TVDrive crushing whites?

Discussion in 'Cable TV & Virgin Media TV' started by cyberheater, Mar 18, 2006.

  1. cyberheater

    cyberheater
    Well-known Member

    Joined:
    Apr 3, 2003
    Messages:
    3,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Location:
    Scotland
    Ratings:
    +189
    I've noticed on my TV drive that it seems to be crushing whites. No contrast adjustment on my TV affects the crushing so I know it must be at source.

    Is yours the same?
     
  2. meltonboy

    meltonboy
    Well-known Member

    Joined:
    May 31, 2005
    Messages:
    3,144
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    London
    Ratings:
    +582
    sorry for being stupid, but can you explain "crushing" and i'll look out as well :)
     
  3. cyberheater

    cyberheater
    Well-known Member

    Joined:
    Apr 3, 2003
    Messages:
    3,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Location:
    Scotland
    Ratings:
    +189
    An example would be the tops of bald mens head where the light bounces of it.
    You should be able to see a small circle of bright white but i'm getting a kind of white bloom effect.

    Imagine a grey scale on your TV. You should be able to see black at the bottom end and white at the top end but I'm suspecting that the TVdrive is forcing the top 5-10% of lightest grey into full white which means your crushing the grey scale.
     
  4. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    Crushing is usually used when describing black rather than white level detail loss - effectively the detail of the picture is "crushed" below black level, and thus no longer visible.

    White level detail loss can be described as clipping - though this is a particular aspect of it.

    The most usual causes of detail loss in the black and white levels of purely digital signals is a clash between "broadcast" and "PC" video levels.

    In broadcast standards (in 8 bit terms) - black is 16, and white 235.
    In PC terms - black is 0 and white 255.

    (The same is true for RGB representation in broadcast and PC circles. In broadcast areas Cr and Cb are 16-240 centred around 127 AIUI)

    This can cause all sorts of problems - especially if you feed PC video into a DVI/HDMI input expecting broadcast video, and vice versa.

    NB I may be +/- 1 on these values - I'm slightly rusty.
     
  5. antfuller

    antfuller
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    i wonder if this has happened to my tv? i have an evesham 27" HD ready(??) lcd tv with just a d sub which is connected to my XBOX360. All of a sudden the screen has gone really bright! No matter how many times i AUTO ADJUST the picture(cos it wont allow me to do it manually) it doesnt work just flashes between colours. Is it reversible?? I have only had the tv for a week!:mad:
     
  6. cyberheater

    cyberheater
    Well-known Member

    Joined:
    Apr 3, 2003
    Messages:
    3,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Location:
    Scotland
    Ratings:
    +189
    Thanks Stephen. I think your right. It's probably more accurate to call it white clipping as oppose to white crush, it's all the same in my head ;). I'm also aware of the PC vs broadcast values but this is a consumer box sending out (via hdmi) a clipped white signal. It shouldn't be so.

    I'm going to try normal scart tomorrow and if the whites are not clipped then i'm raising a bug with TW.
     
  7. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    Interesting that it is doing it via HDMI -wonder if the encoder/source is faultless...

    Suspect it is a bug/implementation feature rather than a fault though, if it in the HDMI output. After all you'd expect the HDMI output to have remained digital throughout - so any clipping would be a firmware/software/DSP kind of thing?
     
  8. Go_Deano

    Go_Deano
    Active Member

    Joined:
    Nov 28, 2005
    Messages:
    327
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    31
    Ratings:
    +46
    not a problem on my sagem g4 picture through hdmi is 1st class
     
  9. cyberheater

    cyberheater
    Well-known Member

    Joined:
    Apr 3, 2003
    Messages:
    3,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Location:
    Scotland
    Ratings:
    +189

    That's good to know. I'll check my tv settings again.
     
  10. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,199
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,241
    Just for reference.
    Crushing is when intensity values are mapped onto each other ; "crushed" together. This can happen with any part of the intensity range ( blacks , whites midtones wherever) and resolves itself as flattening and most likely banding/posterisation. Can also occur to individual colour channels.

    Clipping is when you map the available intensity range beyond the limits of your display paradigm ,ie : above your white point and below your black point and can apply to either the blacks or whites and the individual colour channels themselves.

    Video is always clipped and crushed somewhere relative to the original film material ( in the case of film based material of course). A large part of a telecine operators job is effectively to introduce clipping and crushing (relative to the film range) in a way that is least offensive.

    Human beings are more sensitive to crushing and clipping in the darker areas of the image than the whites. Film and video intensity response curves are actually designed to exploit this. (recording more variation towards black than towards white).
     
  11. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    I've never heard of white crush used as a term in broadcast video circles - only black crush. I guess that is because stuff like knees get in the way in cameras, to avoid it. There are also black stretch devices in many cameras to allow more detail to be seen in the blacks. (Crushed blacks are very popular with some entertainment lighting directors...)

    In the days of analogue cameras it was routine to run clippers above 100%, as you could carry video ranges outside the formal standard. However in digital terms, once you go above 8bit 235 you're in dangerous ground, and you can't go over 255 at all...

    I have to say I approach this from a video camera, rather than telecine approach.
     
  12. Chris Muriel

    Chris Muriel
    Well-known Member

    Joined:
    Jun 14, 2002
    Messages:
    7,280
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Manchester
    Ratings:
    +763
    Some early encoders (I am referring to the chips containing the video DACs) failed to add saturation logic ; the result was that it was possible for some DVB-S signals to cause a "wrap around" from fullscale back to zero again.
    This created some problems for one of our key customers (I work for an IC manufacturer). We solved the problem before most of our competitors.

