For the past few days I've been taking the measure of AutoStakkert with solar system data that I already have. It has found a spot in my suite of tools, but limitations prevent it from replacing either Lynkeos or Nebulosity.

The setup

I used MS Windows for many years as an architect and programmer. I mostly abandoned it about 15 years ago, sick of its bugs, limitations, and poor security. MacOS let me focus on getting the job done with tools that were the same as I used for Linux server side work. When I started imaging I was able to find a great suite of tools running on macOS: Ski Safari, Nebulosity, Lynkeos, Astroplanner, Virtual Moon Atlas, Photoshop, Capture One, ... There is also a great legacy of tools that are limited to the Windows platform.

I was curious about AutoStakkert. My first attempt to run it used Wineskin and I quickly ran into problems. I couldn't get stacks of still image files into the program. The file packer wouldn't let me select multiple files and dragging and dropping files into the virtualized AutoStakkert didn't work. With the new ZWO planetary camera, I decided to revisit this.

Later I installed the production AutoStakkert! 2, into a Windows 10 environment running on Virtual Box. This worked well but was slow and resource intensive. I then tried out a commercial version of the Wine Windows emulator, CodeWeavers Crossover. Wine doesn't require a Windows license and is much faster and lighter weight than a full virtual Windows PC environment. I quickly had working bottles (CodeWeaver's virtual containers) for AutoStakkert!2 and an alpha version the 64 bit AustoStakkert!3, as well as RigiStax 6, and WinJUPOS for later evaluation. They all run as easily and as quickly as native macOS applications.

The only anomaly I found in my initial testing is that virtualized AutoStakkert doesn't handle the ZWO avi files that I tried. I suspect a missing library but I haven't followed up on this. This isn't important to me now, but I intend to follow up later. I decided previously to use the TIFF option for images captured from the ZWO with aoCapture. TIFF images are easier to move between the different tools that I use and I can back them up in free cloud storage on Amazon. It is also simple to cull the bad frames in a still image format.

The good news

AutoStakkert! 2 does a fabulous job on ~ 1 MP iamges from my ZWO ASI120MC planetary camera. At its best it pulls out a bit more subtle detail than Lynkeos stacks, probably because of its Multiple Alignment Points (MAP) algorithm. A major convenience is that it does a better job of automatically rejecting images with defects. With Lynkeos or Nebulosity, I need an eyes on check to make sure that I have the cleanest data in my stack. AutoStakkert does a great job on its own and makes it easier to review the stack images quickly; probably helped by their small size.

Another major convenience is the ability to automatically generate multiple stacks with different numbers of included images. I find the image quality visualization tool to be very useful as well, but its still convenient to try several cuts at the number of images to keep.

My virtualized AutoStakkert does a great job with the 1MP planetary camera images and will be part of my working toolkit.

The bad news - DSLR sized images

When I tried stacks of APS sized images from my two Sony mirrorless cameras, things quickly went down hill with error codes and out of memory messages the order of the day. I tried switching to the alpha 64 bit version of AutoStakkert! 3 and eventually was able to process some heavily cropped (down to around 4 MP using Capture One) versions of these stacks.

I'm not sure if these are application or windows issues. They occurred with the full Windows 10 install in VirtualBox so they are not Wine issues. Windows has virtual memory; I can understand slow, but not out of memory, with a 64bit app and plenty of disk space. It was a flashback to everything I hated about using Windows.

Conclusions for now

The sweet spot for AutoStakkert is small planetary camera images. I already have great native macOS tools for DSLR sized images and 4k ultra high definition video. The large image limitations are not important to me now. Eventually I'd like to compare the results of Nebulosity 8 parrameter affine transform alignment with the AutoStakkert MAP approach on full lunar disk image stacks. For now, AutoStakkert isn't up to big images, but will work just fine for me on planetary and lunar close up stacks.