masked stretch problem

[jamesRC]

I have PI v1.8 and every time I try to do the MaskedStretch, on a variety of images, I get the message "black or too dark image, unable to auto-stretch".
I have looked into this and the solutions I found don't work. For background reference I have tried a preview of a small, dark-sky, also a piece of sky with some stars in it. All from a clone of the image I am trying to operate on. Every time I get the same error message. Entering the median value (the old version of MaskedStretch) seemed to be more reliable!

Suggestions welcomed-

James

[Alejandro Tombolini]

Hi James,
The upper limit in the Background reference in MaskedStretch tool is too low. Check it in the background preview (without stars) and adjust it to the maximum value.

Hope this helps,

Saludos, Alejandro.

[jamesRC]

Thank you for your reply, Alejandro.

Being rather a novice, I'm having some trouble following exactly what you are saying. I have had some success as I work on my understanding of previews as used here. But thank you!

James

[jamesRC]

I do not understand MaskedStretch at all.

For my reference background I try a clone of my linear image (a fits image of the Iris Nebula) stretched with HDRMultiscaleTransform at the latter's default settings. The result of this HDRMST transform is quite dark, appearing almost the same as the starting image.. The MS result using this image as a background reference is a washed out image that's not at all of "similiar brightness and contrast". I'm getting a headache trying to see what's wrong here! I am applying it as suggested in the "Dynamic Range and Local Contrast" tutorial of Vicent Peris. I think this is a very complicated tutorial and if a helpful person suggests a simpler start, I'm willing to listen! 

James.bewildered

[pfile]

hm, HDRMST should probably not be used on linear images.

all you need to do is measure the background in your linear image and adjust the slider in MaskedStretch to slightly exceed the measured background value in your linear image.

sometimes masked stretch leaves the blackpoint way too high, so you can fix that with HistogramTransformation after running MS.

rob

[drmikevt]

James

Allow me to spell things out in a little more detail.  First of all, it does not sound like you are ready for Vincent's tutorial and would be better served by the tutorials on the LightVortexAstronomy website.  There is an tutorial for Masked Stretch there.
Basically, you want to
1) Create an preview in your linear image that is only background - no nebula. 
2) Open Masked Stretch and leave everything at default.  For Background ref choose the preview you just created.
3) Zoom in on the preview and using your cursor, measure the background level. You might need to select 'readout mode'.  On the bottom of the screen there should be a K value - maybe .02 or something.  Enter that value in the Upper Limit box.  Leave the lower limit at 0. 
4)  Disable the screen transfer function on the image and apply Masked Stretch - should work just fine.  You can get more or less contrast by adjusting the clipping fraction. 

I hope this helps
Mike

[jamesRC]

Thank you for your help, rob and Mike!

Mike's brief but explicit instructions get me off to a good working start. Relief! When one is really new, it's difficult to even judge if a tutorial or help material is too advanced, which is true of the "Vincent" tutorial. Needs someone to be honest! I was attracted to this tutorial because of the very impressive results from using it. Heck, later.

I went about working with linear and stretched images in this particular way because of the wording of the tutorial I was trying to use. OK.

Thanks again, fellas! I'm off to LightVortexAstronomy.

James

[jamesRC]

one little problem remaining. I can't find a readout of the preview background "K" value. Only R G and B and something called "MiB" are there at the bottom of the screen. I thought there would be a specific "K" listed but can't find any? Neither is a K value listed when zoomed in on my preview.

Feeling so dense!

James

[pfile]

MiB is the super-technical way to say megabytes - this is the size of the data in the view.

IIRC if the image is RGB, PI is just going to show you the R,G,B values at the cursor. there is an implicit K component (lightness) but i guess it is not computed. so just take the highest of the 3 values. they should be similar to one another if you have done BackgroundNeutralization (and color calibration) which you *should* be doing before you stretch.

rob

[jamesRC]

Thank you, rob.

So many levels of newbie-ism!

Got it this time.

James