If you are waiting for a Draco to arrive, so am I. Mine, the SHO edition, should be here soon, and until it is, the Draco data I can work with comes from samples DWARFLAB shares from its own units. This is the third of those datasets I have written up, and this time the target is Caldwell 34, the Western Veil in Cygnus, which my DWARF 3 has already given me 4 hours 30 minutes on.
That makes this a practical question for anyone waiting on a Draco: what does the extra aperture change on a target you may already have shot with a DWARF 3? Below are both images at matched scale, the numbers behind the difference, and every Siril step each file went through, read from the FITS headers so you can repeat them on your own data.
It is not a controlled test. The two stacks come from different skies, different months and, this time, different processing paths. Once my own Draco is here I will shoot the same targets from the same Bortle 6 driveway on the same nights and process both files identically. Until then, this comparison shows the gap in scale and detail to expect, and the workflow carries over to the first Draco files you capture yourself.


The first two were the Eastern Veil in September and the Heart Nebula in October. The Heart post ran both stacks through identical steps. This one did not, and the processing section shows where the two paths split and what each step does.
The tale of the tape
Capture settings come from the two stacked FITS headers. Aperture, focal ratio and filter width come from the published specs, image scale and field are calculated, and the sky conditions are my own record and DWARFLAB’s.
| DWARF 3 | Draco | |
|---|---|---|
| Aperture | 35mm | 90mm |
| Focal length | 150mm, f/4.3 | 340mm, f/3.8 |
| Filter | built in Duo Band, about 30nm | built in Duo Band, 13nm |
| Subs | 270 × 60 s | 31 × 300 s |
| Gain | 90 | 120 |
| Total | 4 h 30 m | 2 h 35 m |
| Sensor temperature | 30°C | minus 7°C |
| Mount mode | equatorial | alt az |
| Image scale | 2.79″ per pixel | 1.45″ per pixel |
| Field as processed | 2.78° × 1.67° | 1.65° × 1.24° |
| Sky | Bortle 6 | Bortle 3 (per DWARFLAB) |
| Session ended (UTC) | Jul 26, 2026 | Oct 18, 2026 |
Two header details are easy to miss. My DWARF 3 file records 16,200 seconds of exposure but no frame count, so the 270 subs are that total divided by 60. The Draco header names its target HD 335129 instead of C 34, and the stacked file already carried a plate solution when it reached me.
Round one: light and resolution
Collecting area grows with the square of the aperture, so 90mm against 35mm gives the Draco 6.6 times the area. Its 2 hours 35 minutes would take the DWARF 3 about 17 hours to equal on area alone. I gave it 4 hours 30 minutes, which puts the small scope at roughly a quarter of the Draco’s photons (26 percent). Treat that as a rough number. The sensors and filters differ, and the Draco is a catadioptric design: Astroshop lists a 40mm secondary mirror. If that mirror is the whole obstruction, the area ratio drops to about 5.3, the DWARF 3 would need about 14 hours, and my 4 hours 30 minutes comes to about a third of the Draco’s photons.
Image scale is the second number to know, and you can work it out for any setup:
arcsec per pixel = (206.265 × pixel size in µm) / (focal length in mm)
For the DWARF 3 that gives 2.75 (Siril’s plate solve measured 2.79, because it solved the focal length at 147.9mm). For the Draco it gives 1.45, from its 2.394 micron pixels at 340mm (the sensor’s 1.197 micron pixels, binned 2 × 2 for deep sky). Each Draco pixel covers about half the sky width of a DWARF 3 pixel, a factor of 1.92.
The optics set a matching ceiling. The Dawes limit, 116 divided by the aperture in millimeters, is 3.3 arcseconds for the DWARF 3 and 1.3 for the Draco. At 2.79 arcseconds per pixel the DWARF 3’s pixels sit close to its 3.3 arcsecond limit, so lens and sensor both cap its detail near 3 arcseconds, and deconvolution cannot restore what neither recorded. The Draco’s 1.45 arcsecond pixels sit just as close to its 1.3 arcsecond limit. On most nights the atmosphere blurs stars to more than 1.3 arcseconds anyway.
