Quick start

In the browser

Open lapstack.app.ragtux.com. Nothing installs and nothing uploads; the engine runs on your own GPU inside the tab.

Drag the frames of one stack onto the window, or use Add frames… in the toolbar. PNG, JPEG and TIFF (eight- or sixteen-bit) and camera raws all work — see frames for the full list. Add folder… takes a whole folder, subfolders included.

Thumbnails appear in the filmstrip down the left as the frames are decoded. The order in the filmstrip is the stacking order; frames are expected to run from one end of the focus sweep to the other, in either direction.

2 · Run it

Press Run LAP. The defaults are the right starting point for almost every stack: alignment on, brightness equalization on, the pyramid depth chosen from the image size, a 3×3 energy window.

The log shows each frame being decoded, registered and folded in. When it finishes you are looking at the fused result, and the depth pass has already run.

3 · Look at what you got

The viewer's header has three groups:

Source — the aligned frames, as shot. Turn on peaking to see the in-focus band sweep through the scene as you scrub.
Stack — the fused result (LAP).
Depth — the depth map (Focus depth), its Confidence, and In focus, which shows one frame's own pixels with everything the result did not take from it dimmed away.

Scroll to zoom at the cursor, drag to pan, double-click to toggle fit and 100 %. The arrow keys and the wheel scrub through frames on the views that depend on one. Turn on compare to put two layers side by side under one zoom and pan.

Zoom to 100 % and go looking for trouble: soft bands (the focus step was too large), ghosting (something moved), and halos or fringes where a near edge crosses a far background.

4 · Fix what needs fixing

A soft rim beside a bright object → raise halo control from 0 and run again. See fusion.
Crossing structures — hair, bristles, antennae — taking the wrong frame → tick also render from the depth map in the Run ▾ menu and run again, then retouch one rendering into the other.
Dust spots in every frame → load a dust map.

5 · Save

Press 2 Save (or the 2 key). You get a file list: the stacked image, the depth map, the confidence map, any animations, the stereo pair, the 3D model. Tick what you want, choose a format and bit depth, set the name, and save — to a folder you pick (Chrome and Edge) or as downloads.

For a result you will edit further, save sixteen-bit PNG or TIFF. The stacked image carries the first frame's EXIF, ICC profile and XMP.

From the command line

Note

The command-line tool is not released yet. This section is here so you can see what the same run looks like when it is.

The same run, on the native binary:

shell

lapstack --align-coarsen 2 --save-depth -o stacked.tif frames/*.tif

That aligns, fuses, runs the depth pass, writes stacked.tif cropped to the area every frame covers, and writes the depth map beside it. Add --gpu --gpu-align on a machine with CUDA and a build that has the feature:

shell

lapstack --gpu --gpu-align --align-coarsen 2 --save-depth -o stacked.tif frames/*.tif

Raws work directly, and -o stacked.dng writes a linear DNG your raw converter will develop like a raw:

shell

lapstack --align-coarsen 2 -o stacked.dng shoot/*.NEF

lapstack --help lists every option; the option reference groups them by what they do.

Tip

Before a long run, try --draft 2. Every frame is block-averaged by four as it is decoded, so the whole run — alignment, fusion, depth, every output — happens at a sixteenth of the pixels and a few seconds tells you whether the settings are right. The browser app has the same control as draft in the parameter panel.