Hi Johannes,
If the important thing is how closely the curves follow sigma=1 in "flat
noise levels" graphs, then one of my earlier photo shoots actually seems to
be better -- the curves follow 1 more tightly with less fluctuation:
http://emremeydan.com/tempor/550d-denoise_02.tar.gz
Do you think these would work better? I can't directly compare them to the
other ones, because i don't know how to get both sets of presets into
darktable database.
If these are also not good enough, i'm willing to give it one last try, and
shoot another set. I mean, if there's an improvement to be made, it should
be made :) I'm just waiting for a bit of sun, so i can get high contrast
and good clipping.
Thanks!
...
emre
On Mon, Dec 17, 2012 at 10:29 PM, johannes hanika <[email protected]> wrote:
> hey
>
> On Tue, Dec 18, 2012 at 2:33 AM, Emre Meydan <[email protected]> wrote:
> > Hi,
> >
> > I know somebody else was working on a profile for this camera, but i also
> > wanted to try it myself.
> >
> > I had to make three different sets of shoots before i got a nice profile.
> > What i tried last (and what seemed to work) was to shoot the window of my
> > room: the sky was overexposed, while the shadows in the room provided
> > underexposure. I also exhaled on the lens for a bit of extra-blur :)
>
> good idea :)
>
> > I'm not sure if everything is as it should be: looking at the graphs, the
> > red channel is not clipped, and for iso 400 & 1600 the blue channel also
> > doesn't seem to be clipped (though close).
>
> graphs look fine. iso 100 is not quite as overexposed as the others,
> so the fit is not optimal (but who denoises iso 100..).
>
> if roumano agrees, i'd use your values as new values for the 550d.
>
> cheers,
> jo
>
> > But after initial testing it seems to work very well. Up to 400 iso the
> > results are really good. With higher iso the profile removes fine noise,
> but
> > larger color blotches remain -- as other people noted. I guess they
> require
> > a different kind of noise reduction.
> >
> > One small problem i have is that, in the presets section i see two
> different
> > sets of presets: half of the presets are named "Canon 550D", but
> selecting
> > them doesn't change a thing (even though darktable seems to calculate
> stuff
> > when they're selected). The other half are simply called "test" and they
> are
> > the ones that work.
> >
> > You can get the resultant files here, and check if anything is wrong:
> > http://emremeydan.com/tempor/550d-noise_01.tar.gz
> >
> > Included is a camera-jpeg of 400iso, to show you what i used for
> profiling.
> > I can also send the raw files if you need them.
> >
> > Thanks a lot for the hard work!
> >
> > ...
> > emre
> >
> >
> ------------------------------------------------------------------------------
> > LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
> > Remotely access PCs and mobile devices and provide instant support
> > Improve your efficiency, and focus on delivering more value-add services
> > Discover what IT Professionals Know. Rescue delivers
> > http://p.sf.net/sfu/logmein_12329d2d
> > _______________________________________________
> > darktable-devel mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/darktable-devel
> >
>
------------------------------------------------------------------------------
LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
Remotely access PCs and mobile devices and provide instant support
Improve your efficiency, and focus on delivering more value-add services
Discover what IT Professionals Know. Rescue delivers
http://p.sf.net/sfu/logmein_12329d2d
_______________________________________________
darktable-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/darktable-devel
