I wondered - since I had about 75 luminance subs of NGC253 if I could possibly extract more detail from them by performing a drizzle integration (x2) using PixInsight. So following the procedure on the PI forum by Juan Conejero I aligned, drizzled and then stacked them quickly.
The result is a severe crop of the resulting image, but I think you can see that the dust streams and finer detail is much improved in this final image. I have included the original colour version for comparison.
Comments welcomed
Allan
Last edited by allan gould; 31-08-2015 at 05:00 PM.
Drizzle works very well for large amounts of undersampled data. I've been using it for about a year with data from SRO (2 arcsec/pixel image scale and seeing round 1 arcsec on good nights.)
Question: why doesnt everyone capture bin 2 or 3 and then drizzle?.
Good question, Fred.
My take on this is that Drizzle isn't perfect. If you Drizzle up to twice the resolution with good, undersampled data then you don't get quite as good a result as you would have from pixels that were half the size. A small amount of spatial information is lost.
Although Drizzle seems like magic, it isn't. It's just trading off resolution against SNR. That may not be obvious from looking at a Drizzled image because Drizzle introduces correlated noise that isn't as visually intrusive.
so where is the PI colour version Professor Pixel?????
very much improved
Patience young one. Wax on, wax off
Quote:
Originally Posted by RickS
Looks great, Allan!
Drizzle works very well for large amounts of undersampled data. I've been using it for about a year with data from SRO (2 arcsec/pixel image scale and seeing round 1 arcsec on good nights.)
Cheers,
Rick.
Thought that I would give it a trial as I don't use PI in my workflow at all, but I'm slowly realising that I should.
Allan
Quote:
Originally Posted by Bassnut
Stunning (apparent) res and sharpness.
Question: why doesnt everyone capture bin 2 or 3 and then drizzle?.
Don't know to be honest
Quote:
Originally Posted by RobF
Nicely done. What scope were you shooting for this again please Allan?
Thanks Rob. It was a 5" refractor with fl of 950mm
Quote:
Originally Posted by RickS
Good question, Fred.
My take on this is that Drizzle isn't perfect. If you Drizzle up to twice the resolution with good, undersampled data then you don't get quite as good a result as you would have from pixels that were half the size. A small amount of spatial information is lost.
Although Drizzle seems like magic, it isn't. It's just trading off resolution against SNR. That may not be obvious from looking at a Drizzled image because Drizzle introduces correlated noise that isn't as visually intrusive.
Good one Allan. I've had very mixed results with PI's Bayer Drizzle and Drizzle integration. Not behaving the same on all data. I hope they fix it in the next version and take it out of this bloody pre-processing script which is a bit of a dog's breakfast IMHO. Trying to do too many things on one.
Good one Allan. I've had very mixed results with PI's Bayer Drizzle and Drizzle integration. Not behaving the same on all data. I hope they fix it in the next version and take it out of this bloody pre-processing script which is a bit of a dog's breakfast IMHO. Trying to do too many things on one.
Marc, In my case it appears to be beginners luck mainly since I had absolutely no idea what I was doing, just following the script.
I did try DSS but found it lacking, especially in regard to the stars (see attached image). Ringing and halo-ing was evident and thats why I tried PI.
But thanks for the comments.
Allan
Very nice result. I've had good experiences with PI's Drizzle implementation. It does introduce some noise as Rick mentioned, but the fine detail is usually in the high S/N areas anyway so the extra noise is hardly noticeable if you blend a standard 1x1 image with a drizzled version using something like a range mask.
It looks to me that drizzled image has more contrast and the stars are nicer too.
However, I have noticed that one of the stars in the bottom left 'corner' of the galaxy's disk seems to have processing artefacts (dark halos) - just wondering if some of the structures within the galaxy could also be a product of drizzle integration.
Very nice result. I've had good experiences with PI's Drizzle implementation. It does introduce some noise as Rick mentioned, but the fine detail is usually in the high S/N areas anyway so the extra noise is hardly noticeable if you blend a standard 1x1 image with a drizzled version using something like a range mask.
Rolf, thanks for the comments and for the tip regarding noise. I guess there is no free lunch except to get more data to remove the noise when it crops up. I will have to seriously play with this procedure and see what dithering and drizzling parameters are best for my system.
Quote:
Originally Posted by Slawomir
Hi Allan,
It looks to me that drizzled image has more contrast and the stars are nicer too.
However, I have noticed that one of the stars in the bottom left 'corner' of the galaxy's disk seems to have processing artefacts (dark halos) - just wondering if some of the structures within the galaxy could also be a product of drizzle integration.
It is a nice image nonetheless
Thanks Slawomir
The contrast and details are elevated by the procedure but I have compared my image with that of R Jay GaBany (http://apod.nasa.gov/apod/ap060421.html) as this was the one that showed me the elevating dust that I've been trying to capture. Essentially every detail in his image is present in mine except for a lot less dust and it may be one of my subs has something to give the halo?
But thanks for the comment as it's much appreciated. I will have to run the procedure again with slightly different parameters and see what happens.
Allan