Camera Support #11658

Noise profile for Sony Alpha 68

Added by Jan W almost 2 years ago. Updated almost 2 years ago.

Target version:
Start date:
Due date:
% Done:


Affected Version:
git master branch


I created a noise profile for the Sony Alpha 68 (ILCA-68) according to the instructions found here . I did some testing and it looks like it does a good job.
Since this is my first noise profile, I think it would be a good idea to review it before it is even considered to be included in Darktable.

dt-noiseprofile-20170622.tar.gz (8.24 MB) Jan W, 06/22/2017 09:43 PM

dt-noiseprofile-20170624.tar.gz (9.54 MB) Jan W, 06/24/2017 04:48 PM

Associated revisions

Revision 54638f90
Added by Stefan Schöfegger almost 2 years ago

noise profile: Fix Sony Alpha 68 (ILCA-68), fixes #11658

This corrects the wrong added presets.json.
Double iso entries removed


#1 Updated by Stefan Schöfegger almost 2 years ago

Of course, all noise profiled are reviewed before added.
ISO 1000, 1250 and 1600 to corrected to less, all ISO's above 2000 are overcorrected. I am not sure if this cames from your setup our there are other problems.
You images only contain very dark and very bright parts, less between. Please try another setup where there are more between. (but with some over- und underexposed parts) I think it could also be more out of focus.

#2 Updated by Jan W almost 2 years ago

Thank you. Here's the second attempt.
Notice I used 2 different scenes here, which might be a big nono for this purpose (I don't know). Let me know which is best, or if it doesn't matter. I'll redo it if necessary.
I had some trouble getting both under- and overexposed images for the lower ISO values. Any advice on getting good test images is welcome. (like filling the histogram not just with very light and dark parts)

#3 Updated by Stefan Schöfegger almost 2 years ago

  • % Done changed from 70 to 50
  • Status changed from Patch attached to In Progress
  • Assignee set to Stefan Schöfegger

The results look very good now, I add them.
May I use your images for a tutorial about noise profiling?

#4 Updated by Jan W almost 2 years ago

Just one question: When I look at the noise profile data I see there are multiple entries for ISOs 320, 400 and 500. Is that going to be a problem? The values for a do not seem to nicely fit the data there.

You can use the images for a tutorial.

#5 Updated by Stefan Schöfegger almost 2 years ago

I have overseen this in,

I added the wrong profile, will fix this soon. They came from the multiple images for this iso setting.

#6 Updated by Stefan Schöfegger almost 2 years ago

The results don't differ much. You have to compare only the second column (green channel) for a and b values.

            {"name": "ILCA-68 iso 320", "iso": 320, "a": [1.48337366981346e-05, 9.43173343631512e-06, 3.02295004613881e-05], "b": [2.51046792737187e-08, 1.93782339461145e-08, 2.72587932805201e-08]},
            {"name": "ILCA-68 iso 320", "iso": 320, "a": [2.71949832242963e-05, 9.57879716990598e-06, 1.71096542659364e-05], "b": [3.19837641145244e-08, 2.41653060396964e-08, 4.01392690327929e-08]},
            {"name": "ILCA-68 iso 400", "iso": 400, "a": [1.85519534039609e-05, 1.1778464715661e-05, 3.3497819081171e-05], "b": [3.53200653046325e-08, 2.86406104361894e-08, 3.81472543789127e-08]},
            {"name": "ILCA-68 iso 400", "iso": 400, "a": [3.31440286186517e-05, 1.19943083413647e-05, 2.13785576215162e-05], "b": [4.62134955644255e-08, 2.48053562396221e-08, 5.49361160101433e-08]},
            {"name": "ILCA-68 iso 500", "iso": 500, "a": [2.35758031191657e-05, 1.48419005516014e-05, 4.80565709178794e-05], "b": [4.14366443620498e-08, 3.51293934056073e-08, 3.83842919835094e-08]},
            {"name": "ILCA-68 iso 500", "iso": 500, "a": [4.20349931135074e-05, 1.50447268388513e-05, 2.70509147063263e-05], "b": [6.19413917663322e-08, 4.58127934567376e-08, 7.50630629708116e-08]},

9.431*10^-6 -> 9.579*10^-6
1.177*10^-5 -> 1.199*10^-5
The b values are the offsets and are much smaller than the a values.

#7 Updated by Stefan Schöfegger almost 2 years ago

  • Status changed from In Progress to Fixed
  • % Done changed from 50 to 100

#8 Updated by Jan W almost 2 years ago

Super. Thanks for the effort.

#9 Updated by Roman Lebedev almost 2 years ago

  • Target version set to 2.4.0

Also available in: Atom PDF