To deconvulve or not?

[Tim]

For me the deconvulve routines have always been a bit of a hassle. Using CCDSharp meant altering the format of the image, and more often than not the rings around the stars were worse than the stars in the first place, and usually ended up with frustration.

However, like washing the car, its one of those jobs you know you really should do. So I want to get the hang of it really, and have been itching to get to grips with it since RobH demonstrated its importance at Salisbury SP.

Taking the three greyscale images from the tadpole pic, I resolved to reach a workable solution. One problem with them was that my synscan controller got fried somehow, at least the ESC button is, and so I had to hurriedly switch to using eqmod pulse guiding, which I havent had on since September, and I forgot to alter the RA agression, so the stars were a little eggy in the 20 minute subs.

Having bought Pixinsight i'm determined to get some decent use from it, even thought the deconvulve section looks a bit daunting. I ran the decon preview script, and saw a compromise, but then decided to play with the settings myself. Using the scale and rotation sliders I found that I could use the deconvulve tool to bring the stars back a little to a rounder shape, very handy! Also by playing with the deringing sliders you can eliminate the dark haloes pretty much. All the alterations can be applied to the linear image data, but with a preview of how it will affect the stretched image.

The deconvulve tool also really helps to sharpen up the edges of structures, as with these tadpoles, and I have to say it was worth the extra effort. I am looking forward to obtaining the rest of the data now and getting to grips with these tadpoles. I love the way they appear to be swimming towards the dark nebulae!

So, who does, and who doesn't deconvulve?

Cheers

Tim

[harry page]

You are getting all techy on me now !

did you mask the lower signal areas ?

Harry

[Tim]

LOL Harry, you should be able to tell! No, as you can see if you look closely. But later on when the NB data was merged (in Registar) I brought it back to PI and re did the decon using a mask, and only a couple of iterations. There's a lot to learn, but finally feel like making a little headway with it now. I have a natural aversion to anything with numbers in, so the tools all look a bit daunting, but encouraged by your vids to have a play, thats what I did.

I'll bung up a little crop of the stars as they were in their eggy state for comparison. I really didnt want to throw away 9 hours of data!

[steppenwolf]

I have never done convolution on my images but this is mainly because I use an OSC. I have played with this on some mono images using both MaximDL and CCDSharp but have never really found a satisfactory improvement, preferring to use some actions of my own design in PhotoShop. I'll be interested to see how you get with this though, Tim.

[Tim]

Here we are, original stars from Ha Layer. You can see that PI has pulled them back quite a bit.

I have been using the masks more and more, but is it possible to hide it while it is applied, so that the result of the process can be seen clearly?, I usually CTRL+Z & CTRL+Y to review the changes.

Cheers

Tim

[harry page]

Hi

If you look at the top menu bar , there is a mask section and there is a icon with a star in the corner , this will hide or show the mask

Harry

[Tim]

Aha, thanks. That will make it easier.