Page 1 of 2

All white

Posted: Sun Jun 10, 2018 8:15 pm
by Star Dad
Hi all -
I'm a newbie to Star Tools and have been reprocessing some of my pictures. I bought ST because I was able to pull M-51 out of the photos so much better than Nebulosity could. But In It and almost all others I've reconstructed (using the original .fit files and stacking only) I am seeing only white stars. No colors at all. The color tool can change *all* the stars - but it's an overall effect. Using Nebulosity's Digital Development I get blue, red and sometimes green stars. I've follow the tutorial on the main site here at ST.org, and one other a fellow posted as much as I can to the letter.

In desperation I tried redoing M-57 the Ring Nebula which Nebulosity's DDP pulled out as a beautiful mid toned blue-green. I don't have the original .fit files so I used the preprocessed and pre-stacked image (No DDP). Doing the step by step, ST kept the colors well, until I hit the Decon step when it attempted to make it gray. Being careful about it I managed to get the final output to look like Nebulosity.

The curious thing - and perhaps why I'm having trouble is that the original loading of the images and doing the initial AutoDev for all pics except M-57 are just a gray scale images. M-57 showed a yellowish image much like the tutorial here. Everything I've read says that I should stack the images and not do any other pre-processing. IS this gray image vs yellow/red of the tutorial a phenomenon of the latest ST or am I doing something wrong? Any idea why I am losing the star colors during the initial AutoDev? - or somewhere else down the line of processing the image?

Re: All white

Posted: Tue Jun 12, 2018 1:56 pm
by admin
Star Dad wrote:Hi all -
I'm a newbie to Star Tools and have been reprocessing some of my pictures. I bought ST because I was able to pull M-51 out of the photos so much better than Nebulosity could. But In It and almost all others I've reconstructed (using the original .fit files and stacking only) I am seeing only white stars. No colors at all. The color tool can change *all* the stars - but it's an overall effect. Using Nebulosity's Digital Development I get blue, red and sometimes green stars. I've follow the tutorial on the main site here at ST.org, and one other a fellow posted as much as I can to the letter.

In desperation I tried redoing M-57 the Ring Nebula which Nebulosity's DDP pulled out as a beautiful mid toned blue-green. I don't have the original .fit files so I used the preprocessed and pre-stacked image (No DDP). Doing the step by step, ST kept the colors well, until I hit the Decon step when it attempted to make it gray. Being careful about it I managed to get the final output to look like Nebulosity.

The curious thing - and perhaps why I'm having trouble is that the original loading of the images and doing the initial AutoDev for all pics except M-57 are just a gray scale images. M-57 showed a yellowish image much like the tutorial here. Everything I've read says that I should stack the images and not do any other pre-processing. IS this gray image vs yellow/red of the tutorial a phenomenon of the latest ST or am I doing something wrong? Any idea why I am losing the star colors during the initial AutoDev? - or somewhere else down the line of processing the image?
Hi,

It would be really useful if you could upload the data somewhere so we can diagnose the issue.

From your description where you say you don't have the original FITS files, I'm not sure whether you are truly processing a linear image or not. Indeed, what you have read is correct; you should just stack the image and avoid any furtrher processing (even white-balancing if you can).

Generally, when processing a linear stacked data set, you shouldn't worry about coloring at all until you use the Color module (which I would recommend be the last step, before switching Tracking off). Due the Tracking "time traveling" nature of the processing engine, the color module will go back in time when the data was still linear, extract the colors from the linear stack and forward-propagate what those colors would look have looked like as if they had not been stretched along with the luminance/brightness channel (thereby preserving much more coherent coloring, independent of brightness).

If the coloring information is no longer present (or has been tampered with) when you first load the dataset into StarTools, color recovery will fail.

To reiterate, StarTools effectively decouples color processing from your luminance ("detail") processing. It does not stretch and squash the color information along with the stretching and squashing of the luminance information as you try to bring out detail. Ergo, StarTools should be showing more star colour than other programs, rather than less. I suspect something might be amiss with the data you are importing.

Would love to have a look!


That said,

Re: All white

Posted: Tue Jun 12, 2018 10:55 pm
by Star Dad
Thanks for the Info. I was up all last night taking new photos. Again the processing shows nothing but white. I will use your suggestion concerning delaying using the color module. If that fails, I'll post the image on my website and you can have a look-see. I'll bin it down as the file is huge. I'm a zombie right now having gotten only a couple of hours of sleep last night (I just can't let a good night slip by they are so rare). So I'll post tomorrow one way or the other.

Re: All white

Posted: Wed Jun 13, 2018 2:23 am
by admin
Star Dad wrote:Thanks for the Info. I was up all last night taking new photos. Again the processing shows nothing but white. I will use your suggestion concerning delaying using the color module. If that fails, I'll post the image on my website and you can have a look-see. I'll bin it down as the file is huge. I'm a zombie right now having gotten only a couple of hours of sleep last night (I just can't let a good night slip by they are so rare). So I'll post tomorrow one way or the other.
I'd be more interested in the data right after it has been stacked - there is unfortunately much less to glean from looking at a processed image.
What camera are you using? If you are using a DSLR, are you definitely shooting CR2s (or NEFs) and not JPEGs?

Thanks!

Ivo

Re: All white

Posted: Thu Jun 14, 2018 11:33 am
by Star Dad
http://mcleanresearch.com/final_49_stacked.fit is the file location. You may have to right click and download it - it'll probably come up as a text file, but it is a .fit file.

