6 replies to this topic

[dylanear]

I tried my first attempt at imaging using the ASIAIR and got reasonably decent initial results for M31 Andromeda even in very poor conditions, very humid, hazy, partly cloudy, even with no flats, darks or calibration images of any kind, just multiple exposures and basic stacking. But upon actually looking all the resulting files on the ASIAIR today rather than just the jpgs exported to my phone during the night of observing, I was disappointed to find no stacked results other than JPGs, no high quality files other than the .FIT individual images. 

So I tried ASIDeepStack on my PC using the individual .FIT files and I got a very pleasing resulting image given the poor conditions during capture. So I save the results and get a JPG and a TIFF and the JPG opened in other software looks like the stacked results in ASIDeepStack, yet trying to open the TIF for fine tuning in Affinity Photo, or Capture One, or anything else I've tried I get a nearly black image with just a few stars barely bright enough to be seen. Trying to boost brightness a LOT and then getting black levels and contrast/gamma that resembles the JPG output has proven to be nothing but frustration. Affinity Photo gives a message saying it's assigning a sRGB color profile to the TIF on opening it because it has none defined in the file. Also none of the typical TIF file saving options are available like lossless LZW or ZIP compression so you get a rather huge uncompressed file?

Is this the same results others are getting from ASIDeepStack? Seems basically unusable for serious work, only useful for making basic 8bit JPGs with no options for the quality, file size even? Maybe this TIF color problem is a recent bug and not an issue that's plagued every version of the software?

And there's no saving the stacking/color settings from a session into a file or preset? 

This is the JPG output from ASIDeepStack. A reasonable looking JPG file.

 

 

 
This is the nearly all black TIF saved at the same time as the JPG, which unfortunately I don't think Cloudy Nights knows how to make a thumbnail of, or display in any way, maybe because it's lack of color profile and/or it's very minimal metadata? It looks NOTHING like the JPG perhaps because is has no color profile defined in it, likely no conversion was done from the internal data processing color space (linear float?) in ASIDeepStack to a screen/human vision compatible color profile at all like was done for the jpg? That may be a nice option to dump the raw data, but it's a HORRIBLE default for TIF output of some very basic software. The original tif had to be opened and saved in Affinity Photo to add LZW compression as the original TIF was uncompressed, 48 bit (16bit per channel) and 64mb, over the size limit for Cloudy Nights Gallery. I hopefully changed nothing but adding the lossless compression, Affinity forced a particular sRGB color profile onto it on import, but I exported with no color profile or metadata and at the same 16/48 bit depth. It looks the same when opened in Affinity Photo as the original. Apparently you can't view this in the gallery and there's no thumbnail, but if you right click and select "open image in another tab" it will download to your computer if you'd like to inspect it. I hope you can use that "open image in another tab" trick in this post too if you want to inspect this TIF. 
 

 

And here's the gallery to see the images there. 

https://www.cloudyni...eepstack-tests/

I'm trying my luck with DeepSkyTracker, but I'm having no luck getting a stack to have nearly the rather nice look the ASIAIR and ASIDeepStack gives me from the same .FIT files, but ONLY in the JPG file it saves, not the TIF??! The brightness, black level, gamma slider on the top right of DeepSky Tracker is HOPELESSLY imprecise, stays the same small, low precision size even if I full screen the interface on a 4K monitor. There's no numeric input or way to sue a larger, more precise slider so I can't get the fine control needed to get the right values from the FIT files out of the ASIAir. I know it's free software, but that's GLARING GUI design problem and a big limitation on what seems like great software, much more powerful than ASIDeepStack, that keeps it from giving as pleasing images as the simpler and very flawed ASI app?!!!
 

If anyone has a workflow for high quality, lossless, high bit depth stacks with wonderful, precise settings for controlling brightness, black levels, gamma and tones and colors in general, even if it's not especially cheap, I'm very interested! I guess I'm off to research the better options for stacking software now....

[rj144]

You need to stretch the data.  Look into Siril for stacking and processing.

[BQ Octantis]

Hi dylan,

 

Your first link says I do not have permission.

 

The problem you are dealing with is that FITS is meant to store linear data, while TIFF and JPEG are designed to store gamma compressed data. If you simply dump the linear FITS values into a TIFF or JPEG, the display driver of the operating system raises them to the power of 2.2 before displaying them onscreen. Since the values are between 0 and 1, ~squaring them makes them smaller and therefore darker.

 

To further confuse things, most FITS viewers will give you a "stretched" preview, which presents what the data could look like if an astrophotography "stretch" transformation is applied. But for the save from FITS to TIFF or JPEG to be WYSIWYG, you have to tell the FITS viewer that you want to see the linear data instead of the stretch preview.

