[argyllcms] AW: Re: Re: Color management -n- web browsers
- From: "Andreas Mock" <andreas.mock@xxxxxx>
- To: <argyllcms@xxxxxxxxxxxxx>
- Date: Fri, 20 Dec 2019 07:19:10 +0100
TRC seems to be Tone Response Curve in this context.
Von: argyllcms-bounce@xxxxxxxxxxxxx <argyllcms-bounce@xxxxxxxxxxxxx> Im Auftrag
von ??????? ??????? (Redacted sender "koraalex" for DMARC)
Gesendet: Freitag, 20. Dezember 2019 06:07
An: argyllcms@xxxxxxxxxxxxx
Betreff: [argyllcms] Re: [argyllcms] Re: Color management -n- web browsers
While making experiments with browsers I’ve found that I don’t understand TRC
term and its place in CM workflow. Could anybody describe it? X-Rite page says
TRC is used for non-CM applications (wow!), other sources say nothing on it (or
they for TRCs).
Среда, 18 декабря 2019, 6:45 +05:00 от Florian Höch <lists+argyllcms@xxxxxxxxx
<mailto:lists+argyllcms@xxxxxxxxx> >:
Am 18.12.2019 um 01:46 schrieb Graeme Gill:
Various websites purport to have this information. i.e.
<https://www.color-management-guide.com/web-browser-color-management.html>
Might be of interest, a while ago I created my own test page to get a
better view on what (common) profile type(s) browsers support. It can be
found here:
https://displaycal.net/icc-color-management-test/
[ the test page is limited to RGB profiles though ]
it appears that Google has deliberately broken color management
in the Chrome browser
[...]
This is nonsense on their behalf.
[...]
i.e. if it doesn't work for them on MSWindows
and Linux then it's because they are messing up their implementation!
Yep. [ warning: rant incoming ]
I have watched that whole trainwreck unfold and was even involved in
some discussion around the fallout of it over the course of several
months/years and must say I'm still speechless at the sheer lack of
competence on their part. Initially my views on Chrome "color
management" weren't as negative as they are now, but after I
(progressively) learned that:
- they basically tried to roll their own CMM instead of using something
proven (as if creating a CMM was such a menial task that it would make
sense to start from scratch... you'd think they would've gotten a clue
after the debacle that Firefox qCMS initially was)
- instead of using the profile parts (tags) appropriately, they tried to
derive "simplified" representations of certain profile tags (like rgbTRC
curves). As if rgbTRC were not already simplified enough...
- in some cases, their simplification attempts introduced rendering
artifacts (gee, you don't say!)
- they seemingly ignored B2A tables of cLUT-type display profiles
completely, instead, at some point, they tried to use the input shaper
curves (1D LUTs) of lut16Type A2B0 tables as a substitute for using
rgbTRC tags (!), ignoring the 3D cLUT data and output 1D LUTs...
(seriously?!)
- due to these reasons, at some point their "color managed" rendering of
content became an ungodly mess laden with graphical artifacts (well gee,
if they really took the 1D LUTs of A2B without even understanding that
they are useless without the cLUT and output 1D LUTs then I'm not
exactly surprised as to why that was!)
- in a conversation with them, when someone came up with the conspiracy
theory that Argyll's diagnostic "swapped matrix" profile type was an
"aggressive" move to highlight their implementation shortcomings, I
finally lost my sh*t and called that for the load of bull that it
clearly is (this existed longer than their implementation was even around).
- realizing all the negative impact of their completely messed up
implementation, they finally removed most of their "color management",
leaving in place a barely functional(?) husk, that is still broken but
probably has less/no rendering artifacts.
Rant over. Apologies everyone.
(take this all with a grain of salt as I only briefly looked at some of
their code and most of my impressions come from the observable results -
which may be anecdotal, even though experienced by a number of different
people - as well as their comments)
Florian.
Other related posts:
- » [argyllcms] AW: Re: Re: Color management -n- web browsers - Andreas Mock