This is *only* stacked in Nebulosity - 49 subs - and then I binned it down to make it smaller - it's 2M as it is. I did NO other processing - not darks,flats, etc.

I tried the suggestion of moving the color to just before untracking and it made no difference. my work flow:

Code: Select all

open
crop image
wipe: autodev temp y
         agg=99%
         dop = 18
         pre=256x256
         D.A. = 5 pixel
Dev - DA 5
        DD = 75% home in to about 82%
Decon - Radius 2.3
Sharp - size = Large
            small detail = 90%
HDR - soft
Contrast - dark background - just a tweak
Life - Isolate
         Sat = 125%
         strength 50%
color - default - nothing but white stars
tracking - off - default
If you get different results or a different work flow I am super interested to know what you did.

I am using a Canon 70D - 1600 ISO -. In Nebulosity it's saved as a .fit file - normally though it would be a CR2.
I am using ST 1.3 - which is the latest AFAIK

I should mention that in Nebulosity using the Demosaic function I do get colored stars.

And finally - many thanks for helping me!

Re: All white

Posted: Thu Jun 14, 2018 12:53 pm
by admin
Star Dad wrote:http://mcleanresearch.com/final_49_stacked.fit is the file location. You may have to right click and download it - it'll probably come up as a text file, but it is a .fit file.

This is *only* stacked in Nebulosity - 49 subs - and then I binned it down to make it smaller - it's 2M as it is. I did NO other processing - not darks,flats, etc.

I tried the suggestion of moving the color to just before untracking and it made no difference. my work flow:

Code: Select all

open
crop image
wipe: autodev temp y
         agg=99%
         dop = 18
         pre=256x256
         D.A. = 5 pixel
Dev - DA 5
        DD = 75% home in to about 82%
Decon - Radius 2.3
Sharp - size = Large
            small detail = 90%
HDR - soft
Contrast - dark background - just a tweak
Life - Isolate
         Sat = 125%
         strength 50%
color - default - nothing but white stars
tracking - off - default
If you get different results or a different work flow I am super interested to know what you did.

I am using a Canon 70D - 1600 ISO -. In Nebulosity it's saved as a .fit file - normally though it would be a CR2.
I am using ST 1.3 - which is the latest AFAIK

I should mention that in Nebulosity using the Demosaic function I do get colored stars.

And finally - many thanks for helping me!
Thank you for uploading that! It seems that the FITS file only contains one layer (hence looking B&W) - there are no separate red, green and blue channels. This is consistent with the file size; 138 * 928 * 16-bits + a small header = 2.6MB.
My best guess is that something is amiss in your debayer (demosaic) settings in the stacker. It appears the subs are simply not being debayered correctly (or at all?) before aligning and then stacking. The Demosaic function is meant for single Bayered subs only (it's not meant to be used on stacks!). On multi-sub stacks it will simply introduce color artefacts (usually blue and orange or purple and green); these may look to you like "colored stars", but they would just be artifacts.

As a side note, ST 1.4.332 is the latest version as of this writing. It's a "bleeding edge" version, but comes with some cool new features, improvements and bug fixes. It's officially still alpha, but it's the daily driver for many (myself included).

TL;DR; find the right debayering settings for your 70D in Nebulosity and you should be sweet! :)

Re: All white

Posted: Thu Jun 14, 2018 1:26 pm
by Star Dad
Ah Ha! OK, so I *do* need to demosaic all images before stacking. Will do and post results.

Re: All white

Posted: Fri Jun 15, 2018 12:26 pm
by Star Dad
I redid the file and once again - all white. So I thought I'd try M-52 which shows yellow, red, and some blues in a processed Nebulosity image. I binned it down to save transmission size and processed it with ST. Again - nothing but white. I followed the above guide. What am I doing wrong? This image has been pre-processed with darks and de-mosiac each file before stacking.

Here's the image: http://mcleanresearch.com/M52_binned.fit

Re: All white

Posted: Sat Jun 16, 2018 5:00 am
by admin
This time there are 3 channels in the FIT file. However, they seem to be encoded as 3 separate datasets.
I believe this is specific to how ImagesPlus needs its FITS files.

I just downloaded a demo version of Nebulosity 4.1 for any clues with regards to how it saves FITS files.
Could you try a different setting in Nebulosity under Preferences > Output > Color file format (the RGB FITS Maxim setting might be more compatible with StarTools).
Worst case Nebulosity will also allow saving RGB as 3 separate files, which you can import using the LRGB module as red, green and blue.

Let me know if this helps!

Re: All white

Posted: Sat Jun 16, 2018 7:13 pm
by Star Dad
OMG! WOW! Here's a link to the binned image: http://mcleanresearch.com/M52.tiff :thumbsup:

So the answer is: set Nebolisty's Color file format in preferences to RGB Fits: Maxim.

So now I see I have some hot color pixels - Guess I need to retake my darks, etc.

All I can say, is thank you so much for helping me through this. I was regretting spending the money for ST, but now I am convinced I did the right thing. I have SOOOO many old pictures to go and redo. I've saved everyone of the "final" images (those that were pre-processed in Nebulosity before stretch, etc) that I've taken over the past couple of years. If we ever meet - I owe you a beer/wine/drink of your choice. :bow-yellow: