Benefit of Drizzle?

[dpaul]

Hi

Can someone give me advice on whether adding drizzle files during integration will ''in my case'' help?  I have a 30'' F3.5'' scope which when used with a Paracorr means the focal length is just over 3 metres.
I'm using a CMOS camera with 3.8 pixel size and using 2x2 binning. I know this is seriously over-sampled but I'm generally happy with the results, aided by the superb optics from Mike Lockwood.

So here's my question - I think I've seen Juan's original announcement of adding Drizzle in 2014 and there may have been comments where it helps with ''under-sampled'' images. So in my case would I expect any benefit?  I'm still confused about what drizzle is really doing but doesn't it make the image less ''grainy''?  I tried redoing an image for the very first time by adding drizzle files (there were plenty of frames, 25 in this case and I used linear rejection during image integration). It did seem to make a small difference, a little less noisy maybe?

Any comments from experienced users are very welcome. Sometime I use 3x3 or 4x4 binning with narrowband so would that be more beneficial then?

Thanks

David

[pfile]

the idea behind drizzle is to recover lost spatial resolution when your pixels are too big for the resolution the instrument is capable of - the large pixels under-sample the image formed at the sensor.

figure 3 in the original paper (https://arxiv.org/pdf/astro-ph/9808087.pdf) kind of says it all.

in theory the SNR in a drizzled image should be lower than a non-drizzled image per section 7 of that paper, i guess because the noise in any one input pixel spreads to multiple output pixels. in practice though people seem to measure higher SNRs in drizzled images (based on what i've read in various fora.) i'm not sure if this is a measurement error, or if it has something to do with the increase in spatial resolution somehow 'fooling' the noise computation, or if it is somehow real.

so it sounds to me since you are so oversampled that drizzle will not help you much at bin 1x1 or 2x2 but perhaps at 3x3 or 4x4 it makes sense? in the end you'd have to work out what your image scale is for the various binnings and see how they compare to your seeing.

rob
 

[RickS]

Image scale at 2x2 is 0.59 arcsec/pixel, so you'd need seeing better than 2 arcsec to see any benefit in resolution.  At 3x3 or 4x4 it's more likely to be helpful.

[dpaul]

Rob

Thanks for your reply, that's what I expected.

David

[dpaul]

Hi Rick

Thanks for the reply - I do appreciate with 2x2 binning I am oversampling but I still seem to get more detail from deep sky objects with 2x2 compared to 3x3 or 4x4.
Attached are 2 versions of NGC 891 that demonstrate this.

Regards

[dpaul]

Sorry Rick

Here are the attachments

David

[dpaul]

Rick

Here is the 2x2 binning (I can't seem to add two images at once on this forum - they were not too big either?)

David

[dpaul]

Rick

Here is one more 2x2 binned image.

David

[RickS]

I do appreciate with 2x2 binning I am oversampling but I still seem to get more detail from deep sky objects with 2x2 compared to 3x3 or 4x4.
Attached are 2 versions of NGC 891 that demonstrate this

Hi David,

Whether you're oversampled at 2x2 depends on the local seeing conditions.  As I mentioned, your image scale at 2x2 is 0.59 arcsec/pixel.  That's well sampled for seeing of around 2 arcsec.  At 3x3 the image scale is 0.885 arcsec/pixel which is well sampled for seeing around 3 arcsec.  If your seeing is better than 3 arcsec then you will get better resolution with 2x2 binning.  For seeing better than 3 arcsec you also have the option of binning 3x3 or 4x4 and doing Drizzle though it is not going to get you better resolution than imaging at 2x2.

Cheers,
Rick.

[dpaul]

Thanks Rick

David