Re: Conversion of 608/708 captions to WebVTT - how to map the colour green

On Tue, Feb 24, 2015 at 10:24 PM, Philip Jägenstedt <philipj@opera.com>
wrote:

> On Wed, Feb 25, 2015 at 12:16 PM, Glenn Adams <glenn@skynav.com> wrote:
> >
> > On Tue, Feb 24, 2015 at 9:36 PM, Philip Jägenstedt <philipj@opera.com>
> > wrote:
> >>
> >> On Sat, Feb 21, 2015 at 7:10 AM, Michael Borthwick
> >> <mb@michaelborthwick.com.au> wrote:
> >> > On 08/04/2014, at 1:08 PM, Silvia Pfeiffer wrote:
> >> >
> >> >> Note: I've made the change to "lime" and updated the related text,
> see
> >> >> https://dvcs.w3.org/hg/text-tracks/rev/07a843b7f31d
> >> >>
> >> >> Please let me know if there are more changes necessary.
> >> >>
> >> >> Regards,
> >> >> Silvia.
> >> >
> >> >
> >> > Dear list members,
> >> >
> >> > For those who are interested I have made the 22 minute video of my
> SMPTE
> >> > Australia 2013 talk regarding this issue available online with
> >> > appropriate
> >> > amendments to reflect the changes kindly made by Silvia last year to
> the
> >> > mapping of green into CSS lime.
> >> >
> >> >
> >> >
> http://michaelborthwick.com.au/closed_captioning_online_streaming_video_dfxp_webvtt.html
> >>
> >> Thanks Michael, this was a great talk! I'm not sure why, but I found
> >> the trouble with green and conflicting standards very amusing, a very
> >> familiar kind of problem.
> >>
> >> Since WebVTT doesn't have any dedicated syntax for colors it really
> >> comes down to the CSS that goes along with the captions, and of course
> >> any mapping specs. Unfortunately, I suspect you will not be the last
> >> to spend time looking for a suitable green color.
> >
> >
> > Haven't had a chance to watch yet, but from TTML:
> >
> > <named-color>
> >   : ...
> >   | "green"                                 // #008000ff
> >   | "lime"                                  // #00ff00ff
> >   | "olive"                                 // #808000ff
> >
> >
> >>
> >>
> >> Did you ever hear back from the BBC about which the ultimate colors
> >> are? I doubt that #00FF00 is actually used on the Web, it's so bright!
> >
> >
> > Y = 0.299R + 0.587G + 0.114B
>
> Is this BBC green? Multiplying with 255 and rounding, that is rgb(76,
> 150, 29), assuming it's the same RGB colorspace of course...
>

Y (or more accurately Y') is the luminance component of Y'UV or Y'PbPr [1].
My point was that the G component of Y contributes ~60% of brightness. So
an #00FF00 mapping to Y would be bright indeed.

[1] http://en.wikipedia.org/wiki/YUV


>
> In any case, it's a lot nicer than #00FF00, and a bit brighter than
> #008000.
>

In 8-bit Y'CbCr (ITU-R BT.601) [2], which is a typical digital video sample
representation, the 8 bit value of Y' for #008000 would be 81 and for
#00FF00 would be 145 (in a range of [16,235], i.e., Y'(black) = 16,
Y'(white) = 235).

[2] http://en.wikipedia.org/wiki/YCbCr

The brightness (Y') of #00FF00 would be 149, and for #008000 would be 75,
when viewed relative to Y'


>
> Philip
>