Reprocess with new PI tools

[Jockinireland]

This is a reprocess of data from last year to see how much difference/benefit can be gained fro the new WBPP with LN and the NoiseXterminator/StarxTerminator modules.

Its a bit unscientific because, of course, my skill have improved abit but i think it is still interesting.

This was my final image last year - i was quite happy with it.

This is my new image processed last night. I am very happy with it and i believe it shows a marked improvement

But the real difference becomes apparent on zooming in

old

new

Old

new

I think this is a remarkable improvement in terms of noise but also detail, contrast etc.

[messier 111]

splendid work , thx .

[Johnny Carter]

I would agree that your new process gives a better result, well done.

[STEVE333]

Remarkable improvement David. Much more detail and improved contrast. Not sure how much is due to new WBPP with LN and how much is due to the two Xterminator modules, but, the improvements are undeniable!

I too am a fan of the two Xterminator modules.

Steve

[Jockinireland]

Thanks @STEVE333

Yes its unclear to me how much is the new WBPP, how much the Xterminators and how much is just my better ability.

I did a quick comparison between the two WBPP outputs. I applied the extra stretch in WBPP (cant remember the proper technical term for that!) as it showed the differences more.

New is to left.

The obvious difference is the bottom border where I had a mismatch after meridian flip. That is completely resolved by the LN process. The overall background gradient is obviously improved also. This is also expected from the use of LN. After that the difference is more subtle but there is some improvement in contrast in areas of the heart, particularly in the areas to left, bottom and top right of the dense parts. I think that there is also an improvement in the central very faint part of the heart. I think all of this is beneficial to the final processing - but it is fairly subtle.

So I think the most significant improvement is from the XT's. I found that I got even better noise reduction by separating the channels before star removal and subsequent noise reduction. If I did it in other ways ( for example star removal-Noise reduction- separate channels) the noise reduction, while still pretty good, was not quite as effective.

I expect that my improved ability with PI has helped some but I think that the output of NoiseXterminator has given a start point which is much easier to process.

If anyone is interested in doing their own comparison I'd be happy to share the two stacks (the original was shared in one of our processing challenges already)


take care

David.

[Juno16]

[background][/background]Amazing difference! It looks like you added 10 more hours of data too!
I agree. LN works very well. Especially, for me, getting the defaults set correctly.
The Xterminator products are awesome.

Your original process is good. Your reprocess is stellar!

[bobharmony]

That is a fabulous difference, David. Looks like you have lots of work to do on upcoming cloudy nights! I have been using the NXT plugin for PhotoShop and it is an amazing tool, even though in that environment it can only be applied in the non-linear phase. I love seeing these results from PixInsight, it gives me hope for when I finally get around to moving there.

Nice work and please keep us abreast of your progress redoing everything else you have ever shot!

Bob

[Jockinireland]

That is a fabulous difference, David. Looks like you have lots of work to do on upcoming cloudy nights! I have been using the NXT plugin for PhotoShop and it is an amazing tool, even though in that environment it can only be applied in the non-linear phase. I love seeing these results from PixInsight, it gives me hope for when I finally get around to moving there.

Nice work and please keep us abreast of your progress redoing everything else you have ever shot!

Bob

Thanks Bob - yes you are right no shortage of reprocessing in sight - but what else would you do in midsummer at 53N

I normally dont like it when people talk about things in terms of being a "game changer". I prefer the term "quantum leap" and honestly I dont think I have used that term more than a couple of times. Ever. But for for me NoiseXTerminator, allied with starXterminator is indeed a Quantum Leap. (Or game changer if you prefer )

I think that Jim's @Juno16 statement above hits it square, "It looks like you added 10 more hours of data too". I recently posted 2 images each with less than 2 hours integration, which in my view would have been virtually unusable without additional hours of integration but with NX were quite presentable.

Right, I'm off into the archive!!

David

[yobbo89]

very nice, apparently theres a new tool update, starnet 2.0 . i'll also have to give the ones you mentioned ago aswell,your image looks much better.

[Ylem]

That's an amazing improvement, of course the original data is amazing in itself

[STEVE333]

Thanks @STEVE333

Yes its unclear to me how much is the new WBPP, how much the Xterminators and how much is just my better ability.

