[Digikam-devel] tested 0.9 svn raw handling

[Thorsten Schnebeck]

Hi,

I have do to some work on a series of RAW /CR2) photos and I start to do it 
with digikam 0.9 svn and compare it with ufraw.

When comparing the results between ufraw and digikam I get 
better results with ufraw - less noise and a _far_ better visual start with 
correct brightness. This not very exiting when you open a RAW file and have 
nearly a black screen in digikam.

As we know somehow the dynamic range of raw files it should be better to use a 
auto dynamic stretcher per default.

By now I still have not found a workflow to get the same qualitiy like the 
embedded JPEG with digiKam


If have also ask Dave Coffin about denoising with dcraw:

> > what is the benefit when using the bilinear denoiser on dcraw level?
> > Could I get better/same results with an advanced denoiser on application
> > level
>
>      Dcraw applies bilateral filtering before Fuji
> rotation and colorspace conversion.  Applying this
> filter to the final dcraw output might give slightly
> different results.
>
> > Does -B work on every quality level (-q 0,1,3)?
>
>      You could use "-B" with lower quality settings,
> but that would be like buying the cheapest lubricating
> oil for your $100,000 Mercedes.  If AHD interpolation
> is too slow, forget about bilateral filtering.
>
> > When using the verbose output I get the information e.g.
> > Daylight multipliers: 2.098645 0.930145 1.180146
> > Camera multipliers: 2173.000000 1015.000000 1665.000000
> > What does this mean?
>
>      Multiplying red, green, and blue by these numbers
> should give color neutrality.  That is, a grey or white
> card in the photo should have R=G=B after multiplication.
>
>      The "Daylight" values are set within dcraw.c based
> on test images shot under D65 or sunlight.  The "Camera"
> values are set in the camera when the user selects an
> illuminant.  "dcraw" uses the first; "dcraw -w" uses the
> second.
>
>      The difference between 0.930145 and 1015.000000 is
> irrelevant -- only the red/green/blue ratios matter.
>
>                                 Dave Coffin  4/28/2006



Bye

  Thorsten