2-Minute Tutorials: Using LinearFit to Prepare NB Images for ChannelCombination

[@jitterbal]

I only recently found your videos and they are a great help. Thankyou. Hopefully there's one that covers the latter part of this workflow in more detail.

[@cosgrovescosmos]

Glad you find them useful!

[@trw4753]

Hi, thanks for the intersting video. My question: In your workflow (0.21 min into the video) the shown starless images are already nonlinear. Is LF then appropriate to use as it's no linear data anymore?

[@cosgrovescosmos]

Great Question! Linearfit sounds like it is used just for linear data. However, the name is really just referencing the numerical methodology used to put the best straight line through the data being modeled. It can do its job as long as there IS a linear relationship between the two data sets (in this case, the reference image pixels compared to the target image pixels). Now - it is more likely to work best in the linear domain as the mere act of stretching introduces nonlinearities. However, that does not mean it is not useful in the nonlinear domain. Some of the areas of nonlinearity are concentrated towards the toe and shoulder of the curve, so by clipping those, we concentrate on the middle portions of the curve, which can be fairly linear. The example workflow shown here is one I had used extensively for NB images. I should point out that BlurXtermintor has caused me to change my workflow. It does its best job when dealing with three color layers at the same time. So, I now create my SHO images in the linear domain, run BXT, and then go nonlinear. When I want to work on individual color channels, I extract them from the color image, enhance that layer, and then recombine them. So, these days, I tend to do this in the linear domain, and linearFIt is used there. But the key takeaway here is that LinearFit can be useful any time you are attempting to combine unlike images. Thanks, Pat

[@trw4753]

I wanted to follow your workflow of creating a SHO - Image using the Color Combination tool. Unfortunately PI refuses to CC stating that it works only with RGB - Data. I tear my hair out with PI!

[@cosgrovescosmos]

Hmmm - not sure what is happening there. ChannelCombination really does not know what you are feeding it when you select the files for the so-called r, g, and b fields. Are you sure you ae grabbing the right tool? You mention "Color Combination" and "CC" - CC usually refers to a color calibration tool, and that one does care about it being an RGB file. ChannelCombination takes 3 mono images and combines them to form a 3-layer color rgb image. If the files you hand it are really S2, Ha, and O3 ( aks SHO), then it will form a color file with those. But the input files must be single-channel mono files.

[@trw4753]

@@cosgrovescosmos Thank you so much for your answer. Yes I meant Channel combination, CC was just for short. Yes, the files are single channel mono ones in S, H and O respectively.

[@cosgrovescosmos]

Very strange - I am not sure why you are running into this problem. I do this all of the time, and in fact, as a sanity check, I just went over and tried it again. Worked without problems. Sorry - I seem to be no help to you for this one!

[@jitterbal]

I know this is 6 months too late but I had exactly this recently. The answer was simply to apply the process using the global symbol (filled in circle) instead of the square symbol ! I've had PI 3 years and still have plenty of frustration - especially when I bring up the assosciated document only to find it blank!

[@trw4753]

@@jitterbal Thank you for your answer - after much frustration I realised it in the end. So I'm not the only one to tear my hair out with PI......