Something you don't hear very often.

Discussion in 'Projectors, Screens & Video Processors' started by Gordon @ Convergent AV, Dec 26, 2000.

  1. Gordon @ Convergent AV

    Gordon @ Convergent AV
    Distinguished Member AVForums Sponsor

    Joined:
    Jul 13, 2000
    Messages:
    14,216
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Location:
    Living in Surrey, covering UK!
    Ratings:
    +3,071
    Since this forum seems to be a bit quiet these days I thought I'd share some information with you.

    In both demonstrations I've seen of a Snell and Wilcox Interpolater the sound through the Meridian audio system they were using was out of sync with the picture. At the time I put this down to the massive amount of processing that the unit must be doing to create the re-scaled image.

    We have been supplying Focus Enhancements Quadscans for a few months now and have never noticed any video delay caused by its processing. Then we recieved our first Vigatec GVC1280 scaler. Now, it is undoubtedly superior in picture performance to the Quadscan but it also creates a video delay. Having discussed this with Vigatec it seems that it creates a 4 frame delay due to the superior quality of its de-interlacing circuit and scaling. According to them the Faroudja scalers create a 5 frame delay. I presume the Snell and Wilcox is even worse.

    So if you are after an add on video processor for your projector it's worth making sure that your amplifier has the ability to delay the audio to re-synch with the picture! We use a TAG32r for just this purpose, set at 20ms delay.

    As an aside, I've never seen anyone mention this phenomenon on the US web site forums. Perhaps they are so critical of the picture that they don't bother with an audio system to go with it?

    Merry Christmas,

    Gordon

    ------------------
    StereoStereo
    Intelligent Solutions for Intelligent Homes !
     
  2. Ludae

    Ludae
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    This is true, there will always be a delay in the video signal but the duration will differ from device to device depending, of course, on the nature of the processing. The minimum delay you might expect with a 100Hz TV will be at least half a field period and for any de-interlacer one whole field period.

    When it comes to more sophisticated de-interlacing you mught expect two to four field periods and a motion compensated frame rate converter such as Philips Natural Motion processor or the Snell & Wilcox Interpolator you should expect a delay of between 2 and 7 fields, the latter delay being quite rare and most likely only in a broadcast/post production situation where the sound can be easily delayed and sync'd.

    I would guess the Vigatec has a 2 frame (4 field) delay and not 4 frames, and I see no reason for any consumer Faroudja device having a 5 frame delay, most likely 2 or 3 fields depending on the model.

    The DVDO de-interlacers used in many DVD players have the ability to re-sync the digital audio to allow for the three field delay it causes. Other manufacturers chipsets can do this as well. It is not common but is more likely to become so, particularly as motion compensated frame rate conversion becomes more popular.

    I too find it odd that the delay has rarely been an issue with anyone that I have heard of comment on such devices, although it has been mentioned, it has not been dwelt upon in reviews.

    There are well known issues with regards audio delays from some DVD players and within the ADR process of the movies themselves that will be cancelled out by the delay in some video processors. However, there is still the possibility that those processors with longer delays will leave the video lagging behind by a field or two and that should be a disturbing phenomena. The human sensory system is used to seeing the image before hearing/feeling the sound, so I would imagine that some video processor delays could be beyond the threshold of what the human sensory system will tolerate, being the reverse of what is expected. But then, maybe not, judging by the lack of reporting on the subject perhaps some people are more sensitive to such timings than most others.



    [This message has been edited by Ludae (edited 28-12-2000).]
     
  3. Oasis

    Oasis
    Standard Member

    Joined:
    Oct 7, 2000
    Messages:
    28
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    3
    Ratings:
    +1
    Gordon,

    Unfortunatly, your information on the Snell & Wilcox Interpolator Gold and the Merdian demos, where you heard a lip sync error are incorrect - Yes there was an error, but it was not due to the video processing. If you know your Meridian kit you would know that it has continuously adjustable delay and (the actaul cause of the error you heard) that the CD source has, by default a additional 30ms delay to allow it to detect and switch to DTS without pops and bangs.
    On the last demo I am aware of you commenting on, CEDIA UK 2000, there were two Meridian DVDs, a 586.2 running region 1 and an 800 running region 2. the 800 was connected to the CD input, the 586 to the DVD, The 586 had perfect lip sync, the 800 did not because the DTS feature thing was not de-selected by the Meridian chaps who set their gear up, It was fixed for the second day.
    This was pointed out to you on this forum after your original post reviewing the CEDIA show.

