Starfixer: an online tool for star reduction, the correction of elongated stars and noise reduction.

From time to time I like to try out software that I come across that makes use of AI and has the potential for inclusion in my astronomical image processing workflow. StarFixer is a Deep Artificial Neural Network, trained to detect and fix elongated stars in astronomical images, it also reduces the sizes of stars. Starfixer can produce a processed image up to 4000x4000 pixels whilst preserving its aspect ratio. Within the past few days, a denoising funtionality has been added to Starfixer.

The Starfixer project dates back to 2020 and has steadily progressed to the present day.

The developer, Filippo, a 39 years old Italian (at the time of writing), is a computer engineer with a passion for artificial intelligence and astrophotography. He began studying AI when it was (in his own words) still considered futuristic and primarily for nerds.

Starfixer is quite responsive with a simple to use web interface. An image (in png (8 or 16 bit) or jpg format) is uploaded to the system. A link to the results page is provided. The speed of the process depends on the dimensions of the image. The results page has to be refreshed manually and when the processing is complete, the results from three models V1, V2 and V3 are provided to be seen in separate tabs as well as denoising results that are discussed below. You can examine and download any or all of the the three results as a png file. At the moment, whether the input image is 8 bit or 16 bit, the output image is 8 bit. Filippo tells me that he is working on full 16 bit compatibility.

Starfixer is freeware/donationware which is a form that I prefer. Donating the price of a cup of coffee now and again is a good way of supporting the project as well as keeping Filippo awake during the long coding nights.

A series of animations of before and after Starfixing images.

The propeller nebula with elongated stars. 

The elongated stars are also reduced

The Soap Bubble nebula showing star reduction

Reducing the stars gives more emphasis to the Soap bubble nebula.

The cloud Sculpting Cluster NGC 6823 and the nebula NGC 6820

A very busy, star-rich field where the stars obscure the nebulosity to an extent. Reduction of the stars reveals more of the structure of the nebulosity.

Testing the denoising functionality of Starfixer 

The test image is a SeestaS50 image comprising a 2 minute stack of 10s exposures. The stars are clear but the image is noisy.

When an image is uploaded to Starfixer at the moment; when processing is completed, the results page contains these options for viewing the results:

Animation of the original image, The original image denoised and the AI version 3 and denoised image

FULL SIZE ANIMATION

This image did not require star reduction therefore it is important that a denoised original image is provided as well.

A 17 minutes worth of 10s exposure image of M33 captured with a Seestar S50, stacked, debayered, SPCC and ADBE processed in PixInsight, stretched in GraXpert and Gimp, was submitted to Starfixer to examine its image denoising.

Animation of the original image of M33 and the Original denoised in Starfixer

FULL SIZE

Starfixer is a welcome addition to my suite of image processing tools. The enthusiasm of the Developer, Filippo will ensure that the software continues to improve over time and will be well worth trying as another option for any astroimager.

Starfixer can be accessed at starfixer dot org

CaK solar experiment with a Seestar S50

The equipment used

Baader solar filter OD 3.8 was placed at the end of a dewshield for the Seestar S50 and held in place with glue and plastic adhesive tape.

A filter holder for the Seestar S50 with a 2” to 1.25” filter adapter.

The solar filter assembly and the adapter fitted with a Baader Ca K-line double stacked filter 394 nm peak transmission with an 8nm bandpass.

The whole assembly viewed from the bottom.

The Ca K-line filter can be seen in place.

The whole assembly seen from the top.

The Seestar S50 with the Double filter assembly in place

There were initially two unknowns in this experiment:

1) Whether a 1.25” filter would cause vignetting

2) Whether the built-in UV/IR cut filter would be so aggressive that it would prevent the Ca K-line light from being detected by the sensor.

Optical density 3.8 Baader photographic grade solar filter was used because it transmits 16 times the amount of light that is transmitted by an OD 5.0 visual/photographic solar filter. This will give the best chance for the sensor to detect the CaK light if the UV/IR filter will allow it to pass.

As seen below, in a short opportunity between the clouds, the system worked well and revealed more detail of the photosphere than normal white light filtering. There was no noticeable vignetting. 

As this system works, it should also work with the Altair 2nm G-band Solar Contrast Filter 430.3 nm peak transmission.

