[prev in list] [next in list] [prev in thread] [next in thread]
List: kde-kimageshop
Subject: Re: Whither Krita?
From: Sven Langkamp <sven.langkamp () gmail ! com>
Date: 2009-09-22 23:22:49
Message-ID: 478b087a0909221622g5be0689fv1e8ce016fc216da3 () mail ! gmail ! com
[Download RAW message or body]
[Attachment #2 (multipart/alternative)]
On Wed, Sep 23, 2009 at 12:27 AM, Matthew Woehlke <
mw_triad@users.sourceforge.net> wrote:
> Cyrille Berger wrote:
> > On Tuesday 22 September 2009, Matthew Woehlke wrote:
> >> Cyrille Berger wrote:
> >>> On Tuesday 22 September 2009, Sven Langkamp wrote:
> >>>> What would a 'color space change' mask be? Is there a case were it's
> >>>> needed to convert the colorspace manually with a mask?
> >>> Yes. Two cases actually, tone-mapping and raw. For raw, some of the
> >>> algorithms can be applied on a raw colorspace (well curves), some other
> >>> would need to be applied on a RGB one.
> >> I thought we'd said we wouldn't have a "RAW" colorspace, just va16¹?
> > That's not what the wiki says [1].
>
> Really?
>
> "Another option is to ignore the problem, accept that filters will not
> work correctly on bayer data, and import RAW directly into a 16-bit
> single channel (i.e. 'grayscale') layer. /At time of writing, this
> seemed to be the preferred option/ [...]" (emphasis added)
>
> ...and I certainly seem to recall that was the preferred solution. Much
> less special-casing, for one.
>
> > [1]
> >
> http://wiki.koffice.org/index.php?title=Krita/Direct_RAW_Editing#A_Dedicated_Colorspace
>
> >> You
> >> can't debayer as a mask² anyway, it's a specialized convolution filter.
> > Why not ?
>
> You conveniently snipped that part of my previous message:
>
> >> (² at least I think not; wouldn't a cs-conversion mask normally be
> >> necessarily 1:1?)
>
> So... can a mask¹ implement a convolution filter, or a filter with
> configurable parameters? (Bayer -> other cs is not a 1:1 filter like
> most cs conversions, it is a convolution filter³, and there are multiple
> choices of algorithm (and maybe even parameters?).)
>
> (³ ...and not even a 'straight' convolution filter. Even disregarding
> that you use different formulae depending on what channel the pixel
> represents, you have different formulae for the edges as well.)
>
> (¹ Maybe the problem is I don't understand what a "mask" is in krita
> internals. When I hear "mask" I think an extra alpha channel that is
> used to control what parts of a layer are blended, i.e. same as how it
> is used in svg terminology. Clearly the "mask" you are talking about is
> something completely different. It sounds more like a filter that...
> well, ignores masks. Which is confusing.)
>
Krita has filter masks. A filter mask works like adustment layer with a
mask, with the difference that it only works on the layer it's applied to.
[Attachment #5 (text/html)]
<div class="gmail_quote">On Wed, Sep 23, 2009 at 12:27 AM, Matthew Woehlke <span \
dir="ltr"><<a href="mailto:mw_triad@users.sourceforge.net">mw_triad@users.sourceforge.net</a>></span> \
wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, \
204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <div class="im">Cyrille Berger \
wrote:<br> > On Tuesday 22 September 2009, Matthew Woehlke wrote:<br>
>> Cyrille Berger wrote:<br>
>>> On Tuesday 22 September 2009, Sven Langkamp wrote:<br>
>>>> What would a 'color space change' mask be? Is there a case \
were it's<br> >>>> needed to convert the colorspace manually with a \
mask?<br> >>> Yes. Two cases actually, tone-mapping and raw. For raw, some \
of the<br> >>> algorithms can be applied on a raw colorspace (well curves), \
some other<br> >>> would need to be applied on a RGB one.<br>
>> I thought we'd said we wouldn't have a "RAW" colorspace, \
just va16¹?<br> > That's not what the wiki says [1].<br>
<br>
</div>Really?<br>
<br>
"Another option is to ignore the problem, accept that filters will not<br>
work correctly on bayer data, and import RAW directly into a 16-bit<br>
single channel (i.e. 'grayscale') layer. /At time of writing, this<br>
seemed to be the preferred option/ [...]" (emphasis added)<br>
<br>
...and I certainly seem to recall that was the preferred solution. Much<br>
less special-casing, for one.<br>
<div class="im"><br>
> [1]<br>
> <a href="http://wiki.koffice.org/index.php?title=Krita/Direct_RAW_Editing#A_Dedicated_Colorspace" \
target="_blank">http://wiki.koffice.org/index.php?title=Krita/Direct_RAW_Editing#A_Dedicated_Colorspace</a><br>
<br>
>> You<br>
>> can't debayer as a mask² anyway, it's a specialized convolution \
filter.<br> > Why not ?<br>
<br>
</div>You conveniently snipped that part of my previous message:<br>
<div class="im"><br>
>> (² at least I think not; wouldn't a cs-conversion mask normally be<br>
>> necessarily 1:1?)<br>
<br>
</div>So... can a mask¹ implement a convolution filter, or a filter with<br>
configurable parameters? (Bayer -> other cs is not a 1:1 filter like<br>
most cs conversions, it is a convolution filter³, and there are multiple<br>
choices of algorithm (and maybe even parameters?).)<br>
<br>
(³ ...and not even a 'straight' convolution filter. Even disregarding<br>
that you use different formulae depending on what channel the pixel<br>
represents, you have different formulae for the edges as well.)<br>
<br>
(¹ Maybe the problem is I don't understand what a "mask" is in \
krita<br> internals. When I hear "mask" I think an extra alpha channel that \
is<br> used to control what parts of a layer are blended, i.e. same as how it<br>
is used in svg terminology. Clearly the "mask" you are talking about is<br>
something completely different. It sounds more like a filter that...<br>
well, ignores masks. Which is confusing.)<br></blockquote><div><br>Krita has filter \
masks. A filter mask works like adustment layer with a mask, with the difference that \
it only works on the layer it's applied to.<br> </div></div><br>
_______________________________________________
kimageshop mailing list
kimageshop@kde.org
https://mail.kde.org/mailman/listinfo/kimageshop
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic