Please note our new Downloads page here
2023-01-19: APP 2.0.0-beta13 has been released !
!!! Big performance increase due to optimizations in integration !!!
and upgraded development platform to GraalVM 22.3 based on openJDK19
We are very close now to releasing APP 2.0.0 stable with a complete printable manual...
Astro Pixel Processor Windows 64-bit
Astro Pixel Processor macOS Intel 64-bit
Astro Pixel Processor macOS Apple M Silicon 64-bit
Astro Pixel Processor Linux DEB 64-bit
Astro Pixel Processor Linux RPM 64-bit
Hi,
I shoot short exposures with an Alt/Az Dob. I get great results, but the one big issue is stacking/data management in APP. It just chokes on the amount of data for a few reasons:
1) Simply the number of lights can cause issues. Once I get above 2000-3000 lights, if I get a disruption during stacking such as a frame that doesn't register. I can't even deselect the frame because the program kind of hangs with the console window constantly cycling through re-displaying the frame list. It says something like "rebuilding the frame list" or "constructing frames detail list", but I may not have gotten the text right. Basically it's idle, but can't handle the number of frames in th list.
2) Since I get field rotation, the stacking produces a huge frame. However, for mosaics, I kind of want that data that is slightly out of the reference frame, it's good data but perhaps I must integrate into the reference frame only. Sometimes it asks for 1.5TB of disk space! I'm thinking the best thing here is to integrate at a lower scale (which I'm fine with), perhaps 0.5 for 1/4 the amount of data requirements, is that correct?
My data is already bin2, so 6.5megapixel subs, so not giant, but it can't handle the number of frames not the large integration window from a bit of drift and a lot of field rotation.
So I'm wondering which strategy you would recommend:
1) Stack each panel of the mosaic with full frame integration and then reload the finished panels and stack the whole thing? Each panel stacks easily, even in full frame integration mode even with scale set to 1
2) Stack a subset of all the frames together to create a bunch of "sub mosaics" and then stack those? Perhaps with scale <1 to keep data size down.
3) Some other strategy.
When I do smaller sets of data, it works great (good quality results), but it's just choking on the number of subs coupled with the large integration window.
Finally, which of the following options should be set?
Analyze Stars: Set more than 500?
Register: Use triangles versus quads?
Register scale stop: Set to 15?
Normalization Mode: Advanced?
Integration MBB: turn on and set to a high percentage?
Finally, if I want to stack at lower resolution, I see that the Scale in Integration can be set to <1 (ex: 0.5), but it can also be set in Registration and Normalization. Do each of these reduce the resolution and which should I use? If I set it to 0.5 in each, does it downsample further in each step so it's 0.5*0.5*0.5?
My system is a 64Gig RAM, 8 Core Ryzen 5900, 2.5TB SDD Windows system.
Thanks.
Your parameters make sense to me mostly except that in Analyze Stars, put a higher target such as 3000. In register mode, the default pattern recognition method should work for most cases. If your panels have very few overlap stars, pick triangles as suggested.
I just completed a Veil Nebula 2x3 mosaic project not long ago. There is total of 2155 subs (S: 514, H: 991, O: 650). I'm running APP on Apple M1 Mac with 16GB RAM and 1TB SSD. It's basically impossible for me to throw everything in APP and produce the mosaic. So I follow the suggestion I found in this forum by stacking each panel per channel. This is my workflow for each of the narrowband channel (SII, Ha, and OIII):
- Stack the lights for each panel
- Use Dynamic Background Extraction in PixInsight to perform detailed gradient removal for every panel and export as FITS
- Do mosaic integration in APP with the following settings:
- Tab 3 Analyse Stars:
- Target: 3000~3500
- Tab 4 Register:
- Pattern recognition: quadrilaterals
- Scale stop: 10
- Use dynamic distortion correction: Yes
- Same camera and optics: No
- Registration mode: Mosaic
- Tab 5 Normalize:
- model: Advanced
- Tab 6 Integrate:
- LNC degree: start with 1st degree LNC
- LNC iterations: 3
- Enable MBB: Yes, and set the value at 15% for my case (my panels have roughly 15% overlap)
- Tab 3 Analyse Stars:
This is the final result:
Thanks for the reply. I don’t use Pixinsight (but I guess I eventually should), is there any equivalent in APP or an editing program like Affinity/Photoshop?
Also, I tried the mosaic mode on 3 of the 4 panels to test it a couple of times and one of the frames created this half circle arc outside the other frames (way way off, it looks like one of the Hubble gravational lending photos, lol…) but when I stacked with regular mode, not mosaic mode, it worked pretty well. In other words I’m actually getting worse results in mosaic mode and I don’t know why. I think I have most of the settings as described. I’ll keep trying, but non mosaic mode looked ok, it had some boundary issues with replicated stars but fixable.
any idea what I likely did wrong?
Thanks for the reply. I don’t use Pixinsight (but I guess I eventually should), is there any equivalent in APP or an editing program like Affinity/Photoshop?
APP has the very powerful light pollution removal tool. No need for PixInsight at all.