Ca K-line image as seen on the Seestar S50 preview screeen

More work needs to be done to determine the best exposure setting when imaging with this setup. Because of clouds, very little time was available to optimise this. However, automatic exposure and gain seems to give a reasonable result.

The RAW AVI data were debayered and stacked in Autostakkert! with 1.5 drizzle (as the Seestar S50 is slightly under-sampled). The stacked image was wavelet processed in waveSharp, and post processed mainly in the Gimp 2.10.

Seestar S50 Ca K-line solar image August 9

The same situation occurred the next day on August 10 but with 252 debayered and stacked RAW AVI frames and post processing, another acceptable Ca K-line image was obtained 

Seestar S50 Ca K-line solar image August 10

The results of this experiment are very encouraging. Whilst the assembly holding the OD 3.8 Baader solar filter and the double stacked Ca K-line filter worked fine, it may be possible to build a more compact assembly.

It is not known if this system would work with Ca K-line filters such as the Antlia 3nm CaK filter or whether it would yield even better results. It remains for others with different bandpass Cak filters to try.

Once again, the Seestar S50 is shown to be a serious and capable 'smart' telescope. 

Feature release of AstroDMx Capture Version: 2.9.1

 Feature release of AstroDMx Capture Version: 2.9.1

Nicola has released a new version of AstroDMx Capture

Mutatis mutandis

Added: Exposure values can now be entered with floating point accuracy. For example, 3.25ms

Added: High full-well control for supported ToupTek derived cameras

Added: Low noise control for supported ToupTek derived cameras. This control is on by default, turning it off can significantly increase the frame-rates

Added: Taillight control for supported ToupTek derived cameras. This control is able to turn off the telltail lights on the back of the camera

Added: Mount sync function for INDI mount control (see release notes for more information)

Added: Native support for Mallincam (ToupTek-Derived. Linux and macOS only)

Added: Native support for Bresser (ToupTek-Derived. Linux and macOS only)

Added: Improved mount nudge functions. The mount can now be nudged, North-East, North-West, South-East and South-West. This is in addition to the usual North, South, East and West nudge functionality

Changes: Complete rewrite of the ToupTek implementation (see the release notes for more information)

Changes: Improvements to the INDI camera implementation (as well as bug fixes)

Changes: Complete rewrite of the INDI configuator (this is to facilitate the upcoming INDIGO implementation)

Changes: The busy indicator UI component no longer gets in the way of other UI components

Removed: Registax optimised AVI (see release notes for more information)

Updated: Altair SDK

Updated: Toupcam SDK

Updated: Omegoncam SDK

Updated: StarshootG SDK

Updated: Risingcam SDK

Updated: OGMA SDK

Updated: Atik SDK

Updated: SVBONY SDK

Updated: PlayerOne SDK

Bug fixes and other improvements

AstroDMx Capture download and further details can be found HERE

The Sun in Ca K-line light.

Two overlapping 3000-frame AVIs were captured with AstroDMx Capture for Windows through an f/5.5 Ekinox 80mm ED refractor fitted with a Baader OD 3.7 solar filter, with an SVBONY SC432M air-cooled CMOS camera, fitted with a Baader Ca K-line filter and a 5x Balow lens alone (giving 3x increase in focal length). 

The equipment used

AstroDMx Capture for Windows was used to capture the two, overlapping 3000 frame AVIs in 8 bit mono.

Flat frames were captured while the scope was pointing at the Sun. using a material that I have previously used for making H-alpha solar flatfields. It is 0.1mm thick, translucent thermoplastic polyurethane, frosted waterproof material. The material was stretched over the front of the Baader solar filter material and was held in place with a rubber band. The material was a low cost purchase from Amazon and proved to be a very effective diffuser.

The setup for flat field capture

The best 95% of the frames in each AVI were stacked in Autostakkert!4 with 1.5 Drizzle. The two resulting images were stitched in Microsoft ICE. The image mosaic was wavelet processed in waveSharp and post processed in GIMP and ACDSee.

Ca K-line image of the Sun

Full size image

Although the SC432M camera is intended for use with long focal length telescopes, this experiment shows that it can be used successfully with short focal length scopes and a Barlow lens. It must also be remembered that the diffuser material is NOT a solar filter and MUST be used in conjunction with the proper solar filter if the Scope is pointing at the Sun. However, it can be used without the solar filter only if the scope is pointing at a different part of the sky, well away from the Sun in order to capture flat fields.

