A comparison of BXT/NXT vs SyQon

Daniel J. Givens · Jun 15, 2026, 08:03 PM

Also, this isn’t great data. For other targets or cleaner data, Parallax and Prism could be fantastic. I’ll continue to play with it.

This is what i have found also. I think if you have very clean data, Prism has potential. If not it just seems to emphasize defects in the data. Parallax works as advertised but has a completely different look from BXT. I actually get the best result by blending both results together.

I’ve been playing with Parallax this afternoon using the Lum master of a variety of targets. First up was an emission nebula I’m currently working on, SH2-254, and the additional detail Parallax brought out over BXT was striking. I then tried Parallax on another current project, the galaxy NGC 3184, and again the difference was striking, but this time BXT was the winner. BXT also did better with the Draco Triplet, but Parallax outperformed BXT on NGC 4725. It seems when the target is small relative to the image resolution, BXT brings out more detail, but Parallax brings out more with larger targets. In fact, where Parallax did better with 4725, in the same image BXT exceeded Parallax with the much smaller galaxy NGC 4712. FWIW, I’m shooting at 2350mm (9.25 SCT) and an image scale of 0.33”/pixel, but very seeing limited and way over sampled.

Cheers,
Scott

Scott Badger · Jun 17, 2026 at 12:23 AM

I’ve been playing with Parallax this afternoon using the Lum master of a variety of targets. First up was an emission nebula I’m currently working on, SH2-254, and the additional detail Parallax brought out over BXT was striking. I then tried Parallax on another current project, the galaxy NGC 3184, and again the difference was striking, but this time BXT was the winner. BXT also did better with the Draco Triplet, but Parallax outperformed BXT on NGC 4725. It seems when the target is small relative to the image resolution, BXT brings out more detail, but Parallax brings out more with larger targets. In fact, where Parallax did better with 4725, in the same image BXT exceeded Parallax with the much smaller galaxy NGC 4712. FWIW, I’m shooting at 2350mm (9.25 SCT) and an image scale of 0.33”/pixel, but very seeing limited and way over sampled.

Cheers,
Scott

Thank you very much! These comments will help us improve Parallax!

Thought I’d add some comparisons to illustrate my previous comments on Parallax and BlurX at different scales. All three comparisons show non-deconvolved, Parallax, and BlurX from left to right. All are also the Lum master, though for the galaxies they are actually a SuperLum, and for the nebula, it’s the Ha master that I’m using as the luminance. The only post processing step prior to deconvolution for all three is BXT set at correct only, and in all cases I matched the sharpening between BXT and Parallax (0.9 or 1). To show things more clear, I used SetiAstro’s Statistical Stretch on each. Not a stretch tool I use normally, but seemed like the most ‘equal’ stretch to use for comparisons.

Fwiw, I agree with Bill. Where more detail shows, like in SH2-254, it’s not obvious to me whether the difference is more effective deconvolution, or a contrast enhancement.

📷 SH2-254.pngThe additional detail provided by Parallax in the larger cloud is obvious, but I see little difference in the smaller cloud to the right.

📷 NGC 5981.pngNGC 5981 is part of the Draco Triplet and in this case, BlurX is the obvious winner. The effect of Parallax is barely noticeable compared to the non-deconvolved version.

📷 NGC 4712.pngNGC 4712 is a much smaller, more distant galaxy near NGC 4725. Like with 5981, BlurX shows sharper detail than Parallax, but not by as wide a margin.

Cheers,
Scott

A couple more comparisons, first being stars.

📷 NGC 254 stars.pngAt default settings, Parallax doesn’t reduce stars as much as BlurX, and if you push the reduction setting in Parallax, the result are squarish stars with halos that extend out to the stars original size.

📷 NGC 254 Parallax stars at 8.pngOn a positive note though, I find that with some targets (data?) the combination of BlurX and NoiseX can bring out ‘stuff’ in the background. Whether artefacts of some sort, or dim stars and gas/dust that are very close to the noise level (as Adam Block has discussed), it’s pretty unattractive, but Parallax (plus NoiseX), being less effective at smaller scales, doesn’t bring it out nearly as much.

