Tools for Linear vs Non-linear images

[cs_pixinsight]

Is there any reference documentation which explains which PI tools work best on linear images, which work best on non-linear and which can be used equally well on both?

Craig

[Lex]

Craig,

That's a nice question, this would interest me also because i still struggle with the if or not 
Hope the pro's here have an answer...

[Warhen]

Hey guys! I'm newer at this than many here, but... here goes:

My linear tools: STF, Fast Rotation, Dynamic Crop, RGB Working Space assignment, ABE, DBE, Color Calibration, Dynamic PSF and Deconvolution, ATrous Noise reduction

Nonlinear: Auto Histogram and Histogram Transformation, ACDNR NR, HDR Median Transform, Local Histogram Equalization, Morphological Transform, Saturation Curves, SCNR-G if needed

Hope that gets the ball rolling!

[Lex]

Warhen,

Thanks for that, i thought there were more differences 

[cs_pixinsight]

Warren, thank you for taking the time to detail the tools you use.  I wanted to wait in replying to give others a chance to review.  I was hoping they would give their opinions concerning how you felt the tools should be used and add to the list to get most of the tools/scripts covered. 

One question on deconvolution.  Should the support mask be linear or non-linear?  I also assumed that a star mask was used, which means to me that the image should be black for the stars and white elsewhere.  Is that assumption correct?

Here are the tools which are still not covered.  I'm sure a complete list would be helpful to all PI users, especially newbies like myself.  Hopefully others will be able to help complete this list and the authors of the scripts will be available to help itemize where they work best.

Processes:

• AssistedColorCalibration
• BackgroundNeutralization
• ChannelCombination
• ChannelExtraction
• ChannelMatch
• CloneStamp
• ColorSaturation
• Convolution
• DefectMap
• DigitalDevelopment
• DynamicAlignment
• FourierTransform
• GradientHDRComposition
• GradientHDRCompression
• GradientMergeMosaic
• GREYCstoration
• HDRComposition
• HDRMultiscalTransform
• ICCProfileTransformation
• InverseFourierTransform
• LarsonSekanina
• MultiscaleMedianTransformation
• NoiseGenerator
• RangeSelection
• RestorationFilter
• SimplexNoise
• StarAligment
• StarMask - I'm very curious about this one
• UnsharpMask

Scripts:
Image Analysis:

• ExtractWaveletLayers
• ImageSolver
• NoiseEvaluation

Instrumentation:

• CalculateSkyLimitedExposure

Utilities:

• BackgroundEnhance
• CanonBandingReduction
• CosmeticCorrection
• DarkStructureEnhance
• DeconvolutionPreview
• FFTRegistration
• StarHaloReducer

[Warhen]

Hi guys! Yes I'd certainly hope more experienced users will weigh in too. Still, to answer your questions as I understand them- correct, the deringing 'local support' star mask should protect the larger stars. Regarding the general mask to protect areas we don't wish to deconvolve, it's just much easier to define a good mask when the source image is nonlinear. To that end, leave the image in the linear space but use a clone to apply a nonlinear histogram transform to, then use Range Selection on the clone to create your mask and apply it to the original, linear image. CS_PI, of the items in your list and ones I did not mention- I regularly use: Star Mask and Range Selection for masks, and the DSE and BE Scripts.

[cs_pixinsight]

I'll keep updating this list as I come across discussions indicating which type of image they should used upon.  Feel free to update anything you know about too or correcting what I have outlined!

Tools working on linear and non-linear images equally well:

   AutoHistogram
   CurvesTransform
   DynamicAlignment
   HistogramTransformation
   IntegerResample
   Invert
   RangeSelection
   RGBWorkingSpace assignment
   Rotation
   StarAlignment
   Statistics
   processes that only alter the physical shape of the image


Tools best used on linear images:

   AssistedColorCalibration
   ATrousWavelet Noise reduction
   AutomaticBackgroundExtraction
   Color Calibration
   Fast Rotation
   Deconvolution
   DeconvolutionPreview
   DynamicBackgroundExtraction
   Dynamic Crop
   Dynamic PSF
   GradientHDRComposition
   GradientHDRCompression
   GradientMergeMosaic
   GREYCstoration
   HDRComposition
   ImageCalibration
   ImageIntegration
   ScreenTransferFunction

