It did take a long time to have the work finished on this and it will have a major performance boost of 30-50% over 2.0.0-beta39 from calibration to integration. We extensively optimized many critical parts of APP. All has been tested to guarantee correct optimizations. Drizzle and image resampling is much faster for instance, those modules have been completely rewritten. Much less memory usage. LNC 2.0 will be released which works much better and faster than LNC in it's current state. And more, all will be added to the release notes in the coming weeks...
Update on the 2.0.0 release & the full manual
We are getting close to the 2.0.0 stable release and the full manual. The manual will soon become available on the website and also in PDF format. Both versions will be identical and once released, will start to follow the APP release cycle and thus will stay up-to-date to the latest APP version.
Once 2.0.0 is released, the price for APP will increase. Owner's license holders will not need to pay an upgrade fee to use 2.0.0, neither do Renter's license holders.
I hope I can get some help here. I was just processing my first mosaic image, which was of M31. In NINA I set an overlap of 10% for each image.
I followed a tutorial to know which settings are important, and so I went like this:
First off, I have 4 individual integrations that I just stacked regularly with the standard settings.
Then, I deleted everything, and only loaded the 4 integrations in. I used star analysis with 5000 stars, registered with scale start 5 and scale stop 10, dynamic distortion correction, unchecked "same camera and optics", and used "mosaic" as integration mode.
Normalization was done just per default, and for integration I used MBB 10%, and 2nd degree LNC with 10 iterations, rest was kept the same.
My individual
frames look like this
Yes, the bottom one has an orange tint, I think because there were some clouds and also it was approaching dawn.
I tried the stacking both with and without the bottom right panel, but the results were identical.
This was the final result:
As each individual frame looks fine, I am pretty sure there is something with the settings I did wrong. Any advice on it?
As I am not that familiar with mosaic processing. I'd like to write you the first thought that entered my mind seeing your post.
It seems to me very hard to do a mosaic, of one object, if you have the overlapping borders so prominent in parts of the object. Even if one got "perfect" conditions.
Thats said i would try to eliminate lp in each of the 4 stacks (even if you normally should do it afterwards) and make sure background is normalized before integrating them.
I am really curious what other hints you'll get, if there are settings in app to tweak this and how the final image will look like, nice project!
Thank you for your advice. I just tried removing light pollution and also calibrating the star colors, and while the result looks better, it is still the same issue generally:
Hey Tobi! Did you color calibrate in your attempt without lp removal before integration too?
I'd give it another try, lp removal/ background correction, then integrate and do star color afterwards (because this calibrates colors in the background too, and this can add differen color than just background calibration/neutralisation).
If a color cast in the background still exists, you could use the hsl tool to desaturate 0%-background (make sure only background level pixels are selected) to even this out a bit...
There is one thing I just noticed: the gradients on the right side seem to be "reversed". So, I have a bright vignette and dark middle, while on the left it is the other way around.
Not sure how this happens, but I am very sure there is something going on when integrating.
I will wait for Mabula or Vincent, maybe they have any further ideas. 🙂
Usually you do all the processing after the mosaic is created, no need to do it beforehand as each panel may have differences in the background still, that look weird again in the mosaic as APP is normalizing across the mosaic. There could be an issue with calibration or light pollution, I can only tell with the data. Would you mind sharing a subset of the data? Like 5 lights per panel, 20 of each calibration frame (bias or darkflats, flats, darks).
Please make a folder with your forum handle and issue, like "tobigh3-mosaicGradients". The upload server address is :
Sorry, I really don't want to put pressure, but were you able to look into it already? There are some clear nights ahead, and I would love to try out the M31 mosaic with longer exposure times. Since the last 2 tries ended like this, however, I will of course steer away from it until it is being resolved.
So, looking at every step, the data seems ok to me. No overcorrection and no clipping for the flats.
The final mosaic does indeed have gradients in the background. I would suggest to give the panels more overlap (20%) while shooting the data.
Couple of questions popping up: - Are you using any filters? - Is there any light or certain light pollution in your area? - It seems the illumination of especially the top right panel is quite different, would that make any sense when you collected the data?
I didn't try to go without calibration data, but you may try that as well just to rule out any issue in that (I didn't notice them though).
2) Not specifically. I am living on the countryside, prob around Bortle 4. However, there is an airport around 30km to the west, but M31 was towards the east where it is rather dark.
3) Hm no, not sure how that could have happened. The bottom right panel looked a bit different when stacking for me, probably because there were some clouds crossing a couple times.
Might I ask what kind of gradients appeared? I mean, generally gradients shouldn't be an issue, usually APP and also other softwares can get rid of them quite well.
