Mar 28 2026 APP 2.0.0-beta40 will be released in 7 days.
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 am still very green with APP and it is something that I am using for mosaic creation at the moment and it is doing a fantastic job. I do have one issue though and that is ultimately with Registration which I will go into briefly but in a way of fixing that, I ran into a Normalisation Error.
I am creating a 9 panel mosaic with both RGB and L which I have Calibrated and Stacked in PixInsight. The luminance is actually an extracted CIE*L from the RGB frames and then individually processed but they are put in at different scales into APP due to only having 16GB of RAM in my laptop. I'm loading the L at full resolution and the RGB at half resolution and then registering the RGB to L in PixInsight afterwards. Now although they're essentially the same data, there is a minute amount of different distortion correction when stacking the L and RGB individually. It is within a few pixels but when eventually combining the L and RGB it does look a bit wrong.
So, I have loaded the 9 Luminance and 9 RGB frames into APP as different channels and selected them accordingly. Registration of them is quick and easy but the normalisation (Advanced) crashes and I believe this is because the reference frame is a Luminance and APP isn't liking normalising RGB frames against a L.
So then I figured that maybe I could deactivate the L, normalise the RGB and then do the same for normalising the L but it came up with another error.
Finally I deleted the L frames but in doing so you remove the reference frame and you cannot use the "Advanced" normalisation on the RGB data without re-registering against one of the RGB frames and then I'm just back where I started.
I have since tried extracting the RGB channels from the RGB data and then running the 36 LRGB frames but I gave up after an hour as it had only made it 40% of the way through the Registration process.
I am wondering if there is something I am doing wrong or a box I need to tick? Whether you cannot mix L and RGB frames together? Or if there is simply a way of registering the RGB data against the already previously stacked L integration (about 800 megapixels and ~500,000 stars)?
So did you try to do the entire processing in APP or does that run into the mentioned memory problems? It does a great job of processing multiple session and multiple filter data. You just click those options in module 1 (Load) and all will be taken care of. The subsequent combining or RGB and L will then be done in module 9 (Tools). Using Pixinsight to calibrate and register and then trying to do this with APP might work, but I don't think it's the best route.
What I have been doing so far is most of my processing in PixInsight. I am using a Nikon D810 and a 130mm F/5 refractor so I am drizzle integrating in PixInsight, extracting the CIE*L, doing noise reduction and colour calibration on the RGB and deconvolution and noise reduction on the Synthetic L. I'm doing this for all 9 panels and then loading everything into APP.
This afternoon I wanted to try doing a 6 panel mosaic (of 9 planned) of M42. It was 47 calibrated frames in 6 panels and that ran into memory problems (required about 175gb of HD space that I didn't have). What I am running into doesn't seem to be memory issues as it is happening during the Normalisation stage and not even making in to the Integration.
I have previously done 40 panels as a test of just RGB, it took all night but ran into no issues apart from having to do heavy downsizing (could only output a 300mb image) which is all my 16 GB of RAM can handle.
Mmm, a lot of the steps you do in PixInsight are also possible with APP, apart from the deconvolution. Would it be possible for you to try the first steps, including drizzling in APP and then see? You could then take the RGB and L to PixInsight and perform deconvolution (unless that has to be done before integration, I'm not sure about that but @mabula-admin definitely knows that). If possible, you can take those 2 back into APP and process. But I usually try to perform the entire calibration process in APP and then decide to go further with either APP or PixInsight.
I'd need considerably more RAM I'm afraid, 50-60 GB by my calculations. The largest RGB image that APP can output on my computer with 16GB of RAM (using 14 GB of RAM) is about 315 megapixels and as a 9 panel drizzled mosaic about 1.2 gigapixels, I cannot do it in APP without spending a considerably amount on a new computer. I'm using a MacBook Pro so I cannot upgrade it from 16GB of RAM.
Now, before you ask, I can output considerably larger Luminance images because they're only a single channel. I can actually output a 9 panel drizzled mosaic at near full resolution with the luminance. I don't need the RGB and L channels to be the same output size as I can upscale in PixInsight afterwards but I do need identical registration and Dynamic Distortion Correction. If I do the L and the RGB separately they do get close but not close enough. When I am combining them back together the distortion correction is different enough that it is causing some really weird artefacts.
What I need to be able to do is register the RGB and Luminance. The resulting RGB image is going to be downsized on output but that doesn't matter as it is a VERY simple upscale process in PixInsight. The errors in the first post are during the Normalisation and stopping me from getting to the Integration process.
Hi @atmosfearic & @vincent-mod,
"So, I have loaded the 9 Luminance and 9 RGB frames into APP as different channels and selected them accordingly. Registration of them is quick and easy but the normalisation (Advanced) crashes and I believe this is because the reference frame is a Luminance and APP isn't liking normalising RGB frames against a L."
Yes, I should catch this issue and warn the user. Registering monochrome to 3-channel data is possible. Normalization is not however. So that's the issue and that causes the error message.
Regarding issues with size/memory: if you are creating a big mosaic, why are you making it complicated and are use Drizzle integration? Do you need to make a huge canvas for print perhaps? A mosaic will be so large that Drizzle makes it only more complicated. Drizzle with scale increase of 2, needs 4 times as much memory. And drizzle could give artefacts which makes star location calculation less reliable giving worse or even unstable mosaic registration. Furthermore, drizzle integrations will have more noise normally so for a big field of view where details on pixelbasis won't be seen, I would not choose that normally.
Regarding using registered data from other applications: if you created the panels in PI using Thin Plate Splines (Pi's so-called distortion correction) then chances are big that the actual data geometry is screwed up. APP, in some cases, will not be able to fix that, since it assumes correct optical geometry of the data. Thin Plate Splines can bend in all sorts of ways making the data unrealistically/scientifically incorrect. So if you used Thin Plate Splines, I can only suggest to do the registration completely in APP to get a better and more robust result. And registering the RGB data in APP to a L which was created with PI can be problematic due to PI's registration engine with thin plate splines.
I would suggest the following workflow.
1) split the channels of the RGB data into R,G and B frames
2) create the R,G,B and L mosaic panels from the individual frames per panel.
3) register the R,G,B and L mosaic panels into a mosaic. The best way to register a mosaic is to do it in one calculation.
4) integrate it without drizzle. Verify if APP can perform this on your system. If so, you can then try to integrate with a bigger scale factor to increase resolution.
Please let us know if you need more assistance.
Mabula