    Chris Muriel, Manchester
     
  13. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    By encoders you are referring to post-MPEG2 decoding YCrCb to RGB or Composite/S-video encoding I presume?

    Were the "problematic" signals illegal?
     
  14. Chris Muriel

    Chris Muriel
    Well-known Member

    Joined:
    Jun 14, 2002
    Messages:
    7,280
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Manchester
    Ratings:
    +763
    Correct.
    CCIR656 input, RGB, Cvbs or YPrPb output.
    The offending transmissions were mosaics showing several channels ; if one of the PIP'd channels was showing teletext then it was possible to go from black to white on 2 adjacent pixels. This caused problems on RGB output until we put in the saturation logic.

    Chris Muriel, Manchester
     
  15. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    Got you - so two adjacent pixels with black and white samples shared the same colour difference samples (because of the chroma sub sampling)

    Did this cause problems if the sources were legal - or just if the blacks and whites involved were <16 or >235? (Being lazy and not doing the maths myself)

    I guess in theory, teletext (like BBC Micro Mode 7), was 8 primary and secondary colours. If the colours were 16->235 RGB rather than 0->255 RGB there would be less of a problem?
     
  16. Chris Muriel

    Chris Muriel
    Well-known Member

    Joined:
    Jun 14, 2002
    Messages:
    7,280
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Manchester
    Ratings:
    +763
    The problem channels/mosaics were actually officially illegal - but we (and Pace, where I was involved) amongst others decided that it wasn't pratical to tell STB end users that they should complain to the channel operators/providers.
    Illegal streams, technically not fully DVB compliant, are a real pain to STB manufacturers and the ensuing food chain.
    The STB manufacturer has to first analyse the problem to find what's wrong - and very often the problem seen by Joe Public isn't his fault.

    Chris Muriel, Manchester
     
  17. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    Yep - I know most studios have legallisers late in the SDI chain - but edit suites, especially low-cost PC/Mac based stuff may not. Also, I suspect the mosaics are derived from quite low cost multi-screen devices...

    Does MPEG2 deliver illegal in, illegal out, or is it worse than this? (Or is it more of a problem if you assume all sources are legal?)

    There is also the issue that <16 and >235 values should be preserved in some circumstances (particularly in production), as they are there to avoid clipping of over/undershoot effects on sharp transitions, which would cause ringing AIUI. (Particularly an issue when digital component gear was used in analogue production environments - especially composite ones)
     
  18. Chris Muriel

    Chris Muriel
    Well-known Member

    Joined:
    Jun 14, 2002
    Messages:
    7,280
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Manchester
    Ratings:
    +763
    I would have to check the current version of the DVB/MPEG2 and CCIR656 specs to know what they currently state.
    Our original encoder (DENC) design assumed legal parameters only - which is why we had to add the saturation logic after the event. Studios would tend to use higher quality (and more expensive) encoders with bigger (12 bit rather than 8/10 bit) video DACs in them. Their equipment would probably use ASI interfaces etc. and longer coding words so I would hope they have more flexibility.

    Chris Muriel, Manchester
     
  19. Stephen Neal

    Stephen Neal
    Well-known Member

    Joined:
    Mar 29, 2003
    Messages:
    6,595
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +920
    I think the only DACs in use in studios in the programme chain are in the camera head prior to the analogue triax cabling between camera head and CCU. The CCU will have ADCs and the signal will remain digital throughout the studio and station operation. It shouldn't hit analogue again until it hits the DACs in the off-air receiver at home. Any other sources, like VTRs, graphics etc. won't have DACs in the programme chain at all, and camcorders recording digitally won't either. (It is only the triax cabling for studio cameras that is still an analogue path.) Otherwise the only analogue stage is the sampling of the camera CCD.

    (The only other DACs in a studio would be for feeding preview monitoring - but this isn't in the programme chain)

    Within studio infrastructure the interconnects will be CCIR 601 carried usually in SDI (HD-SDI for HD), which may itself be carried over a fibre connection. The MPEG2 encoding will take place massively down stream of the studio - downstream of the presentation suite where the chanel is played out - unless the studio is in a remote location using MPEG2 connections to return it to base.

    AIUI where ASI routing IS used, is for remote satellite contributions. AIUI when the BBC do an interview with a sat-truck in the US, they uplink in MPEG2 (480/60i), downlink this in the US (Washington I believe), but keep it in ASI format (i.e. it remains MPEG2 480/60i). The ASI is then routed across the Atlantic via a fibre circuit, where it arrives at TV Centre in London, and the ASI is then decoded to SDI 480/60i and standards converted to 576/50i, still in SDI format, and routed to the studio or edit suite required. This means only a single MPEG2 encode/decode process in the route, which minimises quality loss.
     
  20. Chris Muriel

    Chris Muriel
    Well-known Member

    Joined:
    Jun 14, 2002
    Messages:
    7,280
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Manchester
    Ratings:
    +763
    Sounds like the signal chain is quite sensible and logical then.
    This is good to hear.

    Chris Muriel, Manchester
     
  21. GrahamMG

    GrahamMG
    Well-known Member

    Joined:
    May 8, 2004
    Messages:
    4,459
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Location:
    UK
    Ratings:
    +194
    Could the fella that asked the original question just have a bit of a DC offset?? Reaching for the "caps" (or diodes in a blunt fashion) might cure it?
     

Share This Page

Loading...