I understand the basic idea of drizzle. And I know that one of the "requirements" is that you be undersampled. For the large majority using small refractors and average sized pixels, drizzle is just a thing that is always done regardless of other conditions because their setup always results in an undersampled image. Unfortunately for me, whether or not I'm undersampled or oversampled 100% depends on the seeing conditions at the time. Because I have a small obsession with printing my images as large as I can get them, I always want the highest resolution possible. So, naturally, drizzle is something I'm interested in. That being said, I'd really like to avoid having to measure the seeing on any given night to decide if I'm undersampled; especially considering that those conditions can change from frame to frame, not night to night. It would be much easier to go the "always drizzle" route. Would there be any adverse effects from drizzling an image that wasn't undersampled?
Should I drizzle? What are the disadvantages of doing so needlessly?
Well Written Engaging
Hi Jaymz! I am also interested in this topic. I also use the drizzle process for a similar reason.
I take photos with OSC camera. According to the developers of Pixinsight, the use of drizzle is strongly recommended for shots taken with OSC camera. I can see you also use OSC camera. I recommend you the following video from the developers of Pixinsight:
https://www.youtube.com/watch?v=JLu3Civ2f3Y&t=255s
I take photos with OSC camera. According to the developers of Pixinsight, the use of drizzle is strongly recommended for shots taken with OSC camera. I can see you also use OSC camera. I recommend you the following video from the developers of Pixinsight:
https://www.youtube.com/watch?v=JLu3Civ2f3Y&t=255s
Helpful
Gábor Galambos:
Hi Jaymz! I am also interested in this topic. I also use the drizzle process for a similar reason.
I take photos with OSC camera. According to the developers of Pixinsight, the use of drizzle is strongly recommended for shots taken with OSC camera. I can see you also use OSC camera. I recommend you the following video from the developers of Pixinsight:
https://www.youtube.com/watch?v=JLu3Civ2f3Y&t=255s
That's new information to me. I'll have to look into that. Thank you!
Well Written Respectful
Particularities around OSC cameras aside, drizzling is the 'old wives tale' of astrophotography that refuses to die.
Drizzling was designed for Hubble and has almost no practical use for increasing details for us amateurs. Hubble is both in space - so no atmosphere - and is capable of very precise movements that our mounts cannot replicate. This is what drizzling was designed around, in addition to the optics and camera Hubble uses.
For the ambitious among you who think your mount can guide as good as Hubble, here is what you are up against: "The level of stability and precision that the FGSs provide gives Hubble the ability to remain pointed at a target with no more than 0.007 arcsecond of deviation over extended periods of time."
You are just better off to resample the image to 200% using bicubic spline or something.
See this thread for a practical example.
Drizzling was designed for Hubble and has almost no practical use for increasing details for us amateurs. Hubble is both in space - so no atmosphere - and is capable of very precise movements that our mounts cannot replicate. This is what drizzling was designed around, in addition to the optics and camera Hubble uses.
For the ambitious among you who think your mount can guide as good as Hubble, here is what you are up against: "The level of stability and precision that the FGSs provide gives Hubble the ability to remain pointed at a target with no more than 0.007 arcsecond of deviation over extended periods of time."
You are just better off to resample the image to 200% using bicubic spline or something.
See this thread for a practical example.
Well Written Helpful Insightful Concise Engaging
On top of what @SemiPro has said I'd like to add that drizzle always comes at a cost, that of SNR. Buyers beware!
Does anyone have any examples of processed images with/without drizzle? My assumption was that it served well to use Drizzle when you are heavily undersampled. (And BXT seems to also take advantage of this) Is that not the case?
Engaging
Is it best to appropriately pair your optics and sensor so you're close to 1"/pix? Imaging at .92 and .61 with OSC cameras I haven't found that drizzle integration does much. People recommended a tool at astronomytools.net to assess these things before buying gear. .61" appears to be on the edge of oversampling.
Nick Grundy, I recently posted a drizzled (Version A) then undrizzled (Version B) of the same target NGC 1788 at https://www.astrobin.com/j3bdj4/B/. I will often develop both, and bottom line for me is that the drizzled version will sometimes have better star morphology as long as you have enough data, but I really only appreciate that morphological difference when I "pixel-peep" really zoomed in. More often than not, I end up preferring my undrizzled versions. Clear skies!
Rob
Rob
Well Written Helpful Insightful Respectful Engaging Supportive
there are at least aspects of drizzling to consider
a) enhancing resolution: this goes back to Hubble, but in practice, arguably the main benefit is getting less blocky stars - as discussed and detailed by many others here and in other similar threads. but there are simpler ways to get less blocky stars for undersampled data, e.g. by upscaling/interpolation. Artifacts of drizzling can be streaks in the image (if the dithering is not good enough), or holes when drop radius is taken to small. in any case, the apparent resolution increase goes along with poorer S/N.
b) there is however another benefit when drizzling, in particular with OSC data. In the case of OSC data, one usually interpolates at least twice, once for the debayering and once for the registering. 1x drizzling can also be seen as an interpolation scheme (more correct an assignment scheme), and drizzling replaces two interpolations with one assignment step thus avoiding some image artifacts in particular for pretty noisy data, such as Moire pattern formation.
so 2x drizzling to increase resolution probably does not really work and has the risk of image artifacts.
1x drizzling in particular for OSC data may however indeed improve the quality of a stacked image.
Matthias
a) enhancing resolution: this goes back to Hubble, but in practice, arguably the main benefit is getting less blocky stars - as discussed and detailed by many others here and in other similar threads. but there are simpler ways to get less blocky stars for undersampled data, e.g. by upscaling/interpolation. Artifacts of drizzling can be streaks in the image (if the dithering is not good enough), or holes when drop radius is taken to small. in any case, the apparent resolution increase goes along with poorer S/N.
b) there is however another benefit when drizzling, in particular with OSC data. In the case of OSC data, one usually interpolates at least twice, once for the debayering and once for the registering. 1x drizzling can also be seen as an interpolation scheme (more correct an assignment scheme), and drizzling replaces two interpolations with one assignment step thus avoiding some image artifacts in particular for pretty noisy data, such as Moire pattern formation.
so 2x drizzling to increase resolution probably does not really work and has the risk of image artifacts.
1x drizzling in particular for OSC data may however indeed improve the quality of a stacked image.
Matthias
Helpful Insightful
Nick Grundy:
Does anyone have any examples of processed images with/without drizzle? My assumption was that it served well to use Drizzle when you are heavily undersampled. (And BXT seems to also take advantage of this) Is that not the case?
I was reading this thread while actually running a test on my OSC (ASI6200MC) with a RASA 11. Drizzle certainly introduced some artifacts, but it did reduce pixelation in the stars. I ran a test with 1x, 2x, and 3x drizzle. This is 6x 10min subs integrated, but no bias, darks, or flats applied.
Well Written Helpful Insightful Concise
I fail to see how noise in those pictures is any different between each other. Besides it's the SNR that really counts and it really counts in the shadows.