Round two: the field
On the Heart Nebula the DWARF 3’s wide field won the round outright. The Western Veil plays out differently. My full DWARF 3 frame, trimmed only at the stacking edges, covers 2.78° × 1.67°, about 2.3 times the sky of the Draco’s 1.65° × 1.24°. I rotated the DWARF 3 image 90 degrees in the final edit, which lines it up with the Draco frame and turns it into a portrait, and across the width of the nebula the two spans are almost identical: 1.67° against 1.65°.
All of the extra sky lands on the other axis. Most of it is the upper third of my DWARF 3 frame, which holds stars only. A narrower strip along the bottom, also outside the Draco’s view, picks up a few faint wisps of nebula. In both images the bright arc and the fainter filaments below it still run off the left and right edges. For this target the Draco’s frame holds nearly all of the nebula the DWARF 3 recorded, at almost twice the scale. For the pair at the top of this post I cropped the DWARF 3 image to the Draco’s field, so both show about the same patch of sky.
Round three: the filaments
To compare detail fairly, I cut the same patch of sky from both final images and enlarged the DWARF 3 crop by the 1.92 scale ratio from round one. The patch centers on the bright ridge just to the right of 52 Cygni.

In the Draco crop the ridge separates into individual strands, teal oxygen emission along the top edge of the ridge with red hydrogen alpha trailing behind it, and faint wisps show above the ridge where the DWARF 3 records only background. The DWARF 3 version traces the same ridge with the same bends, as one wide ribbon. A violet cast sits over its nebula and its stars, and the stars are larger and softer.
Some of that gap is the aperture and image scale from round one. Some of it is processing, because the two stacks did not go through the same steps, and from the finished images alone I cannot say how much of the color difference belongs to each. The next section lists what each file went through.
Same software, two paths
Siril writes a HISTORY line into the FITS header for most operations, so the order below is read from the files themselves. I’m still learning Siril. My recent M51 Through Three Telescopes project, where I processed my own DWARF 3 stack next to public data from Hubble and JWST, has taught me a lot, and I applied it here. Both stacks went through Siril 1.4.5 on November 19 and finished in Affinity 3.3.3.
DWARF 3, 270 × 60 s:
- Crop the stacking edges (3856 × 2180 down to 3586 × 2150)
- Plate solve
- Spectrophotometric color calibration (SPCC)
- Gradient removal
- Deconvolution, three passes
- NL-Bayes denoise at full strength
- Separate stars from the nebula with StarNet
- Denoise the starless layer again
- Stretch the starless layer: a black point move, then six GHS passes
- Saturation boost on the starless layer
- Stretch the star layer with GHS
- Recombine stars and nebula
- Background neutralization
- SCNR green noise removal
- Cosmic Clarity sharpen, then Cosmic Clarity denoise
- Final adjustments and export in Affinity 3.3.3, including a 90 degree rotation
Draco, 31 × 300 s:
- Gradient removal (no crop; the full 4096 × 3072 frame stays)
- Photometric color calibration (PCC)
- Deconvolution, four passes
- NL-Bayes denoise at about half strength
- Separate stars from the nebula with StarNet
- Stretch the starless layer: two black point moves and six GHS passes
- Stretch the star layer with asinh
- Recombine stars and nebula
- Two saturation adjustments
- Cosmic Clarity sharpen, then Cosmic Clarity denoise
- Final adjustments and export in Affinity 3.3.3
Where the paths split
| Step | DWARF 3 | Draco |
|---|---|---|
| Color calibration | SPCC, before gradient removal | PCC, after gradient removal |
| Deconvolution | 3 passes | 4 passes |
| Denoise | full strength, then again on the starless layer | about half strength, once |
| Stars | GHS stretch | asinh stretch |
| Final color cleanup | background neutralization and SCNR | two saturation adjustments, no SCNR |
What each step does
Cropping comes first because the ragged edges left by stacking alignment throw off gradient tools. Plate solving matches the stars to a catalog, which gives the exact pointing and image scale; both color calibration tools need it.
PCC measures star colors in the image and scales the red, green and blue channels until those stars match catalog photometry. SPCC has the same goal but models the spectral response of the sensor and filter against Gaia star spectra. Here the DWARF 3 got SPCC and the Draco got PCC, so the two images were not color calibrated the same way.
Gradient removal fits a smooth model of the background (skyglow, moonlight, leftover vignetting) and subtracts it. Both files used Siril’s automatic gradient removal with the same settings.
Deconvolution estimates how the optics and atmosphere spread a point of light into a star, then partly reverses that blur. It runs on linear, unstretched data. Each extra pass sharpens further and raises the chance of dark rings around stars.