The issue is just that I had these inverse gradients that I couldn't explain.
I will definitely try 20% overlap next time as well, but still I must say it feels kind of odd still.
I talked to someone else who also uses APP, and he said he made a 9 panel mosaic with way stronger gradients without any issues.. Weird
Oh, btw, I also tried to not use the bottom right panel, or alternatively, perform a star color calibration on each of the panels so that the bottom right panel is not such a massive outlier. But yeah, that also did not help, sadly.
This night I did yet another test, this time with 20% overlap instead of 10%. Also, I got a new and bigger flatpanel to make sure the issue weren't faulty flats.
Your issue is in fact a know issue which is on our issue list to be solved. It is caused by 2 factors:
Firstly, you have the mosaic overlap areas in a High Dynamic Range object, the overlap area's are in very bright areas and contain very little sky background, which causes this issue partly. So with the current APP version, whenever you create a mosaic, try to prevent to have the actual object in the overlap areas, it will prevent many issues and it will greatly reduce processing problems in general.
Secondly, I think in 5) Normalize, the option to neutralize the background is making the problem bigger. Did you try processing the mosaic with this disabled? Afterwards, you can get exact background levels with the calibrate background tool.
In addition, 10% overlap is already too little in my opinion for any mosaic, most mosaics will be hard to process correctly in general. Several things are failing in that case, optical distortion correction will be less precise, leading to less correct sky geometry and the information in the small overlap areas will have less data to actually correct the sky background levels to a high degree.
Mabula
This post was modified 3 years ago by Mabula-Admin
I understand that 10% overlap might be not enough, which is why I now tested it with 20% overlap, but sadly not with a great result either. But as you mentioned, this will be most likely due to having little sky background overlap.
However, with M31, is there even a way to plan a mosaic in a good way for it to work properly in the current version of APP, do you have any idea? As most of the image consists of the galaxy in any case. I could try going with 6 panels instead of 4, so I have more sky background on the left and right, but in the middle, I would still have mostly the galaxy.
Also, does it just need more sky background in general, or is it about more sky background in each single panel? Like, would it already work better if I let's say do 10 panels in total, with most of them just from the sky background around the galaxy, or do the individual panels need to contain enough overlapping sky background? Because if the latter is the case, then I don't think (or I wouldn't know how) M31 for example could even be done with my setup, as with the 1000mm focal length I will always have little sky background overlap in some panels.
Lastly, which value of MBB do I have to choose? Is it the same as the overlap? For example, on the 20% overlap I set it to 20%, on the 10% overlap I set it to 10%, is that the correct way of using it?
Please try a reprocess with the data with background neutralization off in 5) normalize, i think that should help because on other mosaics it was creating problems like this as well and disabling background neutralization actually gave better results in those cases. I have on our issue list to remove the background neutralization option and replace it with a more advanced method to get things right in terms of illumination differences in the data.
I think 6 panels will work better, yes, then less area of m31 will be in the overlap areas, so should definitely help 😉
In general, the normalization of the panels should be improved in APP going forward, the actual normalization will be improved so will the LNC algorithm.
I know for a fact that the current LNC algorithm is creating your problem partly here. It can not deal well with object data from bright objects in the overlap areas. This is also on our todo list to improve, by creating LNC 2.0 ;-). SO the best answer right now, is that M31 with your setup right now, is simply complicated for APP in it's current state and we have several things on our issue list that we will improve to make this work much better going forward.
Yes, that is the correct way of choosing the MBB % 😉 But what can also say, that in general, that MBB % is currently static, going forward, this should be dynamic by having MBB actually analyse those overlap areas and then let is adjust locally how much MBB is used... this is on our list to do as well.
The issue here would be the middle 2 panels, as they also overlap over the galaxy center, but therefore the other ones contain more sky background.
That would not be the best way for a 6 panel M31 mosaic, because you will still have M31 in all overlap areas especially in the 2 left and 2 right panels., so it will create less issues if you have less of M31 especially in those overlap areas.
This is sounding great! Thank you so much for the clarifications. 🙂
So, for the integration of teach individual panel, can/should I keep the neutralize BG enabled under 5)?
And then when integrating the panels to the mosaic, I need it disabled, correct?
"I know for a fact that the current LNC algorithm is creating your problem partly here."
So, I chose a LNC of 2nd degree with 10 iterations, as I was just following a tutorial. I obviously don't understand any of this, to be frank, I don't even know what the difference between 1st and 4th degree and the different amount of iterations is, so all I can do is following guides here.
Could you just quickly let me know what you think is the best setting here? Thank you!