📷 stuff.pngCheers,
Scott

Tony Gondola · Jun 18, 2026 at 06:44 PM

Here’s another comparison, close up on the Tadpoles. Syqon is on the left. Stars were sharpened separately using BXT. Only the nebulosity was sharpened by Syqon.

📷 comp1.jpg📷 comp-2.jpgAgain, when it comes to really small scale detail, BX seems to have the edge.

What’s happened to this star? 📷 IMG_3462.jpeg

I’ve been trying both and have found similarly that Parallax does better on large scale items and BlurX much better on smaller scale and star “fixing”. And NoiseX and Prism both have unique qualities as well. The Syqon tools tend to mess more with the contrast/brightness relationships to enhance perception of detail. The RC tools tend to not mess with brightnesses, but just try to bring out finer detail. So I’ve been doing star corrections only in BlurX and then depending on the object, run through both Syqon and RC tools separately then pixel Math the results back together in the proportion I like. The pain with that is having to restart PI after using RC tools otherwise Syqon tools run slow.

I did that on this one to bring out different characteristics of the dust https://app.astrobin.com/u/Ricksastro?i=03jtp7

Tony Gondola · Jun 19, 2026, 08:50 PM

From what I have learned about both processes I think that’s an excellent approach. Take the best aspects of each tool to create something beyond what any single tool can do on its own.

Agreed, but what can fall through the crack are background galaxies if your primary target is large scale and you use Parallax……

Here’s one of my favorites, 2MASX J01203953+0325416, 1 billion light years away. Like before, Original, Parallax, BlurX left to right.

📷 2MASX J01203953+0325416.pngFwiw, here it is processed.

📷 2MASX J01203953+0325416_processed.pngCheers,
Scott

Rick Krejci · Jun 19, 2026, 07:52 PM

So I’ve been doing star corrections only in BlurX and then depending on the object, run through both Syqon and RC tools separately then pixel Math the results back together in the proportion I like.

I just tried this with a slight tweak. BXT for star correction, Parallax with star reduction, separated the stars, then cloned the starless version. I then ran NXT on one, Prism on the other. MAS & curves, independently tweaked. Stretched the stars with StarStretch, then used ImageBlend for both. Parallax still looked kind of blotchy, but the contrast and color saturation was greater. Then used PixelMath (RC*.6)+(SyQon*.4), and the results are nice.

SyQon Astro · Jun 21, 2026, 12:22 PM

At SyQon, we have always followed a very clear philosophy: preserving the information contained in the original data as faithfully as possible.

When discussing very small galaxies or barely resolved structures, it is important to remember that every imaging system is constrained by optical resolution, seeing conditions, sampling, signal-to-noise ratio, and, above all, by the information actually recorded by the sensor.

From an information theory perspective, details that were not captured in the orilaxiesginal dataset cannot be recovered in a strictly deterministic manner. Any representation of structures beyond the available information content inevitably requires some degree of statistical inference or modeling.

These approaches can be highly effective and may produce visually compelling results. However, the goal of Parallax has always been to extract and enhance as much of the real information present in the dataset as possible, while minimizing the introduction of details that are not directly supported by the observed data.

For this reason, especially when dealing with extremely small and marginally resolved objects, we favor a conservative approach that prioritizes fidelity to the acquired data over the reconstruction of details that may be plausible, but were not directly observed.

That really strikes to the core of the discussion.

From that I’ll pose a question: Lucky imaging of let’s say, the Moon, results in a final rendering that’s very close to the physical resolution of the aperture and wavelength and it does so without any tendency to “hallucinate” detail that isn’t there. As you get closer to the resolution limit, contrast declines to the point where detail can longer be rendered, the MTF curve goes to zero and nothing is seen.

I would love to see a tool for deep sky that acts very much the same way, sharpening the large scale structures and the small scale structure as you’d expect from an improved MTF rather than any assumed limitation or preference. I don’t think you could ever make a tool the gets to the diffraction limit with deepsky images without using cameras that can get meaningful data with exposures in the millisecond range but nonetheless, I would expect diminishing but visible improvement at all reasonable scales. The small Galaxies that have been put out as an example seem to be totally untouched by parallax. This is even more noticeable with small sensors where smaller detail dominates.