48 bit support in CyberGraphX/Datatypes/...
  • akl
  • Just looking around
    akl
    Posts: 10 from 2004/7/31
    On the AROS-Dev mailing list I've made a proposal to extend the AmigaOS (Clone) API with support - i.e. a corresponding PIX/PBAFMT - for 48/64 bit bitmaps.

    I'd appreciate everyone interested to participate in the discussion thread there - it would be important to sync this API extension among AOS4/MOS and AROS anyway.

    As a short summary - the idea is to add the following:

    16:16:16 RGB (48 bit)
    16:16:16: RGBA (64 bit)

    in addition to the conventional:

    8:8:8 RGB (24 bit)
    8:8:8:8 RGBA (32 bit)

    This is supported already today by PNG and TIFF and required for professional image processing / video editing systems - even though you won't find any conventional framebuffers which directly support it.

    Still it would be helpful, if not the application but the OS would take care how to render/convert this "high data range" bitmap data.
  • »13.04.07 - 13:43
    Profile Visit Website
  • MorphOS Developer
    CISC
    Posts: 619 from 2005/8/27
    From: the land with ...
    Quote:

    This is supported already today by PNG and TIFF and required for professional image processing / video editing systems - even though you won't find any conventional framebuffers which directly support it.


    But exactly because of this what would be the point? There would be no benefit, and only extra overhead involved in conversion back and forth...

    Besides, there's far more interesting floating-point formats which atleast are displayable with most recentish gfxcards .. either way, all these HDR formats are mainly suitable for editing, and as such I can't say I see the big use in OS handling of them, most of the time you are doing highly specialized tasks with them anyway...


    - CISC
  • »13.04.07 - 14:19
    Profile
  • MorphOS Developer
    Krashan
    Posts: 1107 from 2003/6/11
    From: Białystok...
    From the CyberGraphX documentation: Many of this pixel formats are device specific and should not be used, recommended formats are PIXFMT_LUT8, PIXFMT_RGB16, PIXFMT_RGB24 and PIXFMT_ARGB32.

    Then you are wrong supporting RGBA, ARGB is recommended as device independent intermediate format. As Reggae will be an integral part of forthcoming MorphOS 1.5/2.0, it will incrementally replace datatypes, as it gets more decoders. Reggae uses ARGB32 as processing format (so called "common" format), and is prepared to support ARGB64 as well. ARGB64 will appear in Reggae when the first image format supporting 16 bits per component will be added, it is expected to be PNG.
  • »13.04.07 - 14:22
    Profile Visit Website
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Quote:

    Krashan wrote:

    forthcoming MorphOS 1.5/2.0


    Mmmm...
  • »13.04.07 - 15:46
    Profile
  • MorphOS Developer
    Krashan
    Posts: 1107 from 2003/6/11
    From: Białystok...
    BBRV already have announced MorphOS 2.0 here, so why are you surprised? ;-)
  • »13.04.07 - 15:54
    Profile Visit Website
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Quote:

    Krashan wrote:
    BBRV already have announced MorphOS 2.0 here, so why are you surprised? ;-)


    The term "mmm" does not mean surprise, it's more like... er...

    GIMME GIMME!
  • »13.04.07 - 16:18
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    AmigaMancer
    Posts: 265 from 2005/8/25
    48bit??! Does it really makes a difference to the eye from a 32bit picture?
    Amiga 1200 user.
  • »13.04.07 - 17:20
    Profile
  • MorphOS Developer
    Krashan
    Posts: 1107 from 2003/6/11
    From: Białystok...
    48bit??! Does it really makes a difference to the eye from a 32bit picture?

    There is no such thing like 32-bit picture. It is 24-bit one with 8 bits just ignored or used for transparency, but still you have 8 bits per RGB component. And what is the difference between 24 and 48? You can probably see it only on intentionally prepared images (like some slow changing gradients). The difference is in processing, if you do a lot of operations on 24-bit data, rounding errors (and other errors caused by integer arithmetic) accumulate, and the quality degrades. So possibility to process image in 48 bit (+ optional 16-bit transparency) is an advantage, you get better quality, of course at a price of processing speed. For the same reason recording studios process sound in 24 bits (in spite you can't "hear" the difference between 16 and 24 anyway).
  • »13.04.07 - 19:01
    Profile Visit Website
  • Yokemate of Keyboards
    Yokemate of Keyboards
    magnetic
    Posts: 2129 from 2003/3/1
    From: Los Angeles
    Krashan + CISC

    Just wanted to say glad you guys are in the community! Very interesting thread.

    magnetic

    BTW you can call is "1.5" "2.0" "OS5" or "XrayAlphaOmega8" if you want but when the hell is it going to come out?

    :-?
    Pegasos 2 Rev 2B3 w/ Freescale 7447 "G4" @ 1ghz / 1gb Nanya Ram
    Quad Boot: MorphOS 2.7 | Amiga OS4.1 U4 | Ubuntu PPC GNU/Linux | OS X 10.4
  • »13.04.07 - 19:20
    Profile Visit Website
  • DDJ
  • Butterfly
    Butterfly
    DDJ
    Posts: 99 from 2003/2/24
    From: Brasil
    Quote:


    BTW you can call is "1.5" "2.0" "OS5" or "XrayAlphaOmega8" if you want but when the hell is it going to come out?
    :-?


    Let me try... :-D , when its done? :-) .
    Peg2 G4 1Ghz rev. 1.2 - CPU rev 1.2 - Built on 20050808
    OF-version 1.2 (20050808153840)
    Radeon 9250 (9200Pro) 256MB/128bits AGP - OEM PCVIEW (ProLink)
    RAM 1GB - 2x512MB DDR333 CL2 - Kingston HyperX
    40GB ATA100 - Brand ?
  • »13.04.07 - 21:12
    Profile
  • Cocoon
    Cocoon
    srbin
    Posts: 44 from 2006/11/23
    From: Serbia
    Quote:


    DDJ wrote:
    Quote:


    BTW you can call is "1.5" "2.0" "OS5" or "XrayAlphaOmega8" if you want but when the hell is it going to come out?
    :-?


    Let me try... :-D , when its done? :-) .



    Wait wait wait, i have another one:

    Both mos 1.5 and OS4 will come on efika


    hmmmm....
    May the force be with you. Don't let the dark side of PC take over you.
  • »13.04.07 - 23:02
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    diezi7
    Posts: 167 from 2005/7/26
    From: Madrid
    (Both mos 1.5 and OS4 will come on efika)

    Well, then I?ll purchase 1.5 Efikas of course !!
    PowerMac G4 MDD 1,25 dual (Registered)
    Pegasos II G4 (not working)
    Powerbook G4 1.0 15"
    -=_=_=_=_=_=_=_=_=_=_=_=_=_=_=_=_=_=_=
    http://amigalandia.blogspot.com.es/
  • »13.04.07 - 23:16
    Profile Visit Website
  • Order of the Butterfly
    Order of the Butterfly
    Posts: 186 from 2003/10/23
    Quote:


    srbin wrote:
    Quote:


    DDJ wrote:
    Quote:


    BTW you can call is "1.5" "2.0" "OS5" or "XrayAlphaOmega8" if you want but when the hell is it going to come out?
    :-?


    Let me try... :-D , when its done? :-) .



    Wait wait wait, i have another one:

    Both mos 1.5 and OS4 will come on efika


    hmmmm....


    wait .........

    i hope mos 1.5/2.0 will work on x86 machines :P:P

    And now the mantra is completed!

    :-D :-D :-D :-D
    I'm nerdy in the extreme
    And whiter than sour cream

    White&Nerdy 2006 Al Yankovic
  • »13.04.07 - 23:16
    Profile
  • MorphOS Developer
    jacadcaps
    Posts: 3092 from 2003/3/5
    From: Canada
    If you use 48bit internally for processing then yes, it does make a lot of difference. (that's obviously a HDR made out of 3 24bit raws)
  • »16.04.07 - 10:03
    Profile Visit Website
  • Order of the Butterfly
    Order of the Butterfly
    CLS2086
    Posts: 168 from 2005/8/25
    From: West of PARIS
    Actual scanners use 48 bits technology.
    So let see what will happen
    Keep the Faith !
    VG 5000/A1000/500/500+/600/2000/1200PPC-GREX/1200PPC-ATEO/1200+1230-IVFPUSCSI/CD32/Pegasos 1 April1 G3/Pegasos 2 G4/ Ahtlon/K6-3/various funny machines too :-)
    Maybe one day a G4 AmigaONE when they will be debugged and without April...
  • »17.04.07 - 20:15
    Profile