Drizzle

[John_Gill]

Hi,

To drizzle or not to drizzle, that is the question.  Yeah, I know it depends ... When should drizzle be applied or avoided.

Look up
John

[Niall Saunders]

I certainly now make the effort to 'drizzle' duriing the image acquisition phase - it doesn't 'cost much' in terms of effort or time, and gives more scope for pre-processing options.

For example, I had been having issues with fixed-pattern readout 'moise' (doe to incorrect Gain and Offset settings on the CCD), but drizzling allowed me to align, stack and eliminate this noise from the Master Light frame.

If you collect lots of Lights, and have drizzle selected during the acquisition phase, then StarAlignment can do that for you. What I haven't (yet) tried is selecting drizzle during the StarAlignment and ImageIntegration proceses. I don't even know what the intended outcome is when using drizzle within PixInsight - the last time I tried any form of 'drizzle' was within Meade's "Envisage" software, working with their DSI cameras, and there then seemed to be two primary choices: use drizzle to get larger images, or use drizzle to get a smaller pixel ratio.

I, too, would be interested to hear from users who have played with the drizzle options.

[Nocturnal]

Don't confuse drizzling with dithering. Drizzling is a stacking operation, dithering happens during acquisition.

Drizzling only makes sense when you are undersampled and want to create a larger image than what you would get without drizzling. If you already intend to create say a 1/2 or 1/3 downsized result image of a 3K x 2K master then there really is no point to drizzling. You could try drizzling if you image a small target with a fast telescope and intend to crop the result. You could get slightly more detail in the final image that way.

Naturally much has been written about this topic and a google search will show you the way.

[Niall Saunders]

Don't confuse drizzling with dithering. Drizzling is a stacking operation, dithering happens during acquisition.

You are, as always, correct Sander - I haven't actually used the Drizzle capabilities of PixInsight, and I had 'dithering' in my mind from past experience with the DSI cameras.

I might actually do as you suggested, and try to actually learn something about the subject!!

[Nocturnal]

Ha, I'm hardly ever right Niall but I'll tell my wife you said that 

In fact the DSI camera *did* one of the first drizzle implementations. I still have a DSI and a DSI Pro though I didn't use the included software a lot. It hawked that they used NASA technology (drizzling) to make better pictures. This was the drizzling as we talked about here and it was done during acquisition without dithering the mount.

[RickS]

I use Drizzle regularly.  As already mentioned, you need undersampled and dithered data.  The advantages of using Drizzle are that:

• you may recover some resolution (if your optics, guiding and seeing are significantly better than needed for your image scale)
• star shapes may be improved (especially at large image scales where stars don't cover many pixels)
• the drizzled data usually looks cleaner (some of the noise gets smeared between pixels)

The disadvantages are that MureDenoise doesn't work so well with Drizzle integrations and it takes longer to process the larger images.

So long as you're undersampled it is worth a try...

Cheers,
Rick.

[pfile]

I use Drizzle regularly.  As already mentioned, you need undersampled and dithered data.  The advantages of using Drizzle are that:

• star shapes may be improved (especially at large image scales where stars don't cover many pixels)

this is a really good point - images with large scale are where the lancosz-3 algorithm seems to create ringing on the small stars, but this is avoided by using drizzle.

rob

[John_Gill]

Hi,

So I took my 40 images @ ISO 400 for 180 seconds of the Eagle Nebula and "rack, pack and stacked" them using the drizzle.  After doing DBE and my usual workflow the result looked bad.  The image was very grainy and the background had faint diagonal streaks.  Now using the same images and the same "rack, pack and stack" setting but without drizzle, and same post-processing workflow, the image is far, far better.

Still confused, but that is normal 

[RickS]

How does the drizzled version compare if you downsample it to match the original?  40 x 180 seconds isn't a huge amount of data.  How did the drizzle_weights image look?  Was it a nice, featureless white?

Cheers,
Rick.

[John_Gill]

Hi,

I did an "IntergerResample" - "Downsample" to the original size and the image is still "grainy"  I will re-stack the images and see what the results are.

Look up
John