 

I echo rj144's comment. Here is a basic Siril workflow:

 

https://siril.org/tu...ls/tuto-manual/

 

BQ

Edited by BQ Octantis, 10 November 2024 - 08:07 AM.

[dylanear]

Hi dylan,

 

Your first link says I do not have permission.

 

The problem you are dealing with is that FITS is meant to store linear data, while TIFF and JPEG are designed to store gamma compressed data. If you simply dump the linear FITS values into a TIFF or JPEG, the display driver of the operating system raises them to the power of 2.2 before displaying them onscreen. Since the values are between 0 and 1, ~squaring them makes them smaller and therefore darker.

 

To further confuse things, most FITS viewers will give you a "stretched" preview, which presents what the data could look like if an astrophotography "stretch" transformation is applied. But for the save from FITS to TIFF or JPEG to be WYSIWYG, you have to tell the FITS viewer that you want to see the linear data instead of the stretch preview.

 

I echo rj144's comment. Here is a basic Siril workflow:

 

https://siril.org/tu...ls/tuto-manual/

 

BQ

I see no way to grant or restrict that image specifically? The album is set to public? Try opening the image into a new tab (which downloads it when I do that on Chrome) from the album link? 

I think you are probably correct about the math, but it's not a FIT file that ASIDeepStack is outputting that I am having trouble with it's a TIF file, which can contain linear data, there's a billion ways to to write a TIF image, but typically they are log/gamma image files and should look like any other typical image format when opened in typical image editors, viewers just like, JPG, PNG, etc.

I'll give Siril another go. I tried it and couldn't get it to set up a sequence to stack without flats, darks, etc. Seemed to not have an obvious way to just stack a basic set of light images like ASIDeepStack allows. But, I should RTFM I suppose! Being script based, surely there's a way!

Edited by dylanear, 11 November 2024 - 04:21 AM.

[BQ Octantis]

It looks like there's just something weird about the first image is all. The album is indeed public and shows two images, but only one displays correctly from the album view:

 

 

You are correct on the TIF. It will gladly store the linear data as well as gamma compressed data—but you need the ICC to tell the OS the details. All standard image processors and OSes (indeed, the entirety of the Internet) assume the data is in gamma-compressed sRGB space. So if there is no ICC and the data is linear, then it will get crushed with a gamma correction RGB^2.2. As a workaround, you can use pixel math to gamma compress RGB^(1/2.2), so displayed you get (RGB^(1/2.2))^2.2 = RGB^(2.2/2.2) = RGB^1 = RGB.

 

BQ

 

[gsaramet]

 

 

If anyone has a workflow for high quality, lossless, high bit depth stacks with wonderful, precise settings for controlling brightness, black levels, gamma and tones and colors in general, even if it's not especially cheap, I'm very interested! I guess I'm off to research the better options for stacking software now....

Hi there!

 

Technically, it's called a PBCAC. It stands for Problem Between Chair And Computer.

 

The stacked file is a linear file. You want a stretched image. Therefore, you should stretch it (NOT preview it) and save it.

 

Workflow: Shoot calibration files: Bias (or dark flats), darks, flats. Feed the lights + calibration files to the stacking software. Stack the lights (hopefully, you won't have to calibrate by hand). (Auto)crop the stacked image. Plate solve it. Remove the gradients. Color calibrate it. Separate the stars. Stretch it. Apply whatever other corrections you feel necessary (usually a lot of fiddling with curves and masks). Stretch the stars. Recombine the image and save the final product. Somewhere in between creeps in noise removal and sharpening. 

 

In order to do those you need to learn how to use properly at least one software package. Which, generally, is not that intuitive, as you just found out. 

 

Best of luck!

Edited by gsaramet, 11 November 2024 - 08:32 AM.

[rj144]

It looks like there's just something weird about the first image is all. The album is indeed public and shows two images, but only one displays correctly from the album view:

 

Screen Shot 2024-11-11 at 6.20.11 AM.png

 

You are correct on the TIF. It will gladly store the linear data as well as gamma compressed data—but you need the ICC to tell the OS the details. All standard image processors and OSes (indeed, the entirety of the Internet) assume the data is in gamma-compressed sRGB space. So if there is no ICC and the data is linear, then it will get crushed with a gamma correction RGB^2.2. As a workaround, you can use pixel math to gamma compress RGB^(1/2.2), so displayed you get (RGB^(1/2.2))^2.2 = RGB^(2.2/2.2) = RGB^1 = RGB.

 

BQ

It's because you can't upload tif files to the gallery.