NL-Bayes is Siril’s main noise reduction. Its modulation setting blends the result with the original: 1.0 keeps all of the denoised version, 0.5 keeps half. The DWARF 3 stack got full strength plus a second pass on the starless layer; the Draco stack got 0.505, once.
Star separation with StarNet writes two files, a starless image and a star layer, which is why the folder holds starless_ and starmask_ copies of each crop. The nebula can then be stretched hard without bloating the stars.
GHS, the generalized hyperbolic stretch, brightens around a chosen pivot level: the amount sets how strong the stretch is, and the local setting concentrates the added contrast near that pivot. Several small passes at different pivots, as both files show, lift faint filaments while holding back the bright ridge. The asinh stretch used on the Draco stars brightens faint stars while keeping their color ratios.
Background neutralization sets the empty sky to a neutral gray, and SCNR removes the excess green that dual band data often carries. Cosmic Clarity, the machine learning sharpening and denoising tool from Seti Astro, ran last on both stretched images, and both runs are logged in the Siril history.
Experimenting with Cosmic Clarity
Both final images went through Cosmic Clarity sharpening and denoising. I also experimented with its super resolution mode on the DWARF 3 stack, which doubled it from 3520 × 2084 to 7040 × 4168 pixels; that version did not make the final image. I have mixed feelings about Cosmic Clarity, because I don’t really know what the AI model does to the data.
If you are waiting on a Draco
A few things in these files will apply to the Draco data you capture yourself.
- Field: 1.65° × 1.24°. The Western Veil’s filaments ran off both side edges of the Draco frame, so for large nebulae plan for a mosaic or keep the DWARF 3 for the wide view.
- Storage: one Draco frame processed at 32 bits is a 151 MB FITS file, against 92.5 MB for my cropped DWARF 3 frame. The Draco files for this one target take up about 1.2 GB of my folder.
- Mount mode: the header records equatorial mode as off, so these 300 second subs were shot in alt az.
- Image scale: 1.45 arcseconds per pixel, against 2.79 on the DWARF 3.
- Sample data: links to DWARFLAB’s sample datasets can be found in the Draco Facebook group, if you want to try these steps yourself.
Two scopes on one ridge
The Draco data resolves the Western Veil into strands that the DWARF 3 data, at 1.92 times coarser scale and a quarter to a third of the light, shows as a single ribbon. On this target the DWARF 3’s wider field added mostly empty star field. Part of the color difference is mine, since the two stacks did not go through identical steps.
My own Draco should be here soon. The comparisons I run with it will put both scopes in the same driveway on the same night, through the same Siril steps, starting with targets like this one. To a degree, the DWARF 3 data is remarkably close to the Draco’s, except in resolution. I like both images, and there is room for improvement in the processing. Stay tuned.
Clear skies!
AK
About the Draco data
Data supplied by DWARFLAB. Target: Caldwell 34, the Western Veil Nebula (NGC 6960). Captured with DRACO using the built in Duo Band filter, under Bortle 3 skies, with 300 second exposures at gain 120 and 2 hours 35 minutes of effective integration (31 frames). Independently processed by AK in Siril 1.4.5 and Affinity 3.3.3; main steps: gradient removal, photometric color calibration, four passes of deconvolution, NL-Bayes noise reduction, StarNet star separation, GHS stretch of the starless layer, asinh stretch of the stars, recombination, saturation, Cosmic Clarity sharpening and denoising, and final adjustments in Affinity. I did not physically operate the unit for this evaluation.
Frequently Asked Questions
What is Caldwell 34?
NGC 6960, the western arc of the Veil Nebula supernova remnant in Cygnus. It runs past 52 Cygni, the bright star sitting on the ridge in both images.
How long would the DWARF 3 need to match 2 h 35 m on the Draco?
About 17 hours, counting collecting area alone (90mm against 35mm is 6.6 times the area). Sensor, filter and optical differences move that figure in either direction.
Why is the DWARF 3 image cropped?
Its full frame covers about 2.3 times the sky of the Draco’s. I rotated it 90 degrees to line its filaments up with the Draco frame, then cropped it to the Draco’s field so both images show about the same patch of sky.
Who captured the Draco data?
DWARFLAB captured it on a Draco and supplied the data. AK processed it independently and did not operate the telescope.


Leave a Reply