Back to forums

Perfect stars from any type of telescope - "The Mosaic Flattener" ;)

Processing Deep Sky NGC1499 Nebula
11 replies•660 views
Piotr K. avatar
Piotr K. avatar
Hey guys,

Recently I developed a very simple, yet very effective method for getting perfect stars across the entire image, regardless of the telescope or photographic lens one is using I image with an achromatic refractor Startravel 80 (aperture 80 mm, focal length 400 mm, f/5), with the stock 1,25" focuser, so I can't even attach any flattener - and yet I'm able to achieve stars which look similar to stars from refractors at least 10x more expensive than my old achromat. Here's one example, NGC 1499:

NGC 1499 from an achromatic refractor SW 80/400 & Nikon D5100 mod (Piotr K.) - AstroBin

The trick is to shoot stars at f/10, and make a mosaic of panes, which serve as source of stars for the corners in the original image. Here are links to my video-tutorial showing the workflow. I'm using only free software, and I'm not using any A.I. tricks:

The Mosaic Flattener - part 1
The Mosaic Flattener - part 2
The Mosaic Flattener - part 3

Feel free to comment, suggest improvements, etc.!
Helpful Insightful Respectful Engaging
Tony Gondola avatar
Tony Gondola avatar
I have a few images on bin taken with the ST80, I use it as a guide scope now. It's a horrid little scope for imaging but it can be tamed by shooting narrowband. I haven't looked at your videos yet but the way I used to deal with the bad stars was to use Siril's star synthesis option, this is similar to what BlurX does. I would blend that with the real stars to make them acceptable. There was no need to make a mosaic to do it this way. I do think it's a great point that if all you can afford or have, you can still image with a fast achromat. It just takes more effort.
Helpful Supportive
Piotr K. avatar
Piotr K. avatar
Tony Gondola:
the way I used to deal with the bad stars was to use Siril's star synthesis option

I played with this star resynthesis option, but IMHO it produces "stars" which look very artificial. And using resynthesized stars as a mask for real stars, even shot at f/10, gives wrong colors of stars in the corners (since stars are elongated and not sharp, so their colors are distorted, too). Anyway, thanks for your input - I'll probably play with the resynthesis tool even more!
Tony Gondola avatar
Tony Gondola avatar
Piotr K.:
Tony Gondola:
the way I used to deal with the bad stars was to use Siril's star synthesis option

I played with this star resynthesis option, but IMHO it produces "stars" which look very artificial. And using resynthesized stars as a mask for real stars, even shot at f/10, gives wrong colors of stars in the corners (since stars are not sharp and elongated, so their colors are distorted, too). Anyway, thanks for your input - I'll probably play with the resynthesis tool even more!

I agree, I would never use the synth stars alone, it needs to be a blend to look good. If you need to retain original star color just use the synth as a luminosity layer for your original RGB stars, all color will be retained.
Helpful Concise
Piotr K. avatar
Piotr K. avatar
Tony Gondola:
If you need to retain original star color just use the synth as a luminosity layer for your original RGB stars, all color will be retained.

That is exactly the problem Stars in my original RGB image have distorted colors, because they are elongated, and different colors focus in different positions. If one applies the resynthesized stars as a luminance layer, these wrong colors will be retained. Below is an autostretched image from the f/5 stack. One wouldn't want to use these colors ;) (nor would one want to use colors from the f/10 version - they are similar, only the stars are more compact).

Helpful Insightful Respectful
Tony Gondola avatar
Tony Gondola avatar
Well, there are going to be compromises somewhere. For my own dual narrow band work I often just take the stars to pure mono before recombination. Not always, it just depends on the image.
Chris White- Overcast Observatory avatar
Chris White- Overcast Observatory avatar
Thats a clever idea.  Our brains tend to forgive other image deficiencies when the stars look nice and round.  Keep in mind, that your star shape and size is a good representation of what the optics are doing to your DSO as well.   So if you have mushy stars that you replace, the details within your nebula are also going to be mushy.

You may want to consider leaving the bad stars in the image for running your deconvolution to recover what you can in DSO details, and then replace the stars with your method.

I agree with you, your corrected result looks much better than the uncorrected version, so just that step of replacing stars does have value.
Helpful Insightful
Tony Gondola avatar
Tony Gondola avatar
It does, I don't work that way anymore but when the achromat was all I had, you do what you can.
Piotr K. avatar
Piotr K. avatar
Chris White- Overcast Observatory:
Keep in mind, that your star shape and size is a good representation of what the optics are doing to your DSO as well.   So if you have mushy stars that you replace, the details within your nebula are also going to be mushy.

Of course, and I'm well aware of this Still, I can choose targets which do not take the whole frame - e.g. almost all Messier objects fall within this category.

Anyways, what is also interesting, is that even though there is no flattener, the geometry of the corners is not distorted. Some time ago I checked positions of these elongated stars in the corners of my image of M31, taken with SW80-400 achro, with the positions of stars in a similar frame defined in Stellarium, and everything matches perfectly:



I was hoping that I can restore round shapes of stars by simply "squeezing" the corners closer to the central part by the Lens Distortion filter in GIMP (sorry, I don't know what better English word to use instead of "squeezing"... ;) ), but then the positions of stars become shifted. So, as for now, the "Mosaic Flattener" seems to be the best way to achieve realistic images from this little achromat without the flattener...
Helpful Insightful Respectful Engaging
Piotr K. avatar
Piotr K. avatar
Chris White- Overcast Observatory:
You may want to consider leaving the bad stars in the image for running your deconvolution to recover what you can in DSO details, and then replace the stars with your method.

Yeah, I was also thinking about this, however up to now I didn't find good settings in Siril's Deconvolution tool. The effects I obtain after deconvolution are very, very small...
Tony Gondola avatar
Tony Gondola avatar
The default settings in Siril for Decon are very gentle. Make sure you do a PSF from stars. Gradient step size is where you control the process. Start with something like 0.5 and work up and down until you get the effect you want. Also, Graxpert has added a very easy to use decon tool. Anther great tool to try is Cosmic Clarity tool from Seto Astro. There are lots of great options for this out there.
Helpful
Piotr K. avatar
Piotr K. avatar
Tony Gondola:
The default settings in Siril for Decon are very gentle. Make sure you do a PSF from stars. Gradient step size is where you control the process. Start with something like 0.5 and work up and down until you get the effect you want.

In my version of Siril, v. 1-2-4, the maximum value for gradient step size is 0,1. Sure I use PSF from stars Basically, I'm doing what they write in the user manual. What could be useful, though, is some source of practical information, I mean, what effect each setting really has, how it influences the result and the image. I have the impression that the user manual, at least the part regarding deconvolution, is written by a mathematician for other mathematicians. For them, all these mathematical and statistical terms are obvious, but for a normal user all these terms don't mean much... Anyways, I'll keep trying!