A rant against the trend to blaim EVERY fault in a DVD transfer on low bitrate

Discussion in 'Archived Threads 2001-2004' started by Bjoern Roy, Dec 1, 2001.

  1. Bjoern Roy

    Bjoern Roy Second Unit

    Joined:
    Oct 15, 1998
    Messages:
    315
    Likes Received:
    0
    Trophy Points:
    0
    In this thread, I raised the issue of bitrate fetishism. I couldn't help it and spawned a full fletched rant. [​IMG]
    [rant on]
    This post is directed at the recent trend in the community to blame everythin on bitrate:
    Title has too much EE? "Bitrate must have been too low."
    Title too dark, bad shadow detail? "Higher bitrate, less extras, would have improved this."
    Picture soft and filtered? "Why couldn't they have spread the transfer over 2 discs..."
    Muted pale colors, too dark, bad color balance? "Oh, if this only was a superbit transfer."
    Same in the audio domain: Soundtrack has no ultra low bass extension? "If only it would have been full bitrate DTS."
    Give me a break. [​IMG]
    Just read some of the recent threads and you will understand what i mean. After the superbit titles, some people seem to think that higher bitrate alone is a cure to everything.
    Everything else being equal, higher video bitrate does mainly one thing: Get rid of compression artefacts (blocking, mosquito noise etc). If the bitrate is very very low, detail can suffer too, but not nearly as much as it already has through too much pre-filtering.
    People seem to think the line of reasoning is this:
    Bad transfer in whatever way (contrast, colors, soft, EE...) -> increase bitrate -> perfect
    Eh, no!
    Lets say a transfer has 3.5Mbit average bitrate. And the characteristics of the transfer might be:
    - Picture is a little soft, misses fine detail
    - Shadow detail/delineation is not perfect
    - Black level a bit too hot (10-15IRE instead of 7.5)
    - Colors a bit dull and muted
    - too much EE
    - no compression artefacts, only a few in darker scenes
    Sounds like the average Col/Tri transfer, right?
    Now, you want to make a more detailed version of this transfer. For that, you would have to prefilter the picture to a lesser degree. The result is a sharper picture, the softness is gone, hooray. But now your picture shows lots of compression artefacts. Why? Because the 3.5Mbit bitrate is not enough to tame the additional information/detail without artefacts. So in turn, you need to raise the bitrate until those compression artefacts are reduced to an acceptable level again. You might end up at 7-8Mbit average.
    So you don't raise the detail BY increasing the bitrate. Its the other way around. You increase detail (by filtering less) and then you NEED to increase the bitrate to get compression artefacts back to an acceptable level. If you would simply increase the bitrate without changing anything else, the transfer would look the same.
    Note, that all other characteristic of the 3.5Mbit transfer above (colors, shadows, EE) are NOT really affected through this process!
    So now you have a 7-8Mbit transfer that has these characteristics:
    - Picture now very detailed
    - Shadow detail/delineation still not perfect
    - Black level still too hot
    - Colors still a bit dull and muted
    - still too much EE, although the different filtering might have changed its characteristic (thinner halos etc.)
    - still a few compression artefacts, probably even more than the 3.5Mbit version
    Air Force One fits this scenario perfectly. The upper being the normal release, the lower being the superbit release.
    And the Superbit version does indeed have more compression artefacts in some scenes than the original release. If you understand what i wrote, this shouldn't be a surprise.
    On the normal filtered, soft non-SB release of a particular title, a 3.5Mbit average bitrate might have been enough to tame compression artefacts. When the SB version runs the detail full throttle, a 10Mbit average bitrate might be necessary to avoid compression artefacts. If you can only spent 7.5Mbit, you will end up with a sharper picture, but more artefacts.
    Take a look at Harrison Fords first speak in Air Force One to see a Superbit scene that is highly detailed, but exhibits compression artefacts.
    [rant off]
    Regards
    Bjoern
     
  2. PerryD

    PerryD Supporting Actor

    Joined:
    Aug 28, 2000
    Messages:
    737
    Likes Received:
    0
    Trophy Points:
    0
    I have to admit that I'm guilty of this ever since the release of the Superbits. I look at a recent release like Shrek which contains 6 seperate soundtracks (DTS, 3 DD5.1, DD2.0, and commentary track), and think about how much better the picture may have looked if it included only say a DTS and DD2.0 soundtracks.
    I believe Unbreakable was another title that seemed to have an overly soft picture, and by coincidence contained 4 5.1 soundtracks IIRC.
     
  3. Vince Maskeeper

    Vince Maskeeper Producer

    Joined:
    Jan 18, 1999
    Messages:
    6,499
    Likes Received:
    0
    Trophy Points:
    0
    I'm certainly not an expert on video, but I have always gotten the impression that the issues of filtering the video signal were a direct result of a desire to eventually encode at a lower bitrate.

    I always followed the logic as unfiltered video would cause problems for authoring stage, and by filtering some of the content, lower bit rates could be used without visible (or as visible) artifacting.

    If that is the case, the idea of many of these video issues would be, as a byproduct, the result of lower bitrate encodings.