I did a quick comparison between the two WBPP outputs. I applied the extra stretch in WBPP (cant remember the proper technical term for that!) as it showed the differences more.

heartsxs.JPG

New is to left.

The obvious difference is the bottom border where I had a mismatch after meridian flip. That is completely resolved by the LN process. The overall background gradient is obviously improved also. This is also expected from the use of LN. After that the difference is more subtle but there is some improvement in contrast in areas of the heart, particularly in the areas to left, bottom and top right of the dense parts. I think that there is also an improvement in the central very faint part of the heart. I think all of this is beneficial to the final processing - but it is fairly subtle.

So I think the most significant improvement is from the XT's. I found that I got even better noise reduction by separating the channels before star removal and subsequent noise reduction. If I did it in other ways ( for example star removal-Noise reduction- separate channels) the noise reduction, while still pretty good, was not quite as effective.

I expect that my improved ability with PI has helped some but I think that the output of NoiseXterminator has given a start point which is much easier to process.

If anyone is interested in doing their own comparison I'd be happy to share the two stacks (the original was shared in one of our processing challenges already)


take care

David.

Thanks David for explaining the improvements coming from the WBPP LN. As I mentioned before I've never tried it and am looking forward to trying it on my next set of data. I haven't taken any data in some time (I don't shoot during galaxy season because I only have a WFOV setup) so I'm itching to give it a try.

Steve

[Juno16]

Thanks @STEVE333

Yes its unclear to me how much is the new WBPP, how much the Xterminators and how much is just my better ability.

I did a quick comparison between the two WBPP outputs. I applied the extra stretch in WBPP (cant remember the proper technical term for that!) as it showed the differences more.

heartsxs.JPG

New is to left.

The obvious difference is the bottom border where I had a mismatch after meridian flip. That is completely resolved by the LN process. The overall background gradient is obviously improved also. This is also expected from the use of LN. After that the difference is more subtle but there is some improvement in contrast in areas of the heart, particularly in the areas to left, bottom and top right of the dense parts. I think that there is also an improvement in the central very faint part of the heart. I think all of this is beneficial to the final processing - but it is fairly subtle.

So I think the most significant improvement is from the XT's. I found that I got even better noise reduction by separating the channels before star removal and subsequent noise reduction. If I did it in other ways ( for example star removal-Noise reduction- separate channels) the noise reduction, while still pretty good, was not quite as effective.

I expect that my improved ability with PI has helped some but I think that the output of NoiseXterminator has given a start point which is much easier to process.

If anyone is interested in doing their own comparison I'd be happy to share the two stacks (the original was shared in one of our processing challenges already)


take care

David.

Thanks David for explaining the improvements coming from the WBPP LN. As I mentioned before I've never tried it and am looking forward to trying it on my next set of data. I haven't taken any data in some time (I don't shoot during galaxy season because I only have a WFOV setup) so I'm itching to give it a try.

Steve

Hi Steve @STEVE333 ,

When you get a chance to try LN in WBPP, make sure to reset WBPP to default settings. Or, insure that you start out with the new defaults for LN as shown in the image below.
PI will retain the old LN defaults when WBPP is updated which for me caused a gradient or color background splotching issue.

In my case, the background issue was caused by the "Reference frame generation" parameter was set to "Single best frame", instead of the new default Integration of best frames". As long as you reset WBPP, you should be okay.

David @Jockinireland , please add anything especially if your experience differs.

The difference using LN over not using LN is definitely significant.

Thanks and good luck Steve!

[STEVE333]

Hi Steve @STEVE333 ,

When you get a chance to try LN in WBPP, make sure to reset WBPP to default settings. Or, insure that you start out with the new defaults for LN as shown in the image below.
PI will retain the old LN defaults when WBPP is updated which for me caused a gradient or color background splotching issue.

In my case, the background issue was caused by the "Reference frame generation" parameter was set to "Single best frame", instead of the new default Integration of best frames". As long as you reset WBPP, you should be okay.

David @Jockinireland , please add anything especially if your experience differs.


Capture.JPG


The difference using LN over not using LN is definitely significant.

Thanks and good luck Steve!

Thanks Jim. I wouldn't have thought to set everything to defaults. Will do.

I'll report back how it works.

