome 13 beta i can watch the stream with Level 4 in non-fullscreen mode but in IE9 and firefox i can only watch in Level 2. The size of the window is much bigger in chrome 13 than in IE9 or firefox, can't we have the window the same size for all browsers?

    Chrome 13 often won't play the video, the time ticks showing that its playing but no video or sound, refreshing the page sometimes fixes it. The video works fine in IE9 and firefox 5.

    I like to be able to watch in windowed mode sometimes whilst i do other things on my computer, level 2 is unwatchable, please make it level 3 or preferably level 4 for windowed mode for all browsers.

    Even the image on the test page before you press the play button is blurry in IE9 even though it is the same resolution as chrome

  • Comment number 72.

    p.s wouldn't we get better picture quality with 720x400 at 1400bitrate instead of 640x360?

  • Comment number 73.

    Ok, my 17yr old 'steam driven' desktop pc don't want to know. Not surprised really, tis somewhat dated. (the win 95 logo on the case says it all really)
    On the laptop nothing new to report. Goes straight to HD, no issuses. Sorry if thats boring but there you go.

  • Comment number 74.

    it's the version of flash which I believe comes embedded in Chrome, so presumably not necessarily the one the system has. about:plugins reports 10.3 r181 on both Linux and OS X. It's a thought actually, I'll give Firefox a go in a minute too, to use the system flash.

    "in a minute"... On same Linux as above, Firefox 5.0; about:plugins reports flash 10.3 r181 (as well):

    Basically (and unsurprisingly) identical to as reported with Chrome earlier

    On same Mac as above, Firefox 5.0; about:plugins reports flash 10.1 r102:

    Interestingly a bit different: On fullscreen it steps up to HD quite quickly but I'm not convinced it really is: Image actually looks quite low resolution. However, frame rate is better. Presumably if it says HD in the corner there's just one stream I should be getting from the server, but it seems the two versions of Flash have quite different behaviours in response to pressure. Hm. Time for some screengrabs (filenames hopefully self-explanatory):



  • Comment number 75.

    Thanks. I do get some aliasing on the lines because of the resolution. If you could change speed 4 to 800x448 that would be great. I do alot of video encoding for the Android and I get very good pictures at 400 kb/s but I am using x264 for encoding which is miles ahead of any other software encoder. I have always thought it would be an interesting project to put x264 into hardware for real time encoding.

    Do any of your people have a Samsung Galaxy S2 as it has an hdmi output which could be connected to an HD monitor!!!!

    Well my droid has been running for over an hour now stable on speed 4

  • Comment number 76.

    On being reminded by those screenshots that at the last stage I'm also upscaling to 1080p, I tried changing screenmode on the mac down to exactly 1280x720. So no upscaling taking place, except in hardware in the monitor itself.

    In both cases the actual image was the same - ie: obviously degraded on flash 10.1r102, fine on 10.3r181 - but on the chrome/flash10.3 case, the frame rate issue does seem much improved.

    So once again I think we're really seeing the shortcomings of flash as the delivery platform. :-)

    (haven't tried same test on linux as it's a bit more painful to change resolutions there!)

  • Comment number 77.

    Works very well for me in Chrome 12.0.742.100 and Firefox 5.0, including running the video in both at once (quality 4 in one and HD in the other) on a 4.5Mbit/s line.

    Slight hesitation/stutter as it moves from 4 to 5 and then to HD, the process only taking a few seconds. Quality is great - it feels appreciably better than SD broadcast TV.

  • Comment number 78.

    Testing_Times: thank you. In spite of what I said above about wanting to be on every device I think we might make an exception for your teenage PC.

    StrangeNoise: huge difference between your two screen grabs. It says HD in the corner which means that it's the same video in both cases - I don't think there's any way we can be burning the HD icon into any other video. The only difference must be the upscaling method that Flash is using - but, like you, I'm surprised at difference. Passed to the player team and thanks for taking the time to capture the grabs.

    Trevor: thanks for that. Yes, x264 rocks - we like it here too but we're not using it for this trial.

    We'll be tweaking the resolutions after this trial and also looking at the player's upscaling behaviour.

