Project

General

Profile

Bug #11694

Artifacts on edges in tonemapping module

Added by Paolo Astengo 3 months ago. Updated about 1 month ago.

Status:
New
Priority:
Low
Assignee:
-
Category:
Darkroom
Target version:
-
Start date:
08/15/2017
Due date:
% Done:

0%

Affected Version:
git master branch
System:
Ubuntu
bitness:
64-bit
hardware architecture:
amd64/x86

Description

Hi,

using tonemapping I'm encountering an issue on the edges (see attached images). I didn't fount any possibility to overcome it.

DSC_0175_01.jpg (216 KB) Paolo Astengo, 08/15/2017 02:38 PM

DSC_0175.NEF.xmp (3.96 KB) Paolo Astengo, 08/15/2017 02:38 PM

Schermata da 2017-08-15 13-35-57.png (3.64 MB) Paolo Astengo, 08/15/2017 02:40 PM

DSC_0175.NEF (11 MB) Paolo Astengo, 08/15/2017 02:42 PM

Screenshot from 2017-10-13 15-37-37.png (3.93 MB) Ted Hennicke, 10/13/2017 07:40 AM

History

#1 Updated by Roman Lebedev 3 months ago

  • Status changed from New to Triaged
  • Assignee set to Roman Lebedev
  • % Done changed from 0 to 20

The raw crop is incorrect. I'll fix this.

#2 Updated by Roman Lebedev 3 months ago

  • Assignee deleted (Roman Lebedev)

Roman Lebedev wrote:

The raw crop is incorrect. I'll fix this.

Fixed.
As for the edges in the middle of the screen, i don't really know.
Maybe try local-laplacian based stuff - fusion in basecurve, local contrast

#3 Updated by Paolo Astengo 3 months ago

Thank you Roman,
I've tried fusion, local contrast with highlight and shadows, but I cannot reach the same result as tonemapping, especially on clouds.

#4 Updated by Roman Lebedev 3 months ago

Paolo Astengo wrote:

Thank you Roman,
I've tried fusion, local contrast with highlight and shadows, but I cannot reach the same result as tonemapping, especially on clouds.

I was talking about e.g. http://www.darktable.org/2016/08/compressing-dynamic-range-with-exposure-fusion/

#5 Updated by Paolo Astengo 3 months ago

Just to add some information (I hope this can be useful in some way)

Starting from the original image (zooming at 200% on the rightmost cross on the top of the roof), you can see the strange effect of the "chromatic aberration" module.

#6 Updated by Roman Lebedev 3 months ago

  • Status changed from Triaged to New
  • % Done changed from 20 to 0

(forgot to unset the status)

#7 Updated by Roman Lebedev 3 months ago

Paolo Astengo wrote:

Just to add some information (I hope this can be useful in some way)

Starting from the original image (zooming at 200% on the rightmost cross on the top of the roof), you can see the strange effect of the "chromatic aberration" module.

That is probably #8318

#8 Updated by Paolo Astengo 3 months ago

I think you're right.

The bad edges appear only in raw images, not in jpg files. I've taken a look on rawtherapee: it uses a slider for "edge stopping" and it works very well. A different algorithm?

#9 Updated by Ted Hennicke about 1 month ago

I'm experiencing the same issue. As you can see in this screenshot, the tonemapping algorithm is also producing areas that are too dark on the rocks of the mountain. Both types of artifact go away when the spatial extent is set to 0%, but they are perceptible from 1% onwards. The global tonemappers don't do nearly as good a job as this module, so these artifacts are a shame.

Also available in: Atom PDF