More LR deconvolution examples

[iceman (Mike)]

Hi guys and gals.

What else to do when it rains all day and is cloudy all night. Reprocess avi's!

Reprocessed some good data from the 4th May, and wanted to experiment with different forms of LR deconvolution.

Each has had identical processing in every way, except for the LR method applied.

Top left: Exponential, 10 iterations @ 1.2
Top right: Guassian, 10 iterations @ 1.2
Bottom left: Guassian, 7 iterations @ 1.4 (this was how I originally processed it)
Bottom right: Extract from Image, 5 iterations @ 2 radius

Interesting results, I guess. The top left image has the most contrast and sharpness. It could be argued that it's oversharpened. I think my favourites are the original, or the top right, which reveal similar detail with similar contrast and sharpness.

Interested in your thoughts.

Don't know what conclusions to draw from it yet, but it's all part of the experimentation to find out what works best on good data, bad data, etc.

[sheeny (Al)]

Beautiful images Mike!

Gee, I dunno. The differences just seem so subtle to me.

Al.

[gbeal]

All look great. If anything, top left, and bottom left.
Get that tutorial done so we can all catch up with you.

[jjjnettie (Jeanette)]

The two on the right are too soft. Detail is lost.
The top left image has stacks more detail, but it has a harshness that comes with a little too much processing.
The piccie on the bottom left hand side is the best for me.
It has all the detail that you bought out in the image above it with enough softness to make it the most natural looking photo out of all of them.

[Robert_T]

My thoughts... geez these are great shots

The choice between "exponential" and "gaussian" I must have missed in applying LR deconvolution. Are these choices in the screen where you set curve radius etc? How do you manage to run 10 iterations without getting ringing around the edge?

I like top right best.

cheers

[janoskiss (Steve H)]

Gaussian should be a much better approximation to the Bessel function, which is what you get with an ideal circular aperture (see attached plot). Is there no Bessel option in your deconv software? Exponential is very unphysical for any optical instrument, though it may be useful for removing pixel "bleeding" from CCD recorded images.

To my eyes the RHS images look much better. The bottom right looks best. I don't think you have lost detail there. You're just not adding detail that was not there in the first place. It seems to be the most artifact free (e.g. ringing at the limb).

[davidpretorius]

I have been using exponential ME between 2 and 5 x 1.1 ME deconvolution after one of our friends on IIS in the USA suggested using exponential

It seems to work better for me at my image scale. I have never liked extract from image.

[davidpretorius]

i am yet to try steve's deconvolution link from the last thread where this was all discussed

[Lester]

Great images Mike,

Loads of detail. I like the first 3 the bottom right one appears not as sharp as the others.