Beauty and the Beast & compression artifacts

Discussion in 'DVD' started by John Alderson, Sep 14, 2003.

  1. John Alderson

    John Alderson Supporting Actor

    Joined:
    Aug 8, 2001
    Messages:
    562
    Likes Received:
    0
    Trophy Points:
    0
    I searched the forum and, surprisingly, didn't find a thread specifically about BatB. So I'm starting a new thread [​IMG]

    I have read a lot of talk about the fact that three full versions of the film were crammed onto one DVD, and how that compromised the video with some noticeable artifacts. Does anyone have screenshots of this? I am curious, because personally I haven't seen these imperfections in either the IMAX or theatrical versions (but I haven't watched it THAT critically). I haven't been able to find such screenshots on the web either.

    Something to understand is that compressing cell animation IS different than live action, or even 3D animation. The colors are more likely to match EXACTLY from one frame to the next, therefore requiring a significantly less bitrate to accurately reproduce the original. That is the theory, anyway.

    So what I am wondering is, does BatB suffer from people having unreasonably lowered expecations because they don't think 4.5 hours should be able to fit on one disc (even though animation is a special case), or is there really something there? Perhaps a bit of both?

    At any rate, this type of artifact should show up easily in a screenshot (so long as the screenshot itself isn't compressed).
     
  2. GlennH

    GlennH Cinematographer

    Joined:
    Sep 28, 1998
    Messages:
    2,132
    Likes Received:
    7
    Trophy Points:
    110
    Real Name:
    Glenn
    No, it's not because of lowered expectations due to the amount of content. There is definite mosquito noise artifacts.

    Look in some of the early scenes with the book for example, or the lines on faces.

    Here is a review that includes some screenshots you can click on:

    http://dvd.ign.com/articles/374/374150p1.html

    Click on the pictures of Belle's face and notice the faint lines around her eyebrows and other lines defining her features. Notice all the strange little hashes.

    These aren't artifacts that absolutely ruin the presentation, but once you notice them it's hard not to keep seeing it. This isn't there on other similar animation.
     
  3. John Alderson

    John Alderson Supporting Actor

    Joined:
    Aug 8, 2001
    Messages:
    562
    Likes Received:
    0
    Trophy Points:
    0
    Ok, I'll have to check mine at those scenes. Those look like JPEG compression artifacts to me; those are fairly heavily compressed images. Thanks for the link!
     
  4. MatthewLouwrens

    MatthewLouwrens Producer

    Joined:
    Mar 18, 2003
    Messages:
    3,034
    Likes Received:
    1
    Trophy Points:
    0
    The good thing about the r4 is that the theatrical and special edition are shown through seamless branching. Therefore, there is much less than 4.5 hours on the disc, and therefore no compressiona artifacts.

    Times like this, I'm glad I live in NZ.
     
  5. John_Berger

    John_Berger Cinematographer

    Joined:
    Nov 1, 2001
    Messages:
    2,489
    Likes Received:
    0
    Trophy Points:
    0
    Please, let's not get into the MPEG compression vs. comparative JPEG compression of screen shots bit again. We had enough of that discussion in the official BatB review thread.

    John, the artifacting is without question there on the R1 DVD. My wife and I watched it the day that it came out on our 55" 16:9 Mitsu in progressive scan mode. I was surprised at how apparent the artifacting was, particularly around high-contrast areas such as the black character outline against a bright white or blue background.

    The nature of MPEG compression definitely makes animation the real benefactor of DVD over live-action, so you would think that even with all of the versions on the disc that the video quality would have been outstanding, but alas it could have been better. I should run it through a bit rate analyzer to see what bit rate they used, just out of curiosity.
     
  6. Artur Meinild

    Artur Meinild Screenwriter

    Joined:
    Aug 10, 2000
    Messages:
    1,294
    Likes Received:
    0
    Trophy Points:
    0
     
  7. Dick

    Dick Lead Actor
    Supporter

    Joined:
    May 22, 1999
    Messages:
    6,594
    Likes Received:
    2,662
    Trophy Points:
    9,110
    Location:
    Maine
    Real Name:
    Rick
    The stained glass windows at the beginning of the film are what bother me most in terms of artifacts. Very distracting. The widescreen laser actually looked better during the first several minutes.
     
  8. Mark_vdH

    Mark_vdH Screenwriter

    Joined:
    May 9, 2001
    Messages:
    1,035
    Likes Received:
    0
    Trophy Points:
    0
     
  9. Dan Rudolph

    Dan Rudolph Producer

    Joined:
    Dec 30, 2002
    Messages:
    4,044
    Likes Received:
    0
    Trophy Points:
    110
    Mark, it uses seamless branching starting at the new song, but ending when Belle leaves. The castle was cleaned up during the song, so there are alternate backgrounds with a clean castle until Belle leaves and Beast wrecks up the place again.
     
  10. Ryan Wong

    Ryan Wong Stunt Coordinator

    Joined:
    Oct 6, 2002
    Messages:
    134
    Likes Received:
    0
    Trophy Points:
    0
    Besides "Human Again" and the end credit, I don't see any different between Special Edition and Theatrical Edition. I believe they are using seamless branching.

