Unable to remove severe Gradient

[tdgm]

Hello

I have been having difficult removing a  gradient from 3 hour Ha Image ( 9-1200 second images) of the Jelltfish Nebula in Gemini. Have tried ABE and DBE but can not seem to get the gradient removed. I have provided a link to an xisf file so someone might be able to try to remove the gradient. Below is a stretched jpg of the image so you can see what the problem is.

Link:https://www.dropbox.com/s/u6efcf39vjtvqe9/Boosted%20Auto-Stretch%20HA%20Master.xisf?dl=0

[img]Boosted Auto-Stretch HA Master.jpg[img]


Thanks for any suggestions


Steve TDGM

[tdgm]

HI

OOPS did not send a jpg properly

Here is the attachment

Steve Herrman

[wadeh237]

Here is my 60 second attempt with DBE.  This was just with changing some tolerances and dropping samples manually.  I'm sure that I could do better if I spent some time and did multiple passes (with one pass just to get the radial symmetry).

The DBE settings were as follows:

var P = new DynamicBackgroundExtraction;
P.data = [ // x, y, z0, w0, z1, w1, z2, w2
   [0.95345455, 0.05636364, 0.04258749, 0.979, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.02981818, 0.02272727, 0.04457539, 0.890, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.03854545, 0.94545455, 0.04246331, 0.981, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.95927273, 0.95454545, 0.04165712, 0.956, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.48436364, 0.95363636, 0.04176206, 0.965, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.97163636, 0.47909091, 0.04193973, 0.972, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.01090909, 0.46727273, 0.04291686, 0.972, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.52509091, 0.05818182, 0.04309147, 0.964, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.49018182, 0.45727273, 0.04092786, 0.923, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.69600000, 0.65363636, 0.04125897, 0.940, 0.00000000, 0.000, 0.00000000, 0.000], // row 10
   [0.74836364, 0.18727273, 0.04236361, 0.978, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.14690909, 0.33636364, 0.04312258, 0.963, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.23345455, 0.80454545, 0.04261778, 0.978, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.65745455, 0.22909091, 0.04196299, 0.972, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.60072727, 0.12272727, 0.04259168, 0.980, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.25963636, 0.48363636, 0.04217704, 0.980, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.14545455, 0.84363636, 0.04236858, 0.981, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.74909091, 0.84181818, 0.04153661, 0.954, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.60363636, 0.77454545, 0.04140647, 0.946, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.86327273, 0.46090909, 0.04201731, 0.975, 0.00000000, 0.000, 0.00000000, 0.000], // row 20
   [0.07709091, 0.70636364, 0.04274053, 0.976, 0.00000000, 0.000, 0.00000000, 0.000],
   [0.62472727, 0.70454545, 0.04130535, 0.942, 0.00000000, 0.000, 0.00000000, 0.000]
];
P.numberOfChannels = 1;
P.derivativeOrder = 2;
P.smoothing = 0.500;
P.ignoreWeights = false;
P.modelId = "";
P.modelWidth = 0;
P.modelHeight = 0;
P.downsample = 2;
P.modelSampleFormat = DynamicBackgroundExtraction.prototype.f32;
P.targetCorrection = DynamicBackgroundExtraction.prototype.Divide;
P.normalize = false;
P.discardModel = true;
