Strata: Planetary Processing

Free during beta · macOS 14.6 or later · Apple Silicon and Intel

The Problem

Planetary imaging on the Mac has been an afterthought.

The best stacking software runs on Windows. The best sharpening software runs on Windows. Derotation runs on Windows. The format converters that stitch them together? Also Windows.

So Mac astronomers learned the workarounds. A Boot Camp partition that never quite worked. A Parallels VM that crashed mid-session. A second machine in the corner just to run AutoStakkert. Three apps in sequence: AutoStakkert for stacking, RegiStax for sharpening, WinJUPOS for derotation, each saving files to disk for the next one to find. None of them aware that the Apple Silicon chip in your Mac is faster at this kind of math than the PC you're emulating.

Ask on any forum how to process planetary images on a Mac and someone will tell you to buy a PC. I got tired of reading it. Mac Observatory started as the rebuttal, a running record of what does and does not actually run natively. Strata is where that argument stopped being an article and became an app.

The Solution

Seven phases. One window. One GPU.

Strata brings the core of the planetary processing toolchain home to the Mac: analyze, stack, derotate, sharpen, color, export, in one Metal-accelerated app. For the work most planetary imagers do at the desk, you no longer need Windows. Walk the pipeline straight through, or use Cmd+1 through Cmd+7 to move between phases: back to anywhere you have been, forward to anywhere you have earned.

Proof · Speed

The fastest planetary stacker on the Mac.

Every compute-heavy operation runs on the GPU through Metal. On Apple Silicon, unified memory means your frames never copy between CPU and GPU.

Benchmarks by Mac Observatory · Apple M1, 16GB

Measured by me on real captures, not synthetic benchmarks. Both columns are Strata's own code paths, run on the same machine against the same files. Your numbers will vary with file size, frame count and stacking parameters.

Proof · Selection

What's the best video I shot tonight?

Every planetary imager asks it, every session, and until now the only way to answer was to process everything and find out afterward. Filenames and timestamps tell you nothing. Seeing felt better around midnight, but that is a memory, not a measurement. So most of us pick one on instinct, keep the top 25 percent because that is the number everyone uses, and never learn what the other thirteen files were hiding.

Batch Analysis scores every capture from a session and puts them in one table. Each row carries a quality curve, how many frames survived the validity checks, how good the best quarter of them are, how good the very best moments get, any problems detected in the capture, and a single score out of 100 with the night's best capture anchored at the top. Captures that failed outright sort to the bottom with the reason attached, so a run ruined by cloud says so instead of quietly ranking last.

The score is built to resist the obvious trap. A flickery capture with a handful of brilliant frames can look excellent on averages alone, so steadiness counts, and captures without a deep enough pool of clean frames are discounted rather than flattered. The point is to rank what will actually stack well, not what has the prettiest single frame.

Every capture also gets a suggested keep percentage. Strata first checks whether your quality curve is readable at all, comparing its spread against the frame-to-frame noise in the data. If it is, it finds the elbow where quality falls away and recommends from there, held inside sensible bounds for the target: a tighter band for planets, a wider one for the Moon and Sun, where more frames are usually worth keeping. If the curve is too noisy to trust, it says so by falling back to a calibrated default rather than inventing precision it does not have. Either way you get a range to try, not a single number, and you can overrule it.

Then pick your winner, take it through the full pipeline, and come back to the table for the next one. Nothing in the Windows chain does this. AutoStakkert will stack a batch, but it will not tell you which of last night's files deserved the batch in the first place. There is also an optional Normalize Capture pass for captures with brightness or level drift across the run, which is the usual cause of a stack that looks fine in the middle and wrong at the edges.

Proof · Stacking

Strata Warp, the stacker built for Apple Silicon.

AutoStakkert pioneered alignment-point stacking on Windows nearly two decades ago. Strata Warp is my port of that idea to the GPU, rebuilt from scratch in Swift and Metal, tuned for the way Apple Silicon handles unified memory.

