[Digikam-users] Jpeg export changes image

[Hal V. Engel]

On Saturday 20 October 2007 00:17:33 Arnd Baecker wrote:
> Hi Hal,
>
> On Fri, 19 Oct 2007, Hal V. Engel wrote:
>
> [... very helpful explanation snipped...]
>
> > You might be able to find profiles that are specific to other workflows
> > such as the profiles from Nikon that are specific to the Nikon RAW
> > software on the net.  But those profiles will not be correct for my
> > workflow and probably not yours either.
>
> Thanks a lot for the detailed explanation - obviously,
> there is a lot I have to learn about color management ... ;-)

Color management is non-trivial and the learning curve is fairly steep.  So 
don't expect to understand this stuff in just a few hours.  I should add that 
it is normal to feel like you are in over your head when you first start 
trying to understand it.  Hang in there it will get better.

>
> This brings me back to *my* usual "can't this be simpler?"
> (which seems also in the spirit of Bill's last mail ;-).
> Well, to do things really right, one needs to create profiles
> in the way you explained before.
>
> However, for many users (at least for me) it would be sufficient
> (at least to start with) to
>   "for my raw file produce an image which looks
>    *essentially* the same as the jpg"
> The benefit compared to using the jpg are more Bits and
> no jpg artifacts (and no sharpening etc.).
> ((By essentially I mainly mean white-balance and brightness.
> Presumably, there is no way (and it would not make sense)
> to fully mimick the jpg, as this already may involve
> a lot of other post-processing, like levels adjustment,
> some filtering, sharpening, etc.; but yes, this "essentially"
> is a weak point in the argument ;-))

Also how do you tell any software to "produce an image which looks
*essentially" the same as the jpg"?  The conversion from the sensor data to 
the RGB image data is a complex process.  In addition, the camera processed 
jpeg images may not be very accurate since many camera manufacturers, 
particularly for consumer grade cameras, do all kinds of things to the images 
to make them "look better" such as artificially increasing saturation and 
contrast.  I would rather have an accurate representation as a starting point 
even if it looked a little flat and if I really wanted more saturation or 
contrast then I will do these changes myself.

>
> So the question is, whether for example the available ICC profiles
> (e.g. for bibblepro which also uses dcraw) are good enough
> to use with digikam, or what could be done (if possible at all)
> on the side of digikam to make things work as easy as possible
> for the user.

These will be better than using no profile or some generic profile such as 
sRGB.  But not as good as using a good custom profile.  So trade off is you 
lose a little image quality in return for a little convenience.  You also do 
not have the cost of buying a profiling target.

> Of course, this will not be the 100% solution you described
> (as for example it cannot take the light conditions into account),
> but might be ok for many situations ....
>
> Personally, for me the next important step will be to buy
> a device (which works under linux) for screen calibration,
> as to me this seems my biggest problem in the workflow
> of getting reliable prints at the lab.

At this point almost any device you can buy is supported although some of the 
support is very new and may still have some bugs.  LProf should have support 
for Spyder 2 and Huey devices in CVS in the next day or two.  There is 
currently support for DTP-92, DTP-92, and all of the EyeOne meters as well as 
some older devices like the Spectrolino.  In addition, there is even support 
for a DIY meter if you are handy with a soldering iron.

You can get Huey meters on ebay for around $50 plus shipping and EyeOne 
Display Lt meters can be found on the net for $130 to $140 and sometimes 
cheaper if you shop around.  The DTP-94 is discontinued but you may find one 
on sale cheap someplace and most consider this to be a better meter than the 
EyeOne Display 2 and LT.  For the record the only real difference between an 
EyeOne Display 2 and an Lt is the "unlock" string which allows the vendors 
software to pretend that these are really different devices. 

Regardless of which open source software you use to calibrate and profile your 
displays this software will also allow you to profile your cameras and 
scanners.  So you should consider buying a profiling target so that you can 
fully close the circle.  

>
> Thanks a lot for your time and expertise, it is really
> very much appreciated!!
>
> Best, Arnd
> _______________________________________________
> Digikam-users mailing list
> Digikam-users at kde.org
> https://mail.kde.org/mailman/listinfo/digikam-users