Hey All,
I recently came across the concept of over/under sampling. Checked out my rig on astronomy.tools and see I am under sampling quite a bit.
In this case, I have seen that drizzling could help this, to some extent.
Tested it out and it seemed to be somewhat better at 2x drizzle.
When drizzling 2x, do you typically use the stars from that also? The reason I ask is a found a video deep diving into it mentioned maybe the stars are better in a 1x drizzle. Then just remove stars and use nebula from 2x?
Thanks!
📷 image.png
Drizzle when undersampling?
Respectful Engaging
Well, when drizzeling I tend to keep my equipment under covers.
Can’t see the stars in the rain anyway.
Trey Wood:
Hey All,
I recently came across the concept of over/under sampling. Checked out my rig on astronomy.tools and see I am under sampling quite a bit.
In this case, I have seen that drizzling could help this, to some extent.
Tested it out and it seemed to be somewhat better at 2x drizzle.
When drizzling 2x, do you typically use the stars from that also? The reason I ask is a found a video deep diving into it mentioned maybe the stars are better in a 1x drizzle. Then just remove stars and use nebula from 2x?
Thanks!
📷 image.png
Why would you?
That doesn’t make any sense to me but my best advice is to just try it both ways and compare.
To your basic question, yes, when you’re under-sampled, you can apply drizzle.
I have exactly the same issue you have. I have a WO RedCat 51 which I use with an old ZWO ASI 1600 (3.8 micron pixels) . The issue nowadays is that most CMOS sensors come with tiny pixels which are not compatible with short focal length scopes like the RedCat 51.
The undersampled pixelated stars are almost impossible to fix in post processing. The only solution that has worked so far for me (and it is very dependent on how good your seeing conditions are) is drizzling x2.
Here is a comparison of the same RGB master stacked using 1x Drizzle and 2x Drizzle. As you will see, the difference when applying Drizzlex2 is quite noticeable.
Well Written Helpful Respectful Concise Engaging
Drizzling helps when undersampled. I have found it to be a significant benefit, but I am a pixel peeper and those of my images are intended to be viewed by pixel peepers. There are some trade-offs, but I do it every time under the conditions and the objects I am working with.
Why not drizzle the stars? In my rigs, where I use drizzle, the smallest faintest stars are sometime 1, 2 or 3 pixels in extent. If you like your stars looking like little boxes and “L”s then don’t use the drizzled stars. When I drizzle, those stars yield nice simple psf forms. They actually look like stars (photographic stars). And no, simply up-sampling a non-drizzled star field will not be the same.
However, even with very undersampled data, but with a very widefield intent, such as a widefield lens for landscape astrophotography, I really see no point in drizzling. At that point, the drawbacks weigh in favor of not drizzling. And if you want to print large, the the shape of the stars do not matter. It is the overall impression of the full field that is what is important.
Now for images that are not undersampled, I have also tried drizzling in that case. I will say that the drizzling absolutely did not improve detail resolution. However, I still felt that the star forms were much better and in the cases where I needed that, then I used it. Otherwise, no.
Well Written Helpful Engaging
I tend to always do a 2X drizzle when I integrate, so that I have masters of both undrizzled and drizzled. The only drawback for me is increased processing time during integration., and of course larger file sizes. Then, I examine everything closely to see if it helped (both stars and non stellar) and proceed from there.
I have used undrizzled starless with drizzled stars. The drizzling did not give me any additional detail in the nebulosity, but did give nicer star shapes. I then downsampled the drizzled stars to match the undrizzled starless image. But it really depends on each dataset and what I am trying to do with it.
Well Written Helpful Insightful Respectful Concise Engaging Supportive
Wanda Conde · Mar 3, 2026, 07:17 PM
I have exactly the same issue you have. I have a WO RedCat 51 which I use with an old ZWO ASI 1600 (3.8 micron pixels) . The issue nowadays is that most CMOS sensors come with tiny pixels which are not compatible with short focal length scopes like the RedCat 51.
The undersampled pixelated stars are almost impossible to fix in post processing. The only solution that has worked so far for me (and it is very dependent on how good your seeing conditions are) is drizzling x2.
Here is a comparison of the same RGB master stacked using 1x Drizzle and 2x Drizzle. As you will see, the difference when applying Drizzlex2 is quite noticeable.
Aren’t smaller pixels an advantage for short focal lengths to help mitigate undersampling? If you had larger pixels with the RC51, you would be even more undersampled.
Looking at your screenshot, it appears that some of the medium brightness stars have holes in the middle in the drizzled images. Not sure what causes that.
Well Written Engaging
Rick Krejci · Mar 21, 2026, 03:01 PM
Wanda Conde · Mar 3, 2026, 07:17 PM
I have exactly the same issue you have. I have a WO RedCat 51 which I use with an old ZWO ASI 1600 (3.8 micron pixels) . The issue nowadays is that most CMOS sensors come with tiny pixels which are not compatible with short focal length scopes like the RedCat 51.
The undersampled pixelated stars are almost impossible to fix in post processing. The only solution that has worked so far for me (and it is very dependent on how good your seeing conditions are) is drizzling x2.
Here is a comparison of the same RGB master stacked using 1x Drizzle and 2x Drizzle. As you will see, the difference when applying Drizzlex2 is quite noticeable.
Aren’t smaller pixels an advantage for short focal lengths to help mitigate undersampling? If you had larger pixels with the RC51, you would be even more undersampled.
Looking at your screenshot, it appears that some of the medium brightness stars have holes in the middle in the drizzled images. Not sure what causes that.
Rick is correct, smaller pixels increase the sampling. Shorter focal lengths require smaller pixels.
The “holes” in the middle of some of the stars in this image are interesting. If you look carefully, you can see some of the stars clearly are less bright in the center even in the undrizzled image. The drizzling increased the effect. Curious.
Well Written Respectful Concise Engaging
That’s usually caused by certain sharpening processes applied at some point in the work flow. The light distribution making up the stars becomes no-gaussian.
Concise
Under or over sample?
Hey all. So I have this 65phq paired with a 294mm pro. Bin 1 is oversampled, bin2 is undersampled. I’m consistently getting guiding around 1/4” or lower so that’s not a factor.
Relevant - addresses undersampling issue mentioned in source post.
Dose a barlow make sense?
I have a possible rig purchase in mind, one being advertised on CN. It’s for the WO UltraCat / ZWO ASI2600MC Pro combo at Star Front Observatory. If I decide to buy the rig (it’s basically a ready-to-go system) and also pay the SF pier take over, it’...
Barlow lens affects sampling ratio; relevant to undersampling discussion.
How do my stars look?
I have been working on my workflow for OSC Narrowband using the Askar SQA 55 and the 2600MC Pro. I drizzle 2X and then Resample down 2X to the native resolution after DBE and BlurXT. I use Bill Blanshan’s SHO Colorization script. In this case I used ...
Addresses drizzling and resampling strategy relevant to author's 2X drizzle workflow.
Mysterious star trailing solved: a cautionary tale of fan vibration with a fan cooled (especially ZWO) camera.
Suddenly after no problems with star shapes, the next night out they were just terrible! Below I show a gif comparing two 2 hr stacks, one with the good stars, and the next with elongated stars. In this example the distortion was vertical. But later ...
Star quality issues; relevant to drizzling and star handling concerns.