Free focus stacking · in your browser · full resolution
Drop a focus-bracketed series. Your own computer aligns the frames and merges the sharp regions into one photograph — Mac, Windows, Linux, Chromebook — and the frames never leave it.
Nothing is uploaded: the stacker never sends a frame anywhere.
Stacking on this device
Result
Paint over a halo or a soft patch: the round brush copies that area from one aligned source frame — by default the frame that is sharpest where the stroke starts. Choose a frame yourself when the default is wrong. While you retouch, the downloads are put away; they are made again, with your strokes, when you press “Done retouching”.
Keyboard: with the image focused, the arrow keys move the brush ring, Space paints, [ and ] change the source frame, U undoes the most recent stroke. “Move the view” makes dragging pan instead of paint.
At 100 %: drag or use the arrow keys to pan; press B to compare with the sharpest single frame.
No file yet.
Focus breathing scales and shifts every frame a little. Each one is fitted to the middle frame on its edges — scale, rotation and shift — so a 7 % drift across a long bracket still lines up.
Every frame is broken into six scales of detail. At each scale, every region goes to the frame where it is sharpest, and single-pixel mistakes are outvoted by their neighbours.
The winning detail folds back into one image. Two passes keep only the vote, so memory does not grow with the number of frames.
How it works, in detail · How to shoot a series that will merge
The specialised desktop stackers are good, and most of them are licensed: a trial, then a fee. These are the things this page does at no cost, on any computer, as of 4 October 2026.
| What you get | Focus Stacks | Typical paid stacker |
|---|---|---|
| Full-resolution output, no watermark | Yes on a computer (a Fujifilm RAW is read at a third of its size; phones and “Fit to memory” stack smaller) | During a 30-day trial; Helicon Focus then drops to 4 MP with a watermark |
| RAW straight from the camera, Fujifilm included | Yes, decoded in the browser | Helicon: yes; Zerene: convert to TIFF first |
| 16-bit TIFF export | Yes | Yes |
| Runs on a Mac, Windows, Linux or a Chromebook | Yes, nothing to install | An installer for Windows and macOS (Zerene also Linux) |
| Your frames stay on your computer | Yes | Yes |
| Retouching brush for halos | Yes: one round brush, painting from any aligned frame | Yes, with more brushes (Helicon: Pro and Premium) |
| More than one stacking method | Two: pyramid and depth map | Helicon: three; Zerene: two |
Yes. The stacker runs in your browser at full resolution with no account needed, no watermark and no frame limit other than your computer's memory. The site keeps space for advertising on its pages, never inside the tool.
Yes — anything that runs a current browser. The stacking happens on your own processor, so there is nothing to install and no Windows-only download. A phone or iPad is limited to ten frames at reduced size because of memory.
No. The frames are read by the page in your browser and merged there. The page makes no request that carries a frame, a file name or a pixel of one; a second stack works with the network switched off. The one upload is one you ask for: signed in, “Create link” under “Share this stack” uploads a smaller copy of the finished picture and its depth map and, while the before/after is ticked, a copy of the one sharpest frame; never the series. The how-it-works page lists every request the page makes.
JPEG, PNG, WebP, AVIF, TIFF (8- or 16-bit, in sRGB, as Lightroom, Capture One, darktable and RawTherapee export it) and RAW straight from the camera — DNG (mosaic; ProRAW and other already-demosaiced DNGs not yet), CR2, CR3, NEF, ARW, RAF, ORF, RW2 and PEF, decoded here in the browser. HEIC decodes in Safari. Frames must all have the same dimensions and one format: a bracket is stacked as RAW or as developed pictures, not both.
RAW, when you want the most from the sensor: the frames are merged in linear light with the camera's as-shot white balance and its colour matrix, and the tone curve is applied once, at export, with an automatic exposure. Developed frames, when your look is already made in your RAW software — then stack what you exported, as 16-bit sRGB TIFF for the most precision. Either way the frames never leave this tab.
As many as your memory allows: the engine keeps a few frames' worth of memory whatever the count, and the page tells you before it starts what a stack of your size needs. Fifty 24-megapixel frames are fine on a laptop with 8 GB.
That service sends every frame to its server and caps the count. This one does the same job on your computer, at full resolution, for as many frames as fit in memory, and the frames never leave the tab.
It uses the same family of method as enfuse and Zerene's PMax — align the frames, then choose the sharpest source for every region at every scale. Measured on seven hard thirty-frame macro brackets from a public research dataset, it matched Helicon Focus's own results on those series, with no visible halo in any of them. It has both families of method — the pyramid and a depth map — and a round retouching brush that paints from any aligned frame. The paid stackers' retouching is more mature (several brushes, painting from the other method's result), and they add speed on very long stacks, batch processing and rail control, among other things; the comparison page sets out what each one costs and does.
Start with the pyramid: it keeps the finest detail — hairs, bristles, crossing edges — and is the default. The depth map takes each point whole from the frame where it was sharpest, blending only the two nearest frames, so the colour of every point comes straight from a photograph rather than being reassembled from bands of detail; try it on smooth subjects such as coins, shells and petals, and compare the two at 100 %. It smooths its choice of frame, so detail that crosses depths — a hair in front of a leg — is where the pyramid does better. Either way, the retouch brush fixes the odd region.
Yes, when the frames carry them (JPEG, TIFF and camera RAW; HEIC, PNG, WebP and AVIF frames are stacked without) and the box is ticked when you start the stack: the JPEG and the 16-bit TIFF carry the first frame's camera, lens, date, exposure and focal length, plus a note of how many frames were stacked. The GPS position, serial numbers and the maker's private notes are never copied, because a stack is often shared.
Yes. In Chrome or Edge on a computer, use Install in the address bar; on Android, Add to Home screen; on an iPad or iPhone, Share then Add to Home Screen. Installed, it opens in its own window and, once it has been opened online, opens and stacks without a network; installed from Chrome or Edge on a computer, it also appears under “Open with” for photographs. Nothing is uploaded either way.
Yes. Press “Load our 7-frame sample” under the drop zone: the page loads a seven-frame bracket of a circuit board from this site (Petteri Aimonen's MIT-licensed test series, at 1024 × 768), then “Stack these frames” runs the whole stack, and you can inspect it at 100 % and download the result.
After a stack you can save a greyscale picture showing which frame was sharpest at each point — white where the first frame of the bracket won, black where the last one did. It is a rough relief of the subject: useful for checking that the bracket covered it, and as a starting mask in an editor. Flat, featureless areas carry no focus information, so the map guesses there.
Yes. After a stack, “Prepare 3D clip (MP4)” rocks the picture gently from side to side for 4.5 seconds, moving each point by how near it was in the bracket, at up to 1080 pixels on the long side. It is made on your computer, from the result you see (retouching included), and needs a browser that can encode video in a page: current Chrome, Edge or Firefox (Safari is not supported yet). “Prepare 3D clip (GIF)” makes the same rock as a looping GIF, at up to 540 pixels, in any browser, Safari included. It looks best on a subject with clear near and far parts; the map assumes the bracket starts at the nearest point, as camera bracketing modes do.
Yes. Drop one video (MP4, MOV or WebM) of a focus pull, from a microscope or a phone, instead of photographs. The page plays it on your computer, looks at about ten frames a second, keeps the ones that are the sharpest somewhere in the picture (up to 80), and stacks those like photographs. It reads the first 30 seconds; longer videos are a planned Pro feature. The video never leaves your device.