Iris Core-25.4 Hours

[Rodd]

29 minutes ago, PhotoGav said:

Olly, I'm fascinated by your input to this thread and the techniques that you mention. What a shame that you don't have a YouTube channel with tutorial videos...! I'm inclined to agree with your suggestion that it is the outer bits that aren't 'in sync' with the core and I would love to see what you would do to recitfy the 'issue'.

Just to reiterate, Rodd, this is a lovely image that has stimulated some really good discussion of processing techniques in the pursuit of perfection.

Thanks Gav--and I agree about Olly having tutorials.  As a nameless one, he reigns supreme.

Rodd

[Rodd]

33 minutes ago, apophisOAS said:

Oh Yes, super,

Roger

Thanks Roger

Rodd

[fwm891]

Rodd, Looking at the Iris and walking away and back again (my) thoughts tend to feel that the core is superb the background is great it's the dust to me that feels awkward. The core stands on it's won no problem, but the dust merges with the background and gets lost. Lift the dusty areas a little, and I mean 'a little' so the whole Iris is lifted away from the background...

Francis

[Rodd]

8 minutes ago, fwm891 said:

Rodd, Looking at the Iris and walking away and back again (my) thoughts tend to feel that the core is superb the background is great it's the dust to me that feels awkward. The core stands on it's won no problem, but the dust merges with the background and gets lost. Lift the dusty areas a little, and I mean 'a little' so the whole Iris is lifted away from the background...

Francis

Therein lies the problem--I can't lift it anymore due to the data quality.  Its noisy.  I have tried lifting and applying various amounts of noise control, and using arcane pixel math formulae  to no avail.  I need more data.  Also, I a try to process according to Olly's adage of leaving (5% or 10%--can't remember) of the image on the table.   At present, when one views this image, it takes some consideration to come to the conclusion that the dust is "awkward"--maybe a second (or third) look.  If I lift it and it shows its defaults, that determination would come instantly.  Believe me, this is a new approach for me.  Usually I over stretch and reveal an images shortcomings.

Rodd

PS--Just the thought that the background is decent has me doing cartwheels.   Background is usually one of my pitfalls--especially in broad band.

Edited October 11, 2019 by Rodd

[Rodd]

9 hours ago, Xiga said:

Out of interest, what does the image look like before and after you apply DBE? (assuming that's what you've ran).

Forgot to comment on this.  yes I use DBE--the second step after channel combination (first step is cropping edge artifacts).  This target happens to be in the darkest park of my sky--not in the light domes of the south and east, and I made sure not to shoot when the Moon was visible--so gradients were not really a problem.  Therefore, DBE had little effect.  It helped a bit in a few spots, but I had to clone the image before, apply DBE to the original, and look closely at the one with DBE and the one without to really see a difference. 

Rodd

Edited October 11, 2019 by Rodd

[ollypenrice]

28 minutes ago, Rodd said:

This sounds very nice-much easier to control than using TGV Denoise.  Noise control is a funny thing.  When you get it right, its amazing--can't tell any noise control was applied.  Get it wrong, and you experience fatial expressions you didn't know you had.  I think there is a practical limit.  beyond a certain point, a certain amount of noise, noise control will not succeed, no matter how its applied.  Lately, for me noise control is like mercaptan--the odorous gas they put in propane so you can detect it.  It has a distinct, unpleasant smell that can be detected in parts per billion.  When I see noise control, its like getting a big whiff of that smell.

Rodd

This might be a time to remove and replace noise, in that case. I remember a guest asking me, regarding this Ps filter,  'Add noise? Why would you want to add noise?' At the time I agreed with him but what can happen is that noise on a larger scale can only be removed by the usual smoothing algorithms which leave the 'vaseline on the lens' look. This can be mitigated by the careful addition of a little Gaussian noise to put the texture, the grain in smoothed parts, back to the image's overall level after smoothing. I did this in a recent image. I'm not saying which or where because if it didn't offend you I got it right!  More data is the right way but then you're always tempted to stretch the new data a bit harder which means you're tempted by NR which means you want more data... 

In a west country accent:

I like zider.

When I drinks zider it makes I smell.

When I zmells people gives I money to go away.

With this money I buy zider...

 

Olly

[Rodd]

7 minutes ago, ollypenrice said:

This can be mitigated by the careful addition of a little Gaussian noise to put the texture, the grain in smoothed parts, back to the image's overall level after smoothing. I did this in a recent image.

I think this is a very interesting technique.  I did try adding noise in this fashion one time (PI has a tool to add noise and the list of types of noise it can add is ridiculous.  gaussian is one of them).  I didn't really get into it too much though--certainly not enough to really get a feel for it. 

Here is another problem that is not insignificant.  Me being able to achieve comparable results on the core is not  a given.  I have tried reprocessing this data since I posted this image--just seeing if I can improve things and getting a feel for some of the techniques suggested.  Achieving similar result did not happen before I got tired and had to stop for the night.  It wasn't a full on reprocessing attempt--which is what it will take, but still.  It is completely possible that I will collect another 10-12 hours of Lum and still not be able to improve the result.  That would not surprise me at all.  Eventually, I may succeed.  But it could be a "Long and winding road".

Rodd

[jjosefsen]

Nice image first of all, and I think the colors are good and what I would expect to see.

I do however think the the image has some "mottling" going on in the luminance layer, mostly visible in the low signal areas.