William Optics 81 mm ED APO refractor with an Altair Hypercam 533C 14 bit OSC CMOS camera.

The equipment comprised a William Optics 81 mm ED APO refractor with a 0.8 flattener/reducer ED APO refractor with an Altair magnetic 2" filter holder with an Altair 6nm dualband filter Ha/OIII or an LPRO-MAX filter depending on the object being imaged; a  ZWO EAF and an Altair Hypercam 533C 14 bit OSC CMOS camera.  

The equipment used

The data were captured with AstroDMx Capture for Windows. The scope was mounted on an AVX GOTO mount which was controlled by AstroDMx Capture via an INDI server running on the imaging computer indoors.

The mount was placed on permanent marks on the ground which quickly gives quite a good polar alignment if care is taken with the placement of the tripod feet.

An SVBONY SV165 guide scope with a natively connected QHY-5II-M guide camera was used for PHD2 multistar pulse auto-guiding via the INDI server. The auto-guiding was controlled by a separate Linux laptop indoors.

Screenshot of autoguiding whilst imaging M13

AstroDMx Capture sent the scope/mount to a bright star which was used to focus the scope with a Bahtinov mask. The ZWO AEF was controlled by AstroDMx Capture via the INDI server.

Screenshots of the capturing process

Capturing RAW data on M13 showing the general green cast to the preview

The same preview with the non-destructive DMx White Balance turned on, producing a more realistic and pleasing preview

Screenshot of the preview whilst imaging M101

Screenshot of AstroDMx Capture whilst imaging the Chinese Dragon nebula showing a negative preview

Screenshot of the preview of the Chinese Dragon nebula with a normal preview with DMx White balance turned on

The exposures for the three objects were as follows:

M13: 60 minutes of 5 minute RAW exposures

M101: 35 minutes of 5 minute RAW exposures

The Chinese Dragon nebula, NGC 6559: 90 minutes of 5 minute RAW exposures

The data were debayered, calibrated, stacked and partly processed in PixInsight and post processed in Siril, GraXpert and Gimp 2.10 with Starnet++. 

M13

M101

NGC 6559, The Chinese Dragon nebula

RGB

HOO palette

These imaging sessions were for testing AstroDMx Capture following Nicola's major refactoring of the code for Touptek derived cameras, making the program more efficient and stable.

First light for an Altair Hypercam 533C and AstroDMx Capture

The equipment comprised a Stella Mira 66 ED APO refractor with a field flattener and Altair magnetic 2" filter holder with Altair 6nm dualband filters (Ha/OIII' SII/OIII); a  ZWO EAF and an Altair Hypercam 533C 14 bit OSC CMOS camera.  

The data were capture with AstroDMx Capture for Windows. The scope was mounted on an AVX GOTO mount which was controlled by AstroDMx Capture via an INDI server running on the imaging computer indoors.

The mount was placed on marks on the ground which quickly gives quite a good polar alignment if care is taken with the placement of the tripod feet.

An SVBONY SV165 guide scope with a natively connected QHY-5II-M guide camera was used for PHD2 multistar pulse auto-guiding via the INDI server. The auto-guiding was controlled by a separate Linux laptop indoors.

AstroDMx Capture sent the scope/mount to the star Arcturus which was used to focus the scope with a Bahtinov mask. The ZWO AEF was controlled by AstroDMx Capture via the INDI server.

The Altair Hypercam 533C OSC camera was chosen because like all Touptek derived cameras, it can produce true RAW images. That is, there are no destructive controls such as gamma and white balance applied before the data are saved. A result of this is that the RAW data have a green hue when viewed. This is perfectly normal because in the Bayer matrix of colour filters over the pixels, half of the filters are green whilst a quarter are red and a quarter are blue.

Although this can be distracting and even disappointing when previewing the data, AstroDMx Capture's non destructive DMx white balance control (which does not affect the save data) removes the green hue and correctly white balances the preview. The true RAW nature of the data means that no information has been lost during the capture process and everything has to be done during processing and post processing. An advantage of capturing RAW data is that the file size is only a third of the size of  fully debayered RGB data files. Some capture software can only capture RAW data but AstroDMx Capture gives the option of saving fully debayered, 16 bit data, and the user can choose the quality of the debayering algorithm used.

