Isolating DSO

[whitenerj]

I've tried every combination of range mask and star mask in pixelmath to try and make a mask to isolate a DSO only. Say I wanna work on the color of the DSO and wanna exclude the background and the stars, how can I do that?

Jason

[sreilly]

You are inverting the mask aren't you?

[whitenerj]

I tried that I think. Should both masks be inverted?

Jason

[Jason Tackett]

This often works for me. First, extract the luminance and then use histogram transformation to brighten th DSO and darken the background. Then create star mask from your original image. Finally, subtract the star mask from the luminance mask using pixelmath. Applying the expression, $T - star_mask to the luminance mask will subtract the stars. Now the luminance mask will have the DSO selected, but no stars or background.

Cheers,
Jason

[pfile]

if the image you are trying to mask is in a linear state, and you have cloned or extracted the L* of the image to construct the mask, then you need to apply a HT to the mask to stretch it. otherwise the mask won't do much as all the values are very near 0.

rob

[sreilly]

if the image you are trying to mask is in a linear state, and you have cloned or extracted the L* of the image to construct the mask, then you need to apply a HT to the mask to stretch it. otherwise the mask won't do much as all the values are very near 0.

rob

Rob,

Would you even want to do any color adjustments other than color balance in the linear state? I might see that when in the is non-linear after the image has been stretched and then make these types of adjustments but I can't imagine why before that. But I could be wrong.

Steve

[pfile]

no although you might do SCNR at a linear stage.

i guess i only pointed that out because if you are constructing masks for linear-stage work (like perhaps for deconvolution) you need to make sure the mask actually "does something". since it sounded like it was not working for the OP i wanted to throw that out there. not sure his/her level of expertise with PI; it's a potential "gotcha"...

rob

[sreilly]

True, I remember trying to use deconvolution without a stretch mask and there was little to no difference. I have since then always made a clone and stretched to produce the mask, save the mask and delete the clone.