This could be a result of the NR technique used, I know that TGVDenoise when applied with the "Jon Rista Method" in PixInsight can introduce this. I target this noise with MMT and heavy masking afterwards (also as described by Jon).

It could also be sharpening or deconvolution.

When it comes to noise reduction and noise in general I find it really hard to determine the correct amount. Personally I prefer a little smoother or softer image, without getting too much of that oily look to it.

But then again some people seem to think it detracts from the image if it is too smooth.

 

[coatesg]

1 hour ago, Rodd said:

yeah I am pretty familiar with it--but still learning the really delicate adjustments.  There is a tool called statistics that you are supposed to use (according to the Light Vortex tutorial).  The tool gives you teh standard deviation in exponential format and you put that number in Edge Control (second slider).  But so far it has given me values to the -2 power, which is very aggressive (default is -3).  So so far I have not been able to make it work the way the tutorial suggests.

No - that stats thing doesn't work for me either. I just land up playing with the edge protect setting until it is acting on the right scale of noise, and then adjusting smoothness and finally strength until it's just enough to do the job. Use of the image itself as local reference is helpful to target only the low SNR areas. 

[coatesg]

Double post...

Edited October 11, 2019 by coatesg
Double post

[Rodd]

Just now, jjosefsen said:

Nice image first of all, and I think the colors are good and what I would expect to see.

I do however think the the image has some "mottling" going on in the luminance layer, mostly visible in the low signal areas.

This could be a result of the NR technique used, I know that TGVDenoise when applied with the "Jon Rista Method" in PixInsight can introduce this. I target this noise with MMT and heavy masking afterwards (also as described by Jon).

It could also be sharpening or deconvolution.

When it comes to noise reduction and noise in general I find it really hard to determine the correct amount. Personally I prefer a little smoother or softer image, without getting too much of that oily look to it.

But then again some people seem to think it detracts from the image if it is too smooth.

 

I struggle with background.  This just goes to show you that I did use NC and even probably too much--so recommendations for more are definitely problematic.  The level of "motling" in the crop example does not bother me too much.  I don't see mottling in the background--I see it in the dust in front of the background.  that will be fixed with more data.  however, I am not as convinced that it comes from teh luminance layer.  I looked at the RGB and L channels and they all appeared about the same as far as nosiness.

This image definitely is not meant to be zoomed in teh dust--so normal viewing is pretty decent.  that is why I am willing to call this an image.....Its about as good as I can make it.

Rodd

[Rodd]

1 minute ago, coatesg said:

No - that stats thing doesn't work for me either. I just land up playing with the edge protect setting until it is acting on the right scale of noise, and then adjusting smoothness and finally strength until it's just enough to do the job. Use of the image itself as local reference is helpful to target only the low SNR areas. 

I agree

[Rodd]

12 hours ago, fwm891 said:

Rodd, Looking at the Iris and walking away and back again (my) thoughts tend to feel that the core is superb the background is great it's the dust to me that feels awkward. The core stands on it's won no problem, but the dust merges with the background and gets lost. Lift the dusty areas a little, and I mean 'a little' so the whole Iris is lifted away from the background...

Francis

 

11 hours ago, ollypenrice said:

In a west country accent:

I like zider.

When I drinks zider it makes I smell.

When I zmells people gives I money to go away.

With this money I buy zider...

 

 

11 hours ago, jjosefsen said:

do however think the the image has some "mottling" going on in the luminance layer, mostly visible in the low signal areas.

 

11 hours ago, coatesg said:

No - that stats thing doesn't work for me either. I just land up playing with the edge protect setting until it is acting on the right scale of noise, and then adjusting smoothness and finally strength until it's just enough to do the job. Use of the image itself as local reference is helpful to target only the low SNR areas. 

I very much appreciate all the comments and I would add quoates from others but I don't know how to do multiple quoates from multiple pages.  Anyway--I reprocessed the image and not surprisingly, it took about 100th the amount of time the original did.  It probably could use some fiddleing and fine tuning, but I think the dust is less noisy, the background pretty good, the central core less over-powering, and the dust has been lifted a bit more.  To be truthful, I blended the new version with the previously posyed.   I think a new version could be made to improvement without the need to blend--but that will take a bit of tinkering.  Meanwhile--one step forward is better than 1 step bacK!  let me know if you think I have gone bonkers!

Rodd

And a touch of star control--not much (I don't like a lot)

Edited October 12, 2019 by Rodd

[fwm891]

Nailed it - Brilliant Rodd 👍

[ollypenrice]

Agree with Francis - nailed! Super.

I often end up blending two separate processings. Why not?

Olly

[PhotoGav]

Rodd, this is superb!! It now has a fabulous 3D quality to it. The core doesn’t jump out at me now, it sucks me in to the image and deep into space. Sweet, well done! 

[coatesg]

Yes! This is much better - the core sits much better in the dust now - great job!

[Rodd]

4 hours ago, fwm891 said:

Nailed it - Brilliant Rodd 👍

 

4 hours ago, ollypenrice said:

Agree with Francis - nailed! Super.

I often end up blending two separate processings. Why not?

Olly

 

3 hours ago, PhotoGav said:

Rodd, this is superb!! It now has a fabulous 3D quality to it. The core doesn’t jump out at me now, it sucks me in to the image and deep into space. Sweet, well done! 

 

3 hours ago, coatesg said:

Yes! This is much better - the core sits much better in the dust now - great job!

Thanks guys!

Rodd