Tools best used on non-linear:

   ACDNR NR
   CurvesTransformation - for saturation
   HDRMultiscaleTransform
   LocalHistogramEqualization
   MorphologicalTransform
   MultiscaleMedianTransformation
   SCNR

That leaves the following uncategorized:

Processes:

   ChannelCombination
   ChannelExtraction
   ChannelMatch
   CloneStamp
   ColorSaturation
   Convolution
   DefectMap
   DigitalDevelopment
   FourierTransform
   ICCProfileTransformation
   InverseFourierTransform
   LarsonSekanina
   NoiseGenerator
   RestorationFilter
   SimplexNoise
   StarMask - I'm very curious about this one
   UnsharpMask


Scripts:
Image Analysis:

   ExtractWaveletLayers
   ImageSolver
   NoiseEvaluation

Instrumentation:

   BasicCCDParameters
   CalculateSkyLimitedExposure

Utilities:

   BackgroundEnhance
   CanonBandingReduction
   CosmeticCorrection
   DarkStructureEnhance
   FFTRegistration
   StarHaloReducer

Craig

[georg.viehoever]

Here some tools where I would suggest a different classification

   Fast Rotation: Linear and non-linear
   Dynamic Crop: Linear and non-linear
   GradientMergeMosaic: Linear and non-linear
   ScreenTransferFunction: Linear and non-linear
   MorphologicalTransform: Linear and non-linear
   CloneStamp: Linear and non-linear
   DefectMap: Linear
  NoiseGenerator: Generates new image
   ImageSolver: Linear and non-linear
   NoiseEvaluation: LInear
   BasicCCDParameters: Linear
   CalculateSkyLimitedExposure: Generates number only
   CanonBandingReduction: Linear
   CosmeticCorrection: Linear

[cs_pixinsight]

Thanks Georg, your input has been incorporated and I classified a few more.

Tools working on linear and non-linear images equally well:

   AutoHistogram
   CalculateSkyLimitedExposure - just generates numbers based on input
   CloneStamp
   CurvesTransform
   DynamicAlignment
   DynamicCrop
   Fast Rotation
   GradientMergeMosaic
   HistogramTransformation
   ImageSolver
   IntegerResample
   Invert
   MorphologicalTransform
   NoiseGenerator - generates a new image
   RangeSelection
   RGBWorkingSpace assignment
   Rotation
   ScreenTransferFunction
   StarAlignment
   Statistics
   processes that only alter the physical shape of the image


Tools best used on linear images:

   AssistedColorCalibration
   ATrousWavelet Noise reduction
   AutomaticBackgroundExtraction
   BasicCCDParameters
   CanonBandingReduction
   ColorCalibration
   CosmeticCorrection
   Debayer
   Deconvolution
   DeconvolutionPreview
   DefectMap
   DigitalDevelopment
   DynamicBackgroundExtraction
   Dynamic PSF
   ExponentialTransform
   ExtractWaveletLayers
   GradientHDRComposition
   GradientHDRCompression
   HDRComposition
   ImageCalibration
   ImageIntegration
   LinearFit
   NoiseEvaluation

Tools best used on non-linear:

   ACDNR NR
   CurvesTransformation - for saturation
   GREYCstoration
   HDRMultiscaleTransform
   LocalHistogramEqualization
   MultiscaleMedianTransformation
   SCNR
   StarMask - seems to work better on a clone which is non-linear (i.e. stretched), but will work on both types.

That leaves the following uncategorized:

Processes:

   ChannelCombination
   ChannelExtraction
   ChannelMatch
   ColorSaturation
   Convolution
   FourierTransform
   ICCProfileTransformation
   InverseFourierTransform
   LarsonSekanina
   RestorationFilter
   SimplexNoise
   UnsharpMask


Scripts:

Utilities:

   BackgroundEnhance
   DarkStructureEnhance
   FFTRegistration
   StarHaloReducer


Craig

[Lex]

That's nice, a good and correct list with all the great utilities in the right order 

[sreilly]

While this is a great list and most helpful it is by now  over two years old. To keep the information up to date, any additional tools/processes to add? Or has this been done elsewhere I haven't seen?

Thanks,

-Steve