Screenshot of AstroDMx Capture capturing RAW fits files of M16, the Eagle nebula  but the non-destructive DMx white balance is turned on to correctly white balance the preview.

Capturing with the Ha/OIII dualband filter

Capturing with the SII/OIII dualband filter

Calibration frames were also captured

The two sets of data were debayered, cosmetically corrected, stacked  and partly processed in PixInsight and further processed in GraXpert, Gimp 2.10 and Starnet++.  The colour channels were decomposed to produce Ha and OIII data from one filter with SII and OIII from the other filter.

The starless OIII data were combined from both filters and composed back into various palette renderings in the Gimp. The SHO data were further selective colour processed in Photoshop CS2 and all of the palettes were post processed in the Gimp.

Two bicolour palette renderings of the Eagle nebula were made:

HOO palette

SOO palette

SHO Hubble palette

HOS, Canada, France, Hawaii telescope palette

 

Solar imaging

Using the same camera/scope setup with a Baader OD 3.8 solar filter and a UV/IR cut filter in the magnetic filter holder. AstroDMx Capture for Windows was used to send the scope to the Sun and to capture a 1200-frame RAW AVI of the whole solar disk.

Screenshot of AstroDMx Capture capturing RAW, 8-bit solar data

Notice the green hue of the RAW data.

Screenshot of AstroDMx Capture capturing RAW, 8-bit data but with the DMx white balance turned on

The data were debayered and the best 85% of frames stacked in Autostakkert!4, wavelet processed in waveSharp and post processed in the Gimp 2.10.\

The Sun in White Light

Fitting a dovetail bar to a Stella Mira 66 ED APO refractor

The Stella Mira 66 ED APO doublet refractor is a serious and high quality telescope with very good colour correction. However, it doesn't have mounting rings and instead has a ridiculous mounting shoe that could be used to attach it to a photographic tripod or maybe a tracking mount; but when it comes to mounting the scope on an astronomical mount such as the Celestron AVX, the shoe is outside of all concepts of good design. It can fit into the dovetail slot and be held by just one of the two holding bolts. A result of this is that it is virtually impossible to mount the scope parallel to the polar axis as the single screw pushes the back of the shoe to the east side of the slot, pushing the front of the shoe towards the west side of the slot. Moreover, this process also tends to push the back of the mounting shoe down and slightly raise the front of the shoe. The result is a scope that is not pointing true in either axis when in the home position. Also the scope feels very precarious.

There is a Stella Mira vixen style dovetail bar to which the scope is assumed can be attached. However this is not so. The shoe cannot be mounted directly onto the dovetail bar and provide a better balance for the scope because the dovetail bar will collide with the huge focus locking device. This means that spacers have to be improvised as there are none provided for the purpose. Even though the dovetail bar is actually printed with the Stella Mira name, it is just some sort of generic bar with a slot through which attaching 1/4" UNC photographic thread bolts can pass, but is far too wide and so allows the bar to be moved from side to side even when completely tightened down.

In order to solve all of these problems I enlisted the help of a couple of packing spacers, epoxy resin and some black Sugru mouldable glue.

A year ago in another article, I described the fitting of a ZWO AEF focuser to the scope.

The bracket for the focuser had to be modified but not the scope and the fitting of the AEF in no way exacerbated the problem of fitting a dovetail bar.

A thick packing spacer was attached by Epoxy resin to the base of the mounting shoe

A thinner spacer was then attache by Epoxy resin to the first spacer to provide the desired clearance

The packing spacers provided the required clearance from the focus locking device on the right hand side. Of course this does not have to be tightened as the rack and pinion focuser is held firmly by the AEF

\

Two 1/4" UNC photographic thread 1" length hex socket head bolts were used to attach the dovetail bar to the shoe and Epoxy resin was also used between the dovetail bar and the packing spacers.

Black Sugru mouldable glue was used to pack around the screws and between the two rails of the dovetail bar. After a couple of days the Sugru has become hard so that no lateral or twisting motion is possible

The completed modification

The Stella Mira mounted on the AVX mount much further forward than was previously possible providing much improved balance with true and rigid positioning.

The scope can now function the way that we require for our work.

The modified scope in action