Author Topic: DarkStructureEnhance: works differently on whole image  (Read 3852 times)

Offline Nocturnal

  • PixInsight Jedi Council Member
  • *******
  • Posts: 2727
    • http://www.carpephoton.com
DarkStructureEnhance: works differently on whole image
« on: 2008 December 19 20:30:46 »
Hi,

I created a preview of my image containing both nebulosity and background. Then created an image of the preview. I applied DSE and was very pleased with the result so I applied it to the original image. It didn't look nearly as good.

So apparently the process depends on the entire image, not just pixel values or even neighboring pixels. I'd like to understand what aspect of the process works this way if someone could explain it please. I thought my preview was a good cross-section of brightness levels of the image. Is there something I can do to improve how representative my preview is of the whole image?
Best,

    Sander
---
Edge HD 1100
QHY-8 for imaging, IMG0H mono for guiding, video cameras for occulations
ASI224, QHY5L-IIc
HyperStar3
WO-M110ED+FR-III/TRF-2008
Takahashi EM-400
PIxInsight, DeepSkyStacker, PHD, Nebulosity

Offline C. Sonnenstein

  • PixInsight Addict
  • ***
  • Posts: 262
    • http://astrosurf.com/astro35mm
DarkStructureEnhance: works differently on whole image
« Reply #1 on: 2008 December 20 14:11:13 »
Like you have said, the process depends on the entire image, because wavelet transformation works with the structure size of the whole image. If you select a small portion, the DSE process only detect smaller features.
Carlos Sonnenstein

Offline Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 7111
    • http://pixinsight.com/
DarkStructureEnhance: works differently on whole image
« Reply #2 on: 2008 December 25 15:24:32 »
Hi Sander,

Unfortunately, as Carlos has pointed out, the DSE algorithm depends on large-scale structures. You know what this means: a non-previewable procedure.

However, you can also take advantage of the algorithm's multiscale nature: if you downsample the image by an integer factor (f.e. with IntegerResample), say 1/2, you should get an accurate preview of what will happen for the actual image.

In theory, at least  :wink:
Juan Conejero
PixInsight Development Team
http://pixinsight.com/

Offline C. Sonnenstein

  • PixInsight Addict
  • ***
  • Posts: 262
    • http://astrosurf.com/astro35mm
DarkStructureEnhance: works differently on whole image
« Reply #3 on: 2008 December 26 02:19:32 »
Hi,

This is a special case with DSE working with the Rosetta's dark structures nebula:
http://pteam.pixinsight.com/sonnenstein/dse/mouseover_dse_rosetta.htm

In normal conditions DSE can't achieve a good result with the smaller features. The mouseover shows a DSE enhancement after oversampling the whole image by a factor of 4:1, applied DSE process with seven wavelet layers and 3x3 linear interpolation scaling function, and then subsampling the DSE result by the same factor of 1:4

Actually, this is a good way to adapt DSE to the smaller features in the whole image when the algoritm can't do a good job.

Have a nice day.
Carlos Sonnenstein

Offline Nocturnal

  • PixInsight Jedi Council Member
  • *******
  • Posts: 2727
    • http://www.carpephoton.com
DarkStructureEnhance: works differently on whole image
« Reply #4 on: 2009 January 05 14:07:02 »
Nice demo and technique Carlos! I use DSE on almost all my images now. I'll have to try the up/down sampling technique. Very smart.
Best,

    Sander
---
Edge HD 1100
QHY-8 for imaging, IMG0H mono for guiding, video cameras for occulations
ASI224, QHY5L-IIc
HyperStar3
WO-M110ED+FR-III/TRF-2008
Takahashi EM-400
PIxInsight, DeepSkyStacker, PHD, Nebulosity