VideoHelp Forum
+ Reply to Thread
Results 1 to 9 of 9
Thread
  1. Member
    Join Date: Mar 2005
    Location: Melbourne
    Search Comp PM
    Hi guys, I did a search already but couldn't find the specific info I'm after.

    I need some advice about codecs & compositing.

    What I want to do is capture video via my firewire card and then use compositing software (jahshaka), to composite 3d images on to the video footage. Much like they do in the movies.

    What I would like to know is - what codec would be the best to render my 3d models to? If the captured video is DV avi, then is it possible to render my 3d models to the DV codec? If not, would HuffYUV be a better option, or would that mess with the colours a bit?

    Cheers.
    Quote Quote  
  2. Member edDV's Avatar
    Join Date: Mar 2004
    Location: Northern California, USA
    Search Comp PM
    For the programs I use (Premiere and Vegas), uncompressed RGB or YUV would be my choice for overlay graphics. HuffYUV would also work.

    Check a forum that covers jahshaka.
    Recommends: Kiva.org - Loans that change lives.
    http://www.kiva.org/about
    Quote Quote  
  3. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    targa files or tiff ... or if really serious , cineon files with 16bit or floating bit color ...

    otherwise the sony yuv codec or huffyuv are good choices ..

    you would not to use anything less really ..
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  4. Member Cornucopia's Avatar
    Join Date: Oct 2001
    Location: Deep in the Heart of Texas
    Search Comp PM
    You especially DON'T want to use a codec that is a combination of lossy, block-based, and reduced color sampling. But that is just what standard consumer DV is. There will be difficulties specifically with compositing if using a codec of this sort directly.

    You want something with 4:2:2 or 4:4:4 color sampling, if you want to do it right. That's why lots of post houses bump up their DV stuff to DVCPro50 (4:2:2) if they have to composite with that footage.

    Scott
    "When will the rhetorical questions end?!" - George Carlin
    Quote Quote  
  5. Member
    Join Date: Mar 2005
    Location: Melbourne
    Search Comp PM
    Originally Posted by Cornucopia
    lots of post houses bump up their DV stuff to DVCPro50 (4:2:2) if they have to composite with that footage.

    Scott
    Ok, I might do that then. I have Matrox codecs on my comp which has a DVCPro50 option.
    Quote Quote  
  6. Member edDV's Avatar
    Join Date: Mar 2004
    Location: Northern California, USA
    Search Comp PM
    Originally Posted by Cornucopia
    ...
    You want something with 4:2:2 or 4:4:4 color sampling, if you want to do it right. That's why lots of post houses bump up their DV stuff to DVCPro50 (4:2:2) if they have to composite with that footage.

    Scott
    Most compositing programs bump all elements (including DV) up to 4:4:4 RGB or YUV before the compositing filters are applied. You would want your graphics elements at or above these resolutions for best quality. If the graphics elements are rendered to sizes larger than 720x480, they will be scaled down by the compositing software.

    After the composite is created, the program will resample to your specified output format (e.g. 4:1:1 for DV).
    Recommends: Kiva.org - Loans that change lives.
    http://www.kiva.org/about
    Quote Quote  
  7. Member edDV's Avatar
    Join Date: Mar 2004
    Location: Northern California, USA
    Search Comp PM
    Originally Posted by thegooddale
    Originally Posted by Cornucopia
    lots of post houses bump up their DV stuff to DVCPro50 (4:2:2) if they have to composite with that footage.

    Scott
    Ok, I might do that then. I have Matrox codecs on my comp which has a DVCPro50 option.
    No need to do that. Post houses usually convert everything to uncompressed 4:2:2 for local storage on their multichannel SDI servers. DVCPro50 or Digital Betacam are used to save 4:2:2 material to tape in compressed form (approx 2:1).

    Your software will handle these decompression and resizing steps internally. I usually create graphic elements in HD square pixel sizes 1920x1080 (16:9) or 1440x1080 (4:3) and let Premiere or Vegas handle the downscale to 480i or 480p.
    Recommends: Kiva.org - Loans that change lives.
    http://www.kiva.org/about
    Quote Quote  
  8. Member Cornucopia's Avatar
    Join Date: Oct 2001
    Location: Deep in the Heart of Texas
    Search Comp PM
    Maybe I didn't make myself clear. edDV is correct. If you ALREADY have created your output with standard DV codecs, the damage has already pretty much been done. Most compositing apps do apply their blending/keying etc. in an uncompressed and full color sampling space (4:4:4:4 is common).

    If you are starting fresh with source graphics/renders/vector-based media, it is MUCH better to export with the best quality possible.
    I would list the types (in decending order of quality) this way:

    1. RGBA, uncompressed
    2. RGB, uncompressed
    3. RGB, losslessly compressed
    4. YUV, uncompressed
    5. YUV, losslessly compressed
    6. YUV or RGB with very slight lossy compression (avoid if possible)

    Similarly, the color sampling would be:

    1. 4:4:4 (:4)
    2. 4:2:2
    3. 4:1:1 or 4:2:0 (avoid very much if possible)

    And, if your compositor can wrangle it, making use of high bitdepth color is best AFA banding and smooth blending is concerned. Therefore, 16bit/color/pixel is better than 8bit/color/pixel (there are also some inbetween variations: 10bit, 12bit, etc).

    The bump-up thing is a good future-proof archiving procedure for DV-sourced material, but was tangential to what you were asking about.

    Scott
    "When will the rhetorical questions end?!" - George Carlin
    Quote Quote  
  9. Member
    Join Date: Mar 2005
    Location: Melbourne
    Search Comp PM
    Originally Posted by edDV
    Most compositing programs bump all elements (including DV) up to 4:4:4 RGB or YUV before the compositing filters are applied. You would want your graphics elements at or above these resolutions for best quality. If the graphics elements are rendered to sizes larger than 720x480, they will be scaled down by the compositing software.

    After the composite is created, the program will resample to your specified output format (e.g. 4:1:1 for DV).
    Originally Posted by Cornucopia
    Most compositing apps do apply their blending/keying etc. in an uncompressed and full color sampling space (4:4:4:4 is common).

    If you are starting fresh with source graphics/renders/vector-based media, it is MUCH better to export with the best quality possible.
    I would list the types (in decending order of quality) this way:

    1. RGBA, uncompressed
    2. RGB, uncompressed
    3. RGB, losslessly compressed
    4. YUV, uncompressed
    5. YUV, losslessly compressed
    6. YUV or RGB with very slight lossy compression (avoid if possible)

    Similarly, the color sampling would be:

    1. 4:4:4 (:4)
    2. 4:2:2
    3. 4:1:1 or 4:2:0 (avoid very much if possible)

    And, if your compositor can wrangle it, making use of high bitdepth color is best AFA banding and smooth blending is concerned. Therefore, 16bit/color/pixel is better than 8bit/color/pixel (there are also some inbetween variations: 10bit, 12bit, etc).

    The bump-up thing is a good future-proof archiving procedure for DV-sourced material, but was tangential to what you were asking about.

    Scott

    Ok, great. That's what I wanted to know.

    Thanks for all the help guys. This forum is invaluable.
    Quote Quote  



Similar Threads