Document S-01

Support

DepthWeaver 1.2 for iPhone, iPad and Mac · iOS 18+ · macOS 15+ (Apple silicon)

DepthWeaver is a small app maintained by one person. The fastest way to get help, request a feature, or report a bug is the GitHub issue tracker. Everything else lives below.

Quick start

  1. Capture depth. On iPhone, open the Depth tool in the drawer and tap Change depth source — LiDAR (live), Photo (on-device AI), 3D Model (orbit a bundled or imported file), Capture 3D (scan a real object with Object Capture), or a bundled height-map example. A live 3D point cloud shows you the depth you just captured. On Mac, the inspector pane offers the same sources minus LiDAR and Object Capture (no camera-based capture on macOS).
  2. Pick a pattern. In the Pattern tool, choose a built-in texture, import an image from your library as a tile, or tune one of the six procedural generators.
  3. Tune it. Start from a preset — Soft, Standard or Punchy — then move Depth (how far it pops), Spread (eye separation) and Smoothness (oversampling). Resolution (DPI), depth-range remapping and denoising live under Advanced.
  4. Share. The canvas is the result — pinch to zoom, tap to hide the controls, tap again for full screen. Save to Photos or send it through the share sheet on iPhone. On Mac, use File → Save (⌘S) for a PNG, Edit → Copy (⌘⇧C) for the pasteboard, or the standard share menu.

How to actually see the stereogram

Autostereograms (Magic Eye images) need either of two viewing techniques:

On iPhone, the fastest way in is the built-in “Learn to see it” trainer: tap the ? button on the canvas or the result screen. It gives one instruction at a time — hold it at arm's length, look through the screen, let the dots double up — over a live stereogram, and steps up the difficulty each time you say you saw it. It is also offered automatically on first launch.

Without it: hold the device close enough (or sit close to your Mac display) that the screen blurs, then slowly move away while keeping your gaze relaxed. The hidden image will lock in. It gets easier with practice.

Frequently asked questions

I still can't see the hidden image. Where do I start?

Run the “Learn to see it” trainer on iPhone — the ? button on the canvas or the result screen. It walks you through the technique one step at a time and ramps up the difficulty as you go, so you are not fighting a hard stereogram on your first attempt. If a round defeats you, tap “Still nothing” and it eases back off.

Where did the old controls go? (Updating from 1.1)

1.2 rebuilt the iPhone app around the image. The stereogram now fills the screen and every control lives in the glass drawer at the bottom: Depth for the source and depth adjustment, Pattern for textures and procedural generators, Tune for presets and sliders. Pinch to zoom the canvas, tap it to hide the drawer, tap again for a full-screen view. Nothing was removed — DPI, depth ranges and denoising sit under Advanced in the Tune tool. The Mac app is unchanged apart from the new 3D depth preview.

What do Soft, Standard and Punchy change?

They set depth strength, eye separation and oversampling together, so one tap gets you a sensible result. They deliberately leave resolution (DPI) alone, so changing DPI under Advanced does not knock you out of a preset. Move any of the three sliders yourself and the preset row simply shows Custom.

What is the 3D view on the depth screen?

A live point cloud of the depth map you just captured, rendered from your current depth adjustment. It is there so you can tell a good capture from a bad one before you weave it — if the shape looks wrong in 3D, the stereogram will be wrong too. Available on iPhone and Mac.

Does DepthWeaver run on Mac?

Yes — there is a native macOS app (macOS 15+, Macs with Apple silicon) built from the same SwiftUI codebase. The Mac UI is a document-style window with a collapsible inspector, standard menu commands (Open ⌘O, Save ⌘S, Copy ⌘⇧C, Toggle Inspector ⌘⌥I), and PNG export through the standard save panel. LiDAR live capture and Object Capture scans stay on iPhone where the hardware lives; everything else (AI depth, 3D model loading, presets, denoising, the Metal renderer) works identically on both.

Why is the LiDAR option missing on my iPhone (or Mac)?

LiDAR is only available on iPhone Pro models. On non-Pro iPhones the option is hidden because there is no sensor to read. The Mac app does not offer LiDAR live capture at all.

Why is the "Capture 3D" option missing?

Object Capture requires a LiDAR-equipped iPhone or iPad (iPhone 12 Pro and later Pro models, or recent LiDAR iPad Pro) and iOS 18+. On unsupported devices the button is hidden. The Mac app does not include Object Capture; load a pre-made .usdz or other 3D file instead. The Simulator cannot run Object Capture.

Tips for a clean 3D scan?

Use a small, opaque, matte object on a contrasting surface in even, diffuse light. Follow the orbit guides through three passes. Glossy, transparent, thin, or textureless objects (mirrors, glass, single-color plastic) are difficult for photogrammetry. Reconstruction runs after you finish capturing and may take a minute or two.

Where are my captured 3D models stored?

In a private "My captures" library inside the app sandbox. They show up alongside the bundled samples in the 3D Model picker. You can rename or delete them from a long-press menu. Nothing is uploaded.

My imported depth map looks banded or noisy. Can I fix it?

Open Depth Adjustment and enable denoising. The denoiser targets 8-bit quantization (the typical cause of visible bands) and only runs on imported depth maps — LiDAR and AI-generated depths skip it because they have real precision to preserve.

The output is blank or extremely thin.

The stereogram algorithm requires the maximum separation to be smaller than the image width. Lower the depth strength, raise the DPI, or pick a smaller pattern, then try again.

Is the GPU used for rendering?

Yes. The full stereogram pipeline (link-based hidden-surface removal, both centre-outwards pattern fills, OKLab gap fill, and OKLab oversampling downscale) runs as a Metal compute kernel — one thread per output row. If Metal is unavailable for any reason, the app transparently falls back to a parallel CPU path that produces the same output.

Where does the app save my generated images?

On iPhone, wherever you send them through the share sheet — DepthWeaver does not maintain its own gallery. On Mac, use File → Save (⌘S) to write a PNG to disk, Edit → Copy (⌘⇧C) to put it on the pasteboard, or the share menu. By default, nothing is stored outside the app's session unless you explicitly save or share.

Does the app work offline?

Yes — entirely. There is no network code in DepthWeaver. The Depth Anything V2 model is bundled inside the app.

Do you collect data?

No. See the privacy policy. There is no account, no analytics, no advertising.

What languages is it available in?

English and French. The language follows your iOS or macOS system setting.

Known limitations

Reporting a bug

Open an issue on GitHub with: your device model and OS version (iPhone + iOS, or Mac model + macOS), the DepthWeaver version (the ⓘ button on the iPhone canvas, or DepthWeaver → About on Mac), and the steps to reproduce. A screenshot or short screen recording is always welcome.

Issues
github.com/glandais/DepthWeaverApp/issues
Email
gabriel.landais@gmail.com
Source
github.com/glandais/DepthWeaverApp