    I forgot that the S2 has HDMI out - thanks for the lead - I'll go and see if I can find one.

  • Comment number 79.

    Interesting blog. The technologies you talk about imply using server->receiver unicast streams. Do you use multicast for any of your services?

    It seems that multicast is an obvious architectural advantage, provided you can get support from the network and have no issue with who receives the stream.

    I can see why so many streams break up / fail during busy periods if everything is still unicast; a terrible waste of network resources!

  • Comment number 80.

    Absolutely phenomenal quality; I really notice the difference. I'm running windows 7 with chrome and BT fibre optic which averages about 30-36 mps. It is pretty rock solid too. Occasionally, it slows for a second and then catches straight up.

    Many congratulations...

  • Comment number 81.

    Thanks all for your screen grabs and updates. I'm going into a meeting soon so I'm afraid don't have to time to reply to each of you individually but I'll pass them on to the media player team and I'll be back later.

  • Comment number 82.

    ""we might make an exception for your teenage PC""
    It behaves at times like a teenager,throws a strop when I ask it to do something.:-)
    Gonna be a traitor now, off to watch some cricket. Cheerio, have fun.

  • Comment number 83.

    Andy, to reiterate, the firefox/flash10.1 picture quality when reporting HD is no better than that screenshot when played in a 1280x720 screen mode - so I don't think that's an upscaling problem.

    I think the flash upscaling overhead is more implicated in the poor frame rate I see when playing fullscreen on chrome/10.3 (and Linux).

  • Comment number 84.

    Oh, and the Linux system is a quad core i7 3.07GHz with 12GB RAM and a SATA3 SSD system drive. Flash really has no excuses here. :-)

  • Comment number 85.

    Andy Armstrong: To follow up from my screenshot, I have the following observations:

    • Manually resizing the Flash player to 1280*720, or viewing in its own tab, gives a very crisp image (1:1) when Windowed.
    • Full screen from this resolution still exhibits blockiness
    • Enabling 'Hardware Acceleration' (which I think is using Flash's new GPU-decoding and scaling) provides smooth (perfect) video

  • Comment number 86.

    Is there any chance at some point you could make a downloadable version available of 2-3 minutes at HD quality for those of us on slower connections to see what it's like? The maximum I can get is level 4, which looks great, but still struggles with mixes (particularly the wide angle shots of the crowd during breaks in play) - I'm interested to know if that problem is eliminated altogether at the highest bitrate.

    Also, no-one seems to have mentioned sound.. Are there any variations in bitrate or is constant for all the encoding levels?

  • Comment number 87.

    You can hardly call this live I just timed a 3min 47 second delay. This was timed against BBC 1 on satellite which already has significant delays. So all this buffering is the disadvantage of this system.

    I think that some users would like to be able to set upper limit on the bitrate if they are using 3G or have a limited broadband use. It would be nice for the resolution to be a close match to the display.

    #79 Mentions multicasting. I am very suprized the BBC has not pushed this further. Sometime ago the BBC did do some experiments but is has gone very quiet.

  • Comment number 88.

    Viewing on Windows 7 laptop 4Gb RAM, Google Chrome, 1280x800 screen, over pitiful 3Mbps connection (its a bad day!). Took 90 seconds or so to settle down, blocking, buffering and compression errors,stayed on 2, then 3, back to 2, but once it settled, got a constant 4 on the icon. Over a 10 minute period, 2-3 "ghosting" effects occurred, when there was a cut from a close-up shot to a long shot. It doesn't like mixes very much. On a shot zooming out from the moving crowd up to the blimpcam, it lost definition for a second, dropped to 2, then recovered. Overall, given the connection speed, works well. Had no time to put it on the big screen today unfortunately.Good luck with the rest of the testing.

  • Comment number 89.

    This comment was removed because the moderators found it broke the house rules. Explain.

  • Comment number 90.

    StrangeNoise: thanks for the clarification - yes I agree - Flash seems to be implicated.

    Craig: I think your issue really *is* hardware scaling.

    William: I'm afraid in this case we can't make a sample video available to download - we don't have the rights to do that. You make a good point though - in general if people are kind enough to help us test this stuff it would be useful to be able to provide a 'this is what you should see' reference video. I'll look into that for future tests.

    I'll get back to you with technical specs on the sound.

    Trevor: there is certainly some delay due to buffering. Typically we're seeing 10-20 seconds - usually closer to 10 than 20. The delay can build up over time - particularly if the player has had to buffer. Had you been watching for some time when you noted the delay?

    We haven't finalised how the player should behave when it finds that it has drifted behind real time. We'd like it to be able to adjust its frame rate so that it can stay close to live - but that isn't implemented for this test.

    You'll be glad to hear that we definitely don't consider 3:47 behind the live action to be acceptable...

  • Comment number 91.

    pj1024: it sounds as if your network presents us with the kind of conditions that should be handled well by adaptive bit rate streaming. However we know that the current version of the bit rate switching is not as stable or smart as we'd like it to be. The symptoms you describe sound like an example of that. It's very valuable for us to know that it happens in the wild, thanks.

    I hope that if you're able to join us for a future test of http adaptive streaming you'll see an improvement in that area.

  • Comment number 92.

    Andy Armstrong: The symptoms I am seeing are identical to those of fusen and StrangeNoise. I am a Flash developer, and see similar issues with normal video streams. The playback component isn't using '.smoothing=true', thus 'nearest-neighbour' is used, causing the blockiness when rendering at resolutions greater than 1280x720.

    When hardware acceleration is enabled, the decoding and rendering is delegated to the GPU, so this type of blockiness will not occur.

  • Comment number 93.

    @Craig; just to confirm/reiterate, I *am* using hardware accel - without it, it was much worse on the mac, and I was unable to turn it off on Linux as flash itself went unresponsive when I tried until page reload (when it showed on again).

  • Comment number 94.

    Worked fine for me this afternoon, alternating between 5 and HD on my 3.1Mb/s max line.

  • Comment number 95.

    Worked tremendously well for me. Took a few seconds for the HD icon to come up and then the picture was brilliant. Kept me away from my work for half an hour, though :)

    I think this is superb and can't wait for other programmes to be broadcast this way. (Will you be doing the same for the Open coverage in July?)

  • Comment number 96.

    Craig / StrangeNoise: I'm well out of my limited Flash depth now - I'll make sure the player team see your comments. Very helpful.

    Trevor Wright: glad to hear it worked for you. Positive and negative reports are equally important so that we can get a sense of how well it's working overall.

    I know there are a few posts from today that need a response but which I haven't got to yet. I'm going to sign off for the night now but I'll attempt to get to them tomorrow.

    Thanks again!

  • Comment number 97.

    @William: just go on iplayer and watch a show from bbchd as those are also 3200 bitrate 720p. You can download those to your pc and watch using bbc's video player application.

  • Comment number 98.

    @StrangeNoise: There is actually a brilliant binary patch for Flash (on all platforms) to allow it to stay in fullscreen when it loses focus, which makes it better for multi-monitor setups (unfortunately it still goes onto the wrong ones :P).

    Grab it here: https://deve.loping.net/projects/ignoflash/

  • Comment number 99.

    William: we're using the same audio profile for all the bit rates - 96k / 44.1 kHz AAC-LC.

    And we're live again - keep that feedback coming, thanks.

  • Comment number 100.

    Using Chrome (13.0.782.32) on a base-model MacBook Air. It ramps up one quality point every five seconds or so, to stick happily and resolutely on HD (3168kbps, according to the right-click diagnostic information in the EMP).

    While it does it, in fullscreen, there's a consistent jump to a black screen for a few frames between 2 and 3. I've reproduced this three times now, so thought I'd report it.

    Not sure that this hardware shows every frame while in fullscreen HD; but that's something quite different!


Page 1 of 3

More from this blog...

BBC © 2014 The BBC is not responsible for the content of external sites. Read more.

This page is best viewed in an up-to-date web browser with style sheets (CSS) enabled. While you will be able to view the content of this page in your current browser, you will not be able to get the full visual experience. Please consider upgrading your browser software or enabling style sheets (CSS) if you are able to do so.