    Edge enhnacement seems to have a similar logic... EE or simply "enhancement" as I believe it is called in the techincal portion of the telcine process, seems to be an attempt to substitue for clean edge lines on filtered video material.

    Again- I could be smoking crack- but I would assume that there was a reason for the filtering of video signal-- one could assume all studios- especially honest to god professionals would know as much as a poster like Bjoern and I find it impossible to believe there isn't some sort of logic (maybe not one I agreee with but..) behind the decidions they make.

    Since it seems to take more effort to apply filtering to high end video scontent, I have always assumed that the filtering of video content made compression easier- and this would again say bitrates and compression rates were involved, if even indirectly.

    -Vince
     
  4. Mark Zimmer

    Mark Zimmer Producer

    Joined:
    Jun 30, 1997
    Messages:
    4,301
    Likes Received:
    42
    Trophy Points:
    1,610
    I don't dispute what you're saying, Bjoern, but the fact remains that it takes more bits to get that better picture. You just can't do it at a meager 2 Mpbs. You'll either have no detail, or you'll have artifacts galore. Bits may not be the beall and endall, but you aren't going to improve the picture without them.
     
  5. GregK

    GregK Screenwriter

    Joined:
    Nov 22, 2000
    Messages:
    1,025
    Likes Received:
    255
    Trophy Points:
    1,610
    I'm also no DVD compressionist, but when you have multiple
    5.1 DD soundtracks along with DTS. I still think
    something will have to give picture-wise, unless the
    source was crap to begin with. Compressionist either can:
    (A) Try to keep all the detail and know the MPEG-2 encoder
    will muck up here and there, in the form of artifacts.
    (B) Or a compressionist can "filter" high frequency video
    info (by ever so slightly softening the picture). There's
    no artifacts, and most reviewers think the end result is
    great. This (IMHO) is what I think happened to Shrek.
     
  6. DaViD Boulet

    DaViD Boulet Lead Actor

    Joined:
    Feb 24, 1999
    Messages:
    8,800
    Likes Received:
    3
    Trophy Points:
    0
    I for one would love to see a "super bit" Shrek with no prefiltering...just perfect downsampling from the hi-res images to 720x480, a DTS track and one 5.1 English track.

    Computer animation should be the easiest to render *perfectly* onto the DVD medium as no extraneous noise enters in to the image to confound compression any further (no film grain or shutter movement etc).

    I thought Shrek looked pretty soft to my eyes. Not quite the reference level of Dinosaur.

    -dave
     
  7. Ed St. Clair

    Ed St. Clair Producer

    Joined:
    May 7, 2001
    Messages:
    3,320
    Likes Received:
    0
    Trophy Points:
    0
    Bjoern,

    Well done on how a higher bit rate improves a transfer.

    How does a lower bit rate 'improve' a transfer?
     
  8. DaViD Boulet

    DaViD Boulet Lead Actor

    Joined:
    Feb 24, 1999
    Messages:
    8,800
    Likes Received:
    3
    Trophy Points:
    0
    wasn't his point actually how a higher bit-rate does not, in itself, improve image quality?

    But rather how it's less filtering that improves detail...which then *requires* a higher bit-rate in order to conceal MPEG artifacts given the higher complexity of the compressed image.
     
  9. Bjoern Roy

    Bjoern Roy Second Unit

    Joined:
    Oct 15, 1998
    Messages:
    315
    Likes Received:
    0
    Trophy Points:
    0
    Excatly, David.

    Ed, it seems you completely missed my point. Sorry, maybe my explanation wasn't clear enough. I will try to rephrase the bottomline again, but David has it nailed down.
     
  10. ChrisA

    ChrisA Second Unit

    Joined:
    Nov 25, 1999
    Messages:
    478
    Likes Received:
    0
    Trophy Points:
    0
    With 28.2 Mb/s BANDWIDTH, now you can have your cake and eat it to!! Woo hooo! Digital VHS isn't some hypothetical device, it is here next week!! Until thye give us the equivalent in optical, EMBRACE the BANDWIDTH... SMOOCH!
     
  11. ChrisA

    ChrisA Second Unit

    Joined:
    Nov 25, 1999
    Messages:
    478
    Likes Received:
    0
    Trophy Points:
    0
    Bjoern,
    I was trying to grab your attention in the D-VHS thread. I meant to ask you:
    Do you know what they were using in the WSR thread from the output of the DVHS player to the Sony G90? Did they convert to 1080p? How did they get an RGBHV signal? For some dumb arse reason the JVC player doesn't output RGBHV is seems. I'm just wondering what is required to achive an RGBHV (negative sync) for my Dwin HD700. Nine inch CRT owners must be drooling, but I'll live [​IMG]
     
  12. Bjoern Roy

    Bjoern Roy Second Unit

    Joined:
    Oct 15, 1998
    Messages:
    315
    Likes Received:
    0
    Trophy Points:
    0
    Chris, since this off topic here, I will answer you question in the other thread.
     
  13. Ed St. Clair

    Ed St. Clair Producer

    Joined:
    May 7, 2001
    Messages:
    3,320
    Likes Received:
    0
    Trophy Points:
    0
     

Share This Page