Theli an PI Channel Combination

[frasax]

Hello,

i use Theli for my data reduction, and start imageprocessing with already color calibrated Fits Files, out of Theli.
The Channel Combination always fails, and produces very bad looking RGB Images, with totally wrong colors. It seems like PI interprets the Fits Files somehow wrong.

I have to add, that Theli gives out Fits Files with WCS headers and Plate Solvings. These look much different then standard Fits Headers.

What could be the reason for this fail, and what is a possible solution. Right now i open my single R G and B Fits Files in a third party software, combine them to RGB and then open this one in PI.

CS Frank

[Juan Conejero]

what is a possible solution

Use PixInsight for data reduction.

[frasax]

Hi Juan,

maybe i do from time to time in the future, but for now i don´t change my workflow. I´m very happy with Theli and it is hard to beat by any software speaking of quality.

Any other suggestions?

CS Frank

[Juan Conejero]

None.

[vicent_peris]

Hi,

Could you please upload the color image you generate in PixInsight?


Thanks,
Vicent.

[frasax]

Hi,

Could you please upload the color image you generate in PixInsight?


Thanks,
Vicent.

Hi Vicent,

i will, but not before Monday, as i´m not at Home.

Just a suggestion: I suspect the Format Settings more, than the Algorithm of the Channel Combination Module. What about the Override Embedded Settings? Like ICC Profile for example?

CS Frank

[Lateralus]

Hi,

I suffer from exactly the same problem.
You can download a set of files here: www.albireo.at/temp/color_theli.zip

Using for example MaximDL to channel combine the images (Bgd Auto Equalization disabled and channel weights set to 1:1:1), I get a proper image with correct colors.

Playing around with the ICC Profile Setting in the Format Explorer does not change anything for me.

Thank you for investigating,
Michael

[vicent_peris]

Hi,

I suffer from exactly the same problem.
You can download a set of files here: www.albireo.at/temp/color_theli.zip

Hi,

There isn't anything wrong in these images. And this for sure not a fail in our tool:



You cannot expect to combine three images into a color one and have the right color by pure magic. As you see in the screen copy above, the sky background is brighter in the blue and green channels. The only thing you need to do is to uncheck the "Link RGB channels" option in STF and the press the Auto Stretch button to readjust the screen contrast.

In astrophotography, you'll find always two different color biases. One is additive: the sky background in each color. The second is multiplicative: the white balance. The first is corrected with BackgroundNeutralization while the second is corrected with ColorCalibration. Please look for information about these tools because they are the very basic tools to correct color in PixInsight.

As I told, there is no fail in our side. I didn't investigate if the FITS image was correctly written by Theli or if Theli introduces any kind of pedestal during the preprocessing. But, at the end, the pragmatic solution is to use our tools correctly to get the right color.


Best regards,
Vicent.

[Lateralus]

Theli is doing the right calibration already based on star catalogues so you don't have to unlink the channels.
The background is already without color and the stars and objects with the proper color calibration.
I can for example use MaximDL to channel combine (no weighting, calibration, multiplying of the 3 channels) the 3 channels, open this image in PI and it will look perfect after an auto-stretch with the channels linked together.

The brighter skybackground in B happens only if you open the image in PI.

I have attached an example
Left: 3 FITS files combined in PI and auto-stretched with linked channels
Right: exact same 3 FITS files combined in MaximDL, saved as FITS, opened in PI and auto-stretched with linked channels


You can download the Maxim-RGB-combined FITS file here: www.albireo.at/temp/RGBMaxim.zip

[Juan Conejero]

Playing around with the ICC Profile Setting in the Format Explorer does not change anything for me.

Because color management has absolutely nothing to do with color calibration of linear calibrated data. You are mixing completely unrelated concepts. Color management is all about consistent reproduction of color on different media and devices. It has nothing to do with the relative distributions of pixel values among color components.

[Lateralus]

Because color management has absolutely nothing to do with color calibration of linear calibrated data. You are mixing completely unrelated concepts. Color management is all about consistent reproduction of color on different media and devices. It has nothing to do with the relative distributions of pixel values among color components.

Thank you for pointing that out, it was just an act of pure despair in the middle of the night while I was next to tears 

[Juan Conejero]

Theli is doing the right calibration already based on star catalogues so you don't have to unlink the channels.

Well, it has just applied a G2V white balancing procedure. Whether this is "correct" or not is questionable. In my opinion, it is conceptually wrong.

[Juan Conejero]

To open these images in PixInsight you have to define a custom range of floating point data. Do the following:

- Open Format Explorer.

- Double click on FITS

- On the FITS Format Preferences, enter 1000 in Default Floating Point Input Range, Upper range parameter.

- Click OK

Open the images and combine them. That's all. The problem is, as always, the FITS format. The data have been written in an arbitrary range that is unknown because FITS does not provide a standard way to specify it. The result is endless interoperability problems. That's why we have created XISF. But you probably don't care about this.

[Lateralus]

Well, it has just applied a G2V white balancing procedure. Whether this is "correct" or not is questionable. In my opinion, it is conceptually wrong.

I can remember that you have explained your thoughts behind this statement already somewhere.
Do you maybe have a link to this post? I can't find it, but I'm eager to read it.

- Open Format Explorer.

- Double click on FITS

- On the FITS Format Preferences, enter 1000 in Default Floating Point Input Range, Upper range parameter.

- Click OK

Sorry this does not work for me, maybe I made the wrong changes in my settings?

[Juan Conejero]

Do you maybe have a link to this post? I can't find it, but I'm eager to read it.

This post describes our color calibration methodology:

http://pixinsight.com/forum/index.php?topic=2542.0

This post is from 2010. We have accumulated a lot of experience in this field since then. Maybe Vicent can chime in and comment about this.