A uniform grid of alignment points lays over your frame. Each region is independently registered against a quality-scored reference, then accumulated with sub-pixel precision. The result recovers local detail that bulk-frame averaging blurs out: atmospheric seeing variation, residual drift, the slight misalignment that is invisible at full-frame scale but devastating at pixel level. Point placement and sizing are both automatic, with Balanced and Poor Seeing profiles for nights when the atmosphere is not cooperating.

Two paths run under the hood. Planetary targets are stacked patch by patch and feathered back together, with a crossfade across the limb so the disk edge does not betray where one region ended and the next began. The Moon and the Sun have no limb to protect and no sky to reject, so they take a different route: a measured displacement field warps each frame into place, with low-confidence regions falling back to the full-stack average rather than guessing. You do not pick the path. The imaging target you chose at Import does.

For undersampled captures, drizzle super-resolution runs at 1.5x, 2x and 3x, using the sub-pixel offsets between frames to resolve detail finer than any single frame carries. Simple Average is there as well, as a fast first look at whether a session is worth the full treatment.

Proof · Derotation

Is there a Mac alternative to WinJUPOS?

Jupiter rotates a full degree every 100 seconds. By the time you have six good captures across a thirty-minute session, the surface has shifted enough to smear together if you stack everything as one set. The same problem applies to Saturn and Mars over longer sessions.

The standard answer is WinJUPOS, a Windows app that de-rotates each stacked image back to a common reference time before combining. It works. It is also a separate app, on a separate operating system, with its own learning curve. Strata does the job in the same window as stacking and sharpening, for all seven planets. No file shuffling. No second machine.

Where there is surface detail to track, Jupiter and Mars most of all, Strata matches small patches between images by cross-correlation on the GPU and interpolates the results into a smooth displacement field. That approach needs contrast, and Saturn often has none worth speaking of. Its banding is subtle, and patch matching across a low-contrast disk gets least reliable exactly when you need it most.

So when the rings are open enough to see, Strata stops looking for features and measures the geometry instead. The spread of the bright pixels gives the ring axis and its tilt on your sensor. The thickness of the rings at their outer tips gives the ring opening angle, which is how far Saturn's pole is leaning toward you. The height of the disk, against Saturn's known 9.8 percent flattening, gives the equatorial radius. That is enough to know the planet's orientation, and combined with its rotation rate, enough to compute where every point on the globe travelled between two captures and put it back. None of it comes from an ephemeris file. All of it comes from your picture. When the rings are near edge-on and there is no ellipse left to measure, Strata falls back to patch matching.

The rings themselves are left undisplaced. They are a smooth annulus carrying no rotating features, so pushing them around with the globe is how you end up with a soft ring system attached to a sharp planet. Derotation does need to know how much time passed between captures: Strata reads that from the SER recording's embedded timestamp where it exists, and otherwise from the filename, recognizing AutoStakkert, FireCapture, SharpCap and compact date-time naming.

The Rest of the Pipeline

Sharpening, color and the plumbing in between.

Three GPU-accelerated sharpening engines, a target-aware interface, solar and lunar color, and the workflow plumbing that moves a single capture or an entire night's worth through the pipeline cleanly.

See It In Action

Processing a solar video with Solar Tone.

Watch on YouTube

"Thank you so much for your effort in creating such an outstanding program. I'm blown away. Finally, I don't have to live with that cumbersome Parallels world anymore."

— Hardy

Beta tester · MacBook Air M3 · Lunar imaging

Compatibility

Pricing

Built by one Mac astronomer. One price. Every feature.

No subscriptions. No tiers. Buy once, own it forever.

Mac Observatory Suite

Built for the Mac. Built for Astronomers.

Native macOS apps filling the gaps the astronomy community has been waiting for.

The Observing Workflow   Look Up → Capture → Process → Archive

It starts with Transit: live conditions and tonight's events tell you whether the sky is worth it. From there the planetary chain takes over, with Laminar capturing and Strata stacking and sharpening on the GPU. Meridian covers the other half of the hobby, turning a deep sky imaging archive into a catalog you can actually search.

Guiding Diagnostics   Free & Open Source

Swift & SwiftUI · Apple Silicon Native · Metal GPU · No Subscriptions

Common questions about planetary image processing with Strata on macOS.