Steve

[Juno16]

Hi Steve @STEVE333 ,

When you get a chance to try LN in WBPP, make sure to reset WBPP to default settings. Or, insure that you start out with the new defaults for LN as shown in the image below.
PI will retain the old LN defaults when WBPP is updated which for me caused a gradient or color background splotching issue.

In my case, the background issue was caused by the "Reference frame generation" parameter was set to "Single best frame", instead of the new default Integration of best frames". As long as you reset WBPP, you should be okay.

David @Jockinireland , please add anything especially if your experience differs.


Capture.JPG


The difference using LN over not using LN is definitely significant.

Thanks and good luck Steve!

Thanks Jim. I wouldn't have thought to set everything to defaults. Will do.

I'll report back how it works.

Steve

Personally, I am not a big fan of setting anything back to default because I might not remember a change that I made that I wanted to keep before resetting.

If you are hesitant about resetting, you can even give it a try with the existing settings and see how it goes. For me, I had a large dataset that took 2 hours to process, so experimenting was very time consuming!

I posted a link to a thread on CN relating to the issue in an earlier post, but you can look at it if you want to and see where others had a problem with their defaults.

PI has a policy of not overwriting existing settings because users have changed existing settings (defaults) to get better processing results for their images and they don't want to wipe that out. Makes sense.

In the case of LN, the "Single best frame" setting was the default in previous versions of WBPP LN and when the new WBPP came out recently, LN was offered with a new Reference Frame Generation setting of "Integration of Best frames". If someone had "Single best frame" selected (previous version default), then the PI update will not overwrite that setting with the new version default. The PI support team recommends resetting WBPP to defaults to take advantage of the newest recommended defaults. That CN thread is pretty big, but has some good information.

If you want to use the new recommended LN defaults without resetting everything, just make sure that your settings in LN match what is in the image in my previous post.

Good luck Steve and please respond back with your findings.

[STEVE333]

Hi Steve @STEVE333 ,

When you get a chance to try LN in WBPP, make sure to reset WBPP to default settings. Or, insure that you start out with the new defaults for LN as shown in the image below.
PI will retain the old LN defaults when WBPP is updated which for me caused a gradient or color background splotching issue.

In my case, the background issue was caused by the "Reference frame generation" parameter was set to "Single best frame", instead of the new default Integration of best frames". As long as you reset WBPP, you should be okay.

David @Jockinireland , please add anything especially if your experience differs.


Capture.JPG


The difference using LN over not using LN is definitely significant.

Thanks and good luck Steve!

Thanks Jim. I wouldn't have thought to set everything to defaults. Will do.

I'll report back how it works.

Steve

Personally, I am not a big fan of setting anything back to default because I might not remember a change that I made that I wanted to keep before resetting.

If you are hesitant about resetting, you can even give it a try with the existing settings and see how it goes. For me, I had a large dataset that took 2 hours to process, so experimenting was very time consuming!

I posted a link to a thread on CN relating to the issue in an earlier post, but you can look at it if you want to and see where others had a problem with their defaults.

PI has a policy of not overwriting existing settings because users have changed existing settings (defaults) to get better processing results for their images and they don't want to wipe that out. Makes sense.

In the case of LN, the "Single best frame" setting was the default in previous versions of WBPP LN and when the new WBPP came out recently, LN was offered with a new Reference Frame Generation setting of "Integration of Best frames". If someone had "Single best frame" selected (previous version default), then the PI update will not overwrite that setting with the new version default. The PI support team recommends resetting WBPP to defaults to take advantage of the newest recommended defaults. That CN thread is pretty big, but has some good information.

If you want to use the new recommended LN defaults without resetting everything, just make sure that your settings in LN match what is in the image in my previous post.

Good luck Steve and please respond back with your findings.

Thanks Jim - The suggestions and warnings are much appreciated.

Steve

[starfield]

Nice improvement, David. As for my own experience with LN, I've found that for shots taken from my LP backyard, it's really important to take a look at the frame or frames that are being selected. If I go with defaults I can sometimes end up with a less than desirable reference frame. For something shot from a darker site, without my crazy gradients, things might be different.

[Kanadalainen]

StarX and NoiseX are currently the best in the business.

Your images are super nice too.