'Re: OpenGL Canvas or Not'

[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-kimageshop
Subject:    Re: OpenGL Canvas or Not
From:       David Revoy <davidrevoy () gmail ! com>
Date:       2013-10-18 9:30:35
Message-ID: CAPX1LSAkELiETSBBo6-RjGhz_drvp0Lo63_4AqmUSsCnU9FO6w () mail ! gmail ! com
[Download RAW message or body]

[Attachment #2 (multipart/alternative)]

On Fri, Oct 18, 2013 at 11:21 AM, Boudewijn Rempt <boud@valdyas.org> wrote:

> I tried to make that automatic, but it needs regenerating the mipmaps at
> those those zoom level breaks, and that didn't work out on my gpu's, so I
> got really weird results.
>
> The proper thing is to do the scaling in the shader, probably using the
> bicubic or the lanczos algorthm. I started work on that, got stuck...
> Dmitry is going to take a look, too.
>
> For the qpainter canvas, a solution could be to use Krita's internal
> bicubic scaling algorithm, which gives good results (you can check that by
> scaling the image itself), but might be a bit too slow. Of course, that
> might be solvable by using Vc to vectorize the scaling algorithms, which
> wouldn't be a bad idea at all in any case.
>
> Boud
>

Oh, thanks for the news about the rendering. I didn't knew you already
started to work on it. Really sad to learn it will be for you a lot of
work, as rewriting the scaling algorithm.   :/
I'm at your disposal if you ever make a branch to test something and need
the comparison of rendering between 'Nvidia + proprietary driver' and
'Intel + Libre Mesa'. I have both here with 2 Krita Git master.

David

[Attachment #5 (text/html)]

<div dir="ltr"><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Oct 18, \
2013 at 11:21 AM, Boudewijn Rempt <span dir="ltr">&lt;<a \
href="mailto:boud@valdyas.org" target="_blank">boud@valdyas.org</a>&gt;</span> \
wrote:<br> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px \
#ccc solid;padding-left:1ex"><div id=":ym" style="overflow:hidden">I tried to make \
that automatic, but it needs regenerating the mipmaps at those those zoom level \
breaks, and that didn&#39;t work out on my gpu&#39;s, so I got really weird \
results.<br>

<br>
The proper thing is to do the scaling in the shader, probably using the bicubic or \
the lanczos algorthm. I started work on that, got stuck... Dmitry is going to take a \
look, too.<br> <br>
For the qpainter canvas, a solution could be to use Krita&#39;s internal bicubic \
scaling algorithm, which gives good results (you can check that by scaling the image \
itself), but might be a bit too slow. Of course, that might be solvable by using Vc \
to vectorize the scaling algorithms, which wouldn&#39;t be a bad idea at all in any \
case.<br>

<br>
Boud</div></blockquote></div><br></div><div class="gmail_extra">Oh, thanks for the \
news about the rendering. I didn&#39;t knew you already started to work on it. Really \
sad to learn it will be for you a lot of work, as rewriting the scaling algorithm.   \
:/<br>  I&#39;m at your disposal if you ever make a branch to test something and need \
the comparison of rendering between &#39;Nvidia + proprietary driver&#39; and \
&#39;Intel + Libre Mesa&#39;. I have both here with 2 Krita Git master.<br> \
<br>David<br> </div></div>

_______________________________________________
Krita mailing list
kimageshop@kde.org
https://mail.kde.org/mailman/listinfo/kimageshop

[prev in list] [next in list] [prev in thread] [next in thread] 