Seems chrome has implemented color management has anyone done some comparisons..

Home Forums General Discussion Seems chrome has implemented color management has anyone done some comparisons..

Viewing 12 posts - 1 through 12 (of 12 total)
  • Author
    Posts
  • #11953

    A.ces
    Participant
    • Offline

    Seems chrome has implemented color management has anyone done some comparisons between chrome and firefox and checked which software manages colors better?

    I’ve done some quick checks between chrome and firefox on lagom black/gradient test and it seems there are some noticeable differences in banding, and black level with chrome doing better in the black level test displaying the first 1-5 better than on firefox, as for the bandient test i’m unsure if one is better than the other as the banding looks different depending on the browser.

    #11955

    A.ces
    Participant
    • Offline

    UPDATE: seems Firefox doesn’t handle LUT /v4 profiles so well blacks are kind of clipped so chrome actually does a better job at color management than firefox now.

    #11960

    Florian Höch
    Administrator
    • Offline

    No, Chrome color management with cLUT (and anything other than single curve + matrix) profiles is broken (see color management test page). Firefox needs gfx.color_management.enablev4 true (in about:config) for cLUT profile support, and gfx.color_management.mode 1 so untagged content is color managed as well.

    #11966

    A.ces
    Participant
    • Offline

    Hmm your right chrome does not use the clut part of the correction, I checked but I still find that black looks somewhat clipped in Firefox with the prefs above set to true compared to Photoshop etc .

    #11970

    MW
    Participant
    • Offline

    I read somewhere PS and Firefox apply BPC differently. I can’t say if this explains the differences you see, I haven’t compared myself.

    #11971

    A.ces
    Participant
    • Offline

    No, Chrome color management with cLUT (and anything other than single curve + matrix) profiles is broken (see color management test page). Firefox needs gfx.color_management.enablev4 true (in about:config) for cLUT profile support, and gfx.color_management.mode 1 so untagged content is color managed as well.

    Hmm it doesn’t really explain the differences in black levels i see if i disable v4 support in firefox making it only use the matrix profile, even with v4 support enabled i can only see the 4th black square while in chrome square 1-3 is also visible but not to bright.

    Seems like both chrome and firefox has tradeoffs with black level being better on chrome while firefox has full color correction.

    #11974

    Florian Höch
    Administrator
    • Offline

    I read somewhere PS and Firefox apply BPC differently.

    The accuracy with which the transform is created is different. Firefox uses a slightly faster, slightly less accurate code path. In practice, this doesn’t matter much though. Also, Photoshop always uses relative colorimetric + BPC for display rendering (unless you set up a custom softproof), while Firefox defaults to perceptual.

    even with v4 support enabled i can only see the 4th black square while in chrome square 1-3 is also visible but not to bright

    I really wished the lagom.nl test page would go away. It’s usefulness in a color managed environment is limited to begin with (see the respective entry in the FAQ in this forum), and the test pictures are not designed in a sensible way that takes into account how human vision works and how real display devices behave: For the lagom “black test” levels 1-3 on a standard LCD display should be very hard to distinguish or see, and that is expected and inherent to the way they have designed the test images. Level 1 in sRGB (assuming a display with ideal zero black and 100 cd/m2 white) is 0.03 cd/m2, level 2 is 0.06 cd/m2, level 3 0.09 cd/m2. But the human visual response isn’t linear, and real displays have an above zero black level, let’s say 0.1 cd/m2 for a 1000:1 contrast display. Adding that in, the differences now become insignificant: 0.103 cd/m2, 0.106 cd/m2, and 0.109 cd/m2.

    The test I’ve designed is much more useful, because the above black test circles are spaced in a perceptually uniform manner.

    Seems like both chrome and firefox has tradeoffs with black level being better

    Chrome’s color management doesn’t work correctly, simple as that. You can check the respective bugs on the chrome bug tracker and threads in this forum to get an idea about the problem.

    #11989

    A.ces
    Participant
    • Offline

    I read somewhere PS and Firefox apply BPC differently.

    The accuracy with which the transform is created is different. Firefox uses a slightly faster, slightly less accurate code path. In practice, this doesn’t matter much though. Also, Photoshop always uses relative colorimetric + BPC for display rendering (unless you set up a custom softproof), while Firefox defaults to perceptual.

    even with v4 support enabled i can only see the 4th black square while in chrome square 1-3 is also visible but not to bright

    I really wished the lagom.nl test page would go away. It’s usefulness in a color managed environment is limited to begin with (see the respective entry in the FAQ in this forum), and the test pictures are not designed in a sensible way that takes into account how human vision works and how real display devices behave: For the lagom “black test” levels 1-3 on a standard LCD display should be very hard to distinguish or see, and that is expected and inherent to the way they have designed the test images. Level 1 in sRGB (assuming a display with ideal zero black and 100 cd/m2 white) is 0.03 cd/m2, level 2 is 0.06 cd/m2, level 3 0.09 cd/m2. But the human visual response isn’t linear, and real displays have an above zero black level, let’s say 0.1 cd/m2 for a 1000:1 contrast display. Adding that in, the differences now become insignificant: 0.103 cd/m2, 0.106 cd/m2, and 0.109 cd/m2.

    The test I’ve designed is much more useful, because the above black test circles are spaced in a perceptually uniform manner.

    Seems like both chrome and firefox has tradeoffs with black level being better

    Chrome’s color management doesn’t work correctly, simple as that. You can check the respective bugs on the chrome bug tracker and threads in this forum to get an idea about the problem.

    ahh okay i always thought the lagom.nl was a good test page for monitors.

    Do you know if firefox also supports multi monitor color correction or only on the primary monitor ?

    #26168

    A.ces
    Participant
    • Offline

    I read somewhere PS and Firefox apply BPC differently.

    The accuracy with which the transform is created is different. Firefox uses a slightly faster, slightly less accurate code path. In practice, this doesn’t matter much though. Also, Photoshop always uses relative colorimetric + BPC for display rendering (unless you set up a custom softproof), while Firefox defaults to perceptual.

    even with v4 support enabled i can only see the 4th black square while in chrome square 1-3 is also visible but not to bright

    I really wished the lagom.nl test page would go away. It’s usefulness in a color managed environment is limited to begin with (see the respective entry in the FAQ in this forum), and the test pictures are not designed in a sensible way that takes into account how human vision works and how real display devices behave: For the lagom “black test” levels 1-3 on a standard LCD display should be very hard to distinguish or see, and that is expected and inherent to the way they have designed the test images. Level 1 in sRGB (assuming a display with ideal zero black and 100 cd/m2 white) is 0.03 cd/m2, level 2 is 0.06 cd/m2, level 3 0.09 cd/m2. But the human visual response isn’t linear, and real displays have an above zero black level, let’s say 0.1 cd/m2 for a 1000:1 contrast display. Adding that in, the differences now become insignificant: 0.103 cd/m2, 0.106 cd/m2, and 0.109 cd/m2.

    The test I’ve designed is much more useful, because the above black test circles are spaced in a perceptually uniform manner.

    Seems like both chrome and firefox has tradeoffs with black level being better

    Chrome’s color management doesn’t work correctly, simple as that. You can check the respective bugs on the chrome bug tracker and threads in this forum to get an idea about the problem.

    Florian it seems that chrome has implemented LUT/3DLUT correction now in chrome on your test page i get this result

    Attachments:
    You must be logged in to view attached files.
    #26182

    MW
    Participant
    • Offline

    Can confirm in Brave v1.14.84 (Chrome v85.0.4183.121 based). Near black ramps also look normal now. I’m still not sure if Chrome is using display profile or just assuming sRGB.

    #26188

    MW
    Participant
    • Offline

    I take it back, the grey levels are broken: https://bugs.chromium.org/p/chromium/issues/detail?id=1134280&q=color%20management&can=2 Even
    “Force Color Profile>sRGB doesnt fix this.

    #26189

    A.ces
    Participant
    • Offline

    I take it back, the grey levels are broken: https://bugs.chromium.org/p/chromium/issues/detail?id=1134280&q=color%20management&can=2 Even
    “Force Color Profile>sRGB doesnt fix this.

    Yeah I’m  the one that reported that bug on chromium, they also ideally need a way to completely  disable color management for people using monitors with  hardware 3DLUT support

    Firefox also has fixed the color corruption with v4 profiles that  I and another guy found the regression  window for.

    Though on videos there still seems to be something wrong on chrome as in a webm animation  the gray background was noticeable tinted reddish, unless I put color profile to scSRGB, all other profiles still had the tint.

    • This reply was modified 3 weeks ago by A.ces.
Viewing 12 posts - 1 through 12 (of 12 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS