NGC 7640

[andyschlei]

It's been quite a while since I've had any data to process or to post.  But I finally had some time to image in August and here is what I got.

It is NGC 7640, a spiral galaxy in Camelopardalis.  It is a fairly dim object.  I did practically all of my processing in PixInsight (final touch up and save for web in Photoshop  ), and the  video tutorial on LRGB processing was excellent.  The color balancing and background neutralization tools are great and easy to use.

The tutorial does show use of HDR Wavelets without a star mask (using lightness mask instead) so I used that technique here.  Not sure if that is a best practice.

Another step I added was to crop off the edges of all the sub-frames and then rescale the images prior to combining into the original RGB.  The aligned images have 5-10 pixels around the edge that are not complete from my dithering.  This area tends to have pixels with very low values, lower than the general background of the image.  Doing this, my initial RGB is much closer to balanced then when I do not.  I trust that a rescale is not changing the linear nature of the images -- after all, PixInsight does rescale any floating point image whenever it is opened.

I still struggle with curvature in the field that leads to some odd shaped stars and some color halos.  But overall I am reasonably happy with the result.  A new scope is on order so I expect better optics will help with the curvature.

All comments and suggestions are welcome.



This is 70 minutes luminance and 20 minutes each RGB with an ST-10 on a C-11 @f6.3.

Clicking on the image will take you to my gallery and the ability to get to a full-sized image.

Thanks for looking,

--Andy

[andyschlei]

I forgot to mention in my original post that the main reason I used Photoshop was to get rid of blooms.  The healing brush is very good at that.

Any suggestions for a bloom removal technique in PixInsight?

Thanks,

--Andy

[Juan Conejero]

Hi Andy,

The tutorial does show use of HDR Wavelets without a star mask (using lightness mask instead) so I used that technique here.  Not sure if that is a best practice.

You usually want to apply HDRWaveletTransform to the stars too. The integrated lightness mask in the HDRWT process is normally the best option. Note that this is actually an inverted mask used to protect low SNR areas.

I trust that a rescale is not changing the linear nature of the images

Rescaling is a strictly linear operation.

-- after all, PixInsight does rescale any floating point image whenever it is opened.

No floating point image should be rescaled when opened. When this happens, it is because PixInsight does not know (and has no way to know) the numeric range to which the images are referred. In other words, this happens because PI doesn't know which values correspond to black and white. The FITS standard does not provide any way to specify this information for floating point images.

You could ask the authors of the application that has generated the images about the numeric range to which the pixel values are referred (the black and white points). There are applications that seem to store floating point images in the 16-bit unsigned integer range (0.0=black, 65535.0=white). There are other applications that seem to store images using arbitrary ranges that vary from image to image, which is terrible. PixInsight always stored floating point data in the [0,1] range. If you manage to know the actual range of your images, you can specify it in the FITS Format Preferences dialog (Default Floating Point Input Range), or you can use format hints in the ImageCalibration, StarAlignment and ImageIntegration tools.

If you are loading uncalibrated raw CCD images in PixInsight (e.g. with the ImageCalibration tool) then you should use 16-bit unsigned integer images. Then you should have no rescaling problems.

Hope this helps. Nice image by the way!

[Juan Conejero]

Any suggestions for a bloom removal technique in PixInsight?

The best way to fix bloomings is by rotating your camera so they can be rejected during integration. This replaces bloomings with actual data. I realize this recommendation is most appropriate for square sensors

You can use the CloneStamp tool (Painting category on Process Explorer). We usually use it to fix small bloomings and other minor artifacts. In difficult cases, I prefer to leave a blooming unmodified when I feel that I have to invent too much data to fix it, but this is my personal point of view.

[andyschlei]

Juan,

This is very helpful, thank you!

-- after all, PixInsight does rescale any floating point image whenever it is opened.

No floating point image should be rescaled when opened. When this happens, it is because PixInsight does not know (and has no way to know) the numeric range to which the images are referred. In other words, this happens because PI doesn't know which values correspond to black and white. The FITS standard does not provide any way to specify this information for floating point images.

You could ask the authors of the application that has generated the images about the numeric range to which the pixel values are referred (the black and white points). There are applications that seem to store floating point images in the 16-bit unsigned integer range (0.0=black, 65535.0=white). There are other applications that seem to store images using arbitrary ranges that vary from image to image, which is terrible. PixInsight always stored floating point data in the [0,1] range. If you manage to know the actual range of your images, you can specify it in the FITS Format Preferences dialog (Default Floating Point Input Range), or you can use format hints in the ImageCalibration, StarAlignment and ImageIntegration tools.

If you are loading uncalibrated raw CCD images in PixInsight (e.g. with the ImageCalibration tool) then you should use 16-bit unsigned integer images. Then you should have no rescaling problems.

Hope this helps. Nice image by the way!

I apologize for making a general assumption about what PixInsight did to floating point images.  I noticed this behavior when opening calibrated or integrated images I had created in CCDStack and generalized it to all openings of floating point images.   This was before the calibration, alignment, and integration were available in PixInsight.  My belief was reinforced because when you open a file in PixInsight and then go to close it, it gives a "this image has been modified" error even though nothing has changed.  I believe now this is because my set-up is configured to assign a default color profile to the image.

My workflow now is entirely within PixInsight, so I should not have a problem.  I start with 16 bit images but just accept the default of 32 bit floating point from there on out (disk space is cheap).  Since all the floating point images are created by PixInsight there should be no issues about rescaling.

I have to say again that the LRGB  tutorial was excellent.  The clear explanation of linear vs. non-linear images and why I should care was very helpful.  I've been using (as best as I can) PixInsight since the LE days and it brought my game up a big notch.  The color calibration tools are really good.

Thanks again and clear skies,

--Andy