    Compression artifact didn't bother me much. What bother me is sometime I feel large part of top and buttom has been crop off. (thus the face is too big for me)
     
  11. James Reader

    James Reader Screenwriter

    Joined:
    Mar 10, 2002
    Messages:
    1,465
    Likes Received:
    0
    Trophy Points:
    0
    Let's clear this up:

    All regions of the Beauty and the Beast DVD use EXTENDED branching to show the Special Edition of Theatrical Version of the film.

    You cannot tell it is EXTENDED BRANCHING though as the branching point is at the layer change. So the entire second half of the movie is encoded twice on the disc, once with the musical sequence (and associated animation and background changes) and once without any of these changes (but with WIP angles).

    The WIP is indeed implemented by angles.

    The reason extended branching the whole second half of the movie as opposed to seamless branching the affected sections is due to the inclusion of the WIP version using angles. As both true seamless branching and angles are based on the same bitstream interlacing technology, it would be hard (but not impossible) to author a single title with both alternative angles and seamless branching picture and sound data in a single bitstream.
     
  12. Michael St. Clair

    Joined:
    May 3, 1999
    Messages:
    6,001
    Likes Received:
    0
    Trophy Points:
    0
    The WIP should have been encoded on another disc. I own hundreds of animated discs (Disney, Anime, TV Shows, and more) and this is disc has the worst compression artifacts out of all of them.

    Disney really shit the bed with this one.
     
  13. Malcolm R

    Malcolm R Executive Producer

    Joined:
    Feb 8, 2002
    Messages:
    13,442
    Likes Received:
    1,055
    Trophy Points:
    9,110
    Real Name:
    Malcolm
  14. Artur Meinild

    Artur Meinild Screenwriter

    Joined:
    Aug 10, 2000
    Messages:
    1,294
    Likes Received:
    0
    Trophy Points:
    0
     
  15. Michael St. Clair

    Joined:
    May 3, 1999
    Messages:
    6,001
    Likes Received:
    0
    Trophy Points:
    0
     
  16. James Reader

    James Reader Screenwriter

    Joined:
    Mar 10, 2002
    Messages:
    1,465
    Likes Received:
    0
    Trophy Points:
    0
    The WIP images will be encoded at a low bitrate, and they will not eat into the bitrate available for the 'finished' version of the film as they are interwoven in the bitstream.

    As you know MPEG is a variable bit-rate compression, but when reading from a disc, the speed of the disc's spinning doesn't alter. Instead the player reads 'packets' of data which are then decoded by the processor as needed. Each packet has a small header indentifying it. Typically too these packets are non-sequential on the disc surface, thus granting better error-protection. So reading from a DVD disc is like reading from a computers' hard-disc, rather than a constant non-changing bitstream like a CD. When using angles, unwanted 'packets' are quickly examined via the headers and discarded if unwanted.

    What this means is that while a bitrate of just over 10 is the maximum, the bitstream can have a higher bitrate if some of it is to be discarded before being decoded by the player. So an angle with a video bitrate of 5 doesn't mean that the alternative angle has a maximum of 5 as well. Of course this is only a guide, and the through bitrate is compromised by the presence of multiple angles - but not as much as you would first expect for the reasons I have explained above.

    I doubt the WIP angles had a big effect on the picture quality of the disc, I think that perhaps some of the problems are down to poorer encoding, as the R2 release does have a superior picture, despite being implemented the same as the R1 version. (It also has a different transfer as the R2 version keeps the 'In Association with Silver Screen Partners' credit at the start).
     
  17. Darren Haycock

    Darren Haycock Second Unit

    Joined:
    Nov 13, 2002
    Messages:
    456
    Likes Received:
    0
    Trophy Points:
    0
    I just don't care about the work in progress mode. Maybe some of you would've liked a third disc to flesh everything out, but I would've been just as happy if the work in progress mode was dumped. It's kinda like watching Star Wars I and II and just seeing blue screens everywhere... whoop de doo! (No offense to any of those who really like said mode... )[​IMG]
     
  18. Michael St. Clair

    Joined:
    May 3, 1999
    Messages:
    6,001
    Likes Received:
    0
    Trophy Points:
    0
    James,

    You are correct, it is not linear. The use of alternate angles reduces the peak bitrate for each stream from 9.8mb/s to a max of around 7mb/s, but the latter number actually varies from the inner to outer tracks of the disc.

    This is one permanent angle from start to finish in the case of BatB.

    A peak of 7mb/s is low for an animated feature, especially with multiple audio tracks. The video for BatB peaks at more around 6mb/s, and sadly the picture quality refelects this.



    Darren,

    I love the WIP mode but it shouldn't be on disc 1.


    This largely moot as Disney is likely to never reissue this on regular DVD.
     
  19. James Reader

    James Reader Screenwriter

    Joined:
    Mar 10, 2002
    Messages:
    1,465
    Likes Received:
    0
    Trophy Points:
    0
     
  20. Michael St. Clair

    Joined:
    May 3, 1999
    Messages:
    6,001
    Likes Received:
    0
    Trophy Points:
    0
     

Share This Page