P.replaceTarget = false;
P.correctedImageId = "";
P.correctedImageSampleFormat = DynamicBackgroundExtraction.prototype.SameAsTarget;
P.samples = [ // x, y, radius, symmetries, axialCount, isFixed, z0, w0, z1, w1, z2, w2
   [2622, 124, 10, 0, 6, 0, 0.042587, 0.979, 0.000000, 0.000, 0.000000, 0.000],
   [82, 50, 10, 0, 6, 0, 0.044575, 0.890, 0.000000, 0.000, 0.000000, 0.000],
   [106, 2080, 10, 0, 6, 0, 0.042463, 0.981, 0.000000, 0.000, 0.000000, 0.000],
   [2638, 2100, 10, 0, 6, 0, 0.041657, 0.956, 0.000000, 0.000, 0.000000, 0.000],
   [1332, 2098, 10, 0, 6, 0, 0.041762, 0.965, 0.000000, 0.000, 0.000000, 0.000],
   [2672, 1054, 10, 0, 6, 0, 0.041940, 0.972, 0.000000, 0.000, 0.000000, 0.000],
   [30, 1028, 10, 0, 6, 0, 0.042917, 0.972, 0.000000, 0.000, 0.000000, 0.000],
   [1444, 128, 10, 0, 6, 0, 0.043091, 0.964, 0.000000, 0.000, 0.000000, 0.000],
   [1348, 1006, 10, 0, 6, 0, 0.040928, 0.923, 0.000000, 0.000, 0.000000, 0.000],
   [1914, 1438, 10, 0, 6, 0, 0.041259, 0.940, 0.000000, 0.000, 0.000000, 0.000], // row 10
   [2058, 412, 10, 0, 6, 0, 0.042364, 0.978, 0.000000, 0.000, 0.000000, 0.000],
   [404, 740, 10, 0, 6, 0, 0.043123, 0.963, 0.000000, 0.000, 0.000000, 0.000],
   [642, 1770, 10, 0, 6, 0, 0.042618, 0.978, 0.000000, 0.000, 0.000000, 0.000],
   [1808, 504, 10, 0, 6, 0, 0.041963, 0.972, 0.000000, 0.000, 0.000000, 0.000],
   [1652, 270, 10, 0, 6, 0, 0.042592, 0.980, 0.000000, 0.000, 0.000000, 0.000],
   [714, 1064, 10, 0, 6, 0, 0.042177, 0.980, 0.000000, 0.000, 0.000000, 0.000],
   [400, 1856, 10, 0, 6, 0, 0.042369, 0.981, 0.000000, 0.000, 0.000000, 0.000],
   [2060, 1852, 10, 0, 6, 0, 0.041537, 0.954, 0.000000, 0.000, 0.000000, 0.000],
   [1660, 1704, 10, 0, 6, 0, 0.041406, 0.946, 0.000000, 0.000, 0.000000, 0.000],
   [2374, 1014, 10, 0, 6, 0, 0.042017, 0.975, 0.000000, 0.000, 0.000000, 0.000], // row 20
   [212, 1554, 10, 0, 6, 0, 0.042741, 0.976, 0.000000, 0.000, 0.000000, 0.000],
   [1718, 1550, 10, 0, 6, 0, 0.041305, 0.942, 0.000000, 0.000, 0.000000, 0.000]
];
P.imageWidth = 2750;
P.imageHeight = 2200;
P.symmetryCenterX = 0.500000;
P.symmetryCenterY = 0.500000;
P.tolerance = 5.000;
P.shadowsRelaxation = 3.000;
P.minSampleFraction = 0.050;
P.defaultSampleRadius = 10;
P.samplesPerRow = 10;
P.minWeight = 0.750;
P.sampleColor = 4292927712;
P.selectedSampleColor = 4278255360;
P.selectedSampleFillColor = 0;
P.badSampleColor = 4294901760;
P.badSampleFillColor = 2164195328;
P.axisColor = 4292927712;

[tdgm]

thanks wadeh237

Your result is a lot better than I have gotten but I am not adept at interpreting all your numbers a picture of were samples were placed would help as well as the DBE process.

Thanks again


Steve TDGM

[wadeh237]

What I gave you is a lot better than a screen shot (where the samples would be hard to see anyway).

If you click on the source icon in DBE (it's the first of the icons in the lower right of the window, the squarish one), it will bring up the instance source code.  You can just select whatever is there and replace it with what I posted.  Then, you will have the same settings that I used, right there on your machine.

[pfile]

just fyi you can of course save the process icon, zip it, and then attach it to a post here. your method does work but might be hard to execute for some (not necessarily steve)

rob

[wadeh237]

Ahh, thanks!

I hadn't thought of that.

Attached.

[tdgm]

Thanks for the icon and pfile thanks for the confidence in my knowledge


Steve TDGM