    Sorry, Gordon, if this sounds a bit hard - it's not meant to be, honest! I just wanted to point out the error on this single point.

    The actual delay through the Interpolator is 2 1/2 fields plus or minus 1 field. i.e. for PAL sources (field duration 20ms)the delay is nominally 50ms, but could be between 30 and 70ms depending on any time base correction needed. This is actually quite a short delay in scaler terms and is due to the Interpolator using the SWHD1 chip developed by S&W and IBM which does all the de-interlacing, processing and scaling in a single chip, allowing a single frame buffer to be used. Many other scalers use seperate de-interlacing and scaling chips which requires two frame buffers to be used, and therefore longer delays.

    Better processing does not mean necessarily longer delays, in fact it could be the other way round, but it is completley accurate that ANY processor from any manufacturer will have some delay, and it is certainly very important to check that any delay can be compensated for. I also generally recommend 30 to 40ms delay on the audio.

    I hope this helps clear up a few things !

    I can be contacted for more info on the Interpolator if needed at the e-mail address below (once I get back to the office on the 2nd!)

    Merry Xmas one and All,

    Oasis, Otherwise known as

    Richard Ansell
    Senior Product Support & Development Engineer / Product Manager for Interpolator Gold
    Snell & Wilcox

    E-Mail [email protected]
    web www.interpolator.com
     
  4. lmccauley

    lmccauley
    Well-known Member

    Joined:
    Oct 24, 2000
    Messages:
    3,552
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    Newcastle
    Ratings:
    +337
    Gordon,

    Do you know if it's possible to set a delay for the Arcam DAVE (I don't remember there being a setup option to do this)?

    Cheers,
    Liam

    ------------------
    "Do you hear that, Mr. Anderson? That is the sound of inevitability."
     
  5. Gordon @ Convergent AV

    Gordon @ Convergent AV
    Distinguished Member AVForums Sponsor

    Joined:
    Jul 13, 2000
    Messages:
    14,216
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Location:
    Living in Surrey, covering UK!
    Ratings:
    +3,071
    Richard,

    I'm glad that you continue to take part in this forum and I have no problem in letting you correct my errors. I did say that the audio was delayed in comparison to the picture and it was. As before in my comments about CEDIAuk I had been told this was due to the Meridian not having synchronised its audio properly. I had thought that it had just been set incorrectly to compensate for the Snell + Wilcox delay. I had not appreciated it was due to the weird processing dealy of the audio processor. If the reason is due to the Meridian system then fair enough. I understand the 30milli second DTS delay while working out what the signal is so there are no loud cracks through the speakers but why this would be continually present is a mystery to me! I would have thought it would just be an initial "muting" of the audio signal for about a second then everything would be OK.

    You were using a Meridian system in NewOrleans the other time as well. Perhaps you should consider something better ;-)

    Ludae, the Vigatec manufacturer confirms the delay at 4 frames, so a 20ms delay seems correct yes? They also advised that this was superior to Faroudja scalers which they claim are at 5 frames. I too wonder why no-one ever mentions this.

    Liam, You're out of luck with the Dave I'm afraid. The Quadscan will work with your amp and the "TAW Rock" does not show any video delay according to the manufacturer. Of course, you could just go for it and use an Interpolater .........

    I'll check up on more facts from other scaler manufacturers and we can try and compile a hit list here. I'll also see which processors feature a delayable audio.

    All the best,


    Gordon

    ------------------
    StereoStereo
    Intelligent Solutions for Intelligent Homes !

    [This message has been edited by Gordon, StereoStereo (edited 28-12-2000).]
     
  6. Ludae

    Ludae
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Well 20ms is the period of just one field of a 50Hz video system, such a delay would not disturb most people.

    I find the video delay quoted by VigaTec regarding their own device and Faroudja scalers to seem exessive. Maybe they are quoting the delay in the processors locking onto the video sequence and source type, i.e. 3:2 pulldown cadence? Some devices take up to 15 fields before they will lock on, the Faroudja models can take up to 5 fields but usually only on startup, and can indeed detect changes in the source type within a field period under certain circumstances. The Faroudja systems process between 2 and 4 fields within their de-interlacing algorithms depending upon the source type and device model, i.e. interlace video source is more difficult to de-interlace and can require reference to more fields, de-interlacing film source is trival if the original field pairs are reliably detected.

    I would guess that the Vigatec device has up to a 4 field delay when processing interlaced camera material, but may take up to 4 frames to detect a 3:2 or 2:2 cadence for film source but that won't mean that it will delay the output by 4 frames whilst it decides on the source type.

     
  7. Gordon @ Convergent AV

    Gordon @ Convergent AV
    Distinguished Member AVForums Sponsor

    Joined:
    Jul 13, 2000
    Messages:
    14,216
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Location:
    Living in Surrey, covering UK!
    Ratings:
    +3,071
    It all seems a bit strange to me too. I must note that a 20ms delay between audio and video is obvious to me and everyone we've shown it too. In fact, whenever we've done a dem without dialling in the delay the client has asked why the film was so badly dubbed.

    I posted this over at AVS and so far the only person who's replied has noted that they see no video delay with their Vigatec. In fact, by the tone of the post it appears that I may have upset him somewhat.

    I'd love to get to the bottom of this. The manufacturer says it is there and a user says it isn't. Richard says it's there but not of a high enough magnitude to matter on the Interpolater. TAW say that it isn't there on their unit, or at least not so you'd see.


    Anyone out there got a FAROUDJA 3000/5000?

    Gordon

    ------------------
    StereoStereo
    Intelligent Solutions for Intelligent Homes !
     
  8. Oasis

    Oasis
    Standard Member

    Joined:
    Oct 7, 2000
    Messages:
    28
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    3
    Ratings:
    +1
    Gordon,

    Appreciate your comments, thanks.

    One point I'd like to mention - You said the audio was delayed relative to the picture - i.e. too much audio delay - in our demos. That would be correct for the Meridian additional delay!

    The actual problem you comment on is actually where the audio is not delayed enough and is early relative to picture - this is much more noticable to the brain - it seems to expect that sound can be late (due to experience of the difference between the speeds of light and sound) but sound early really can be very obvious. I personally suspect that the combination of the delay in most audio processing and the distance from speakers to ears actually gives just enough delay for the brain to tie the lips and sound together, but not always, hence the need for some delay but not neccesarily as much as 'the book' suggests. It is surprising that more people don't hear the effect, but, equally at long, thin cinemas (like the odean leicester square) then the delays will be different for the entire audience, so there should be the same problem there - I haven't got my brain in gear to do the numbers but I suspect there is a fair bit more than a couple of (video) frames error over the length of the auditorium!

    Interested to hear any other thoughts!

    p.s. The DTS thing was a bit strange- but then it is Meridian after all! (no offence to Meridian - we are good mates after all) We have a long standing relationship with Meridian, but it's certainly not exclusive either way. We are also announcing or SDI output card for the Proceed PMDT DVD player as well as the Meridian 800. You would also have seen Interpolator Gold being used by Denon (Bristol 2000) and TAG McLAren (Live 2000)as the most recent.

    Hope this all helps!!
    Oasis

    [This message has been edited by Oasis (edited 28-12-2000).]
     
  9. Nick Beecham

    Nick Beecham
    Standard Member

    Joined:
    Oct 23, 2000
    Messages:
    132
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    21
    Location:
    Hockley, Essex, UK
    Ratings:
    +2
    Sorry for being thick but you guys have lost me here. I thought PAL was displayed at 24fps, so that a delay of 5 frames would be 5/24 of a second = 208ms. Please could you explain how I've gone so badly wrong.

    Thanks in advance
    Nick Beecham
     

Share This Page

Loading...
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice