DepthMakerMac converts 2D images into real-time 3D stereoscopic views using CoreML-based AI depth estimation models (Depth Anything V2 / Depth Pro).
Download Installer (DepthMakerMac006.dmg 25th Apr. 2026)
Added functionality to convert 2D and 2D 360° videos to 3D and 3D 360° videos
Fixed a bug where colors turned blue when saving in SBS format
Added a 2D-to-3D conversion and saving feature for spherical images; this feature has also been added to batch conversion
Added focus adjustment when saving in SBS or 3D360 formats
Fixed an issue where Depth Pro did not work on M1 and M4 Pro.
The bundled CoreML model (DepthAnythingV2SmallF16.mlpackage) is compiled on first launch and cached in ~/Library/Application Support/DepthMakerMac/. Subsequent launches use the cached compiled model instantly. License: Apache-2.0.
Selecting Depth Pro triggers an automatic download from HuggingFace (coreml-projects/DepthPro-coreml) on first use. After downloading, it is compiled and cached. Subsequent uses start immediately. License: Apple Depth Pro License.
Depth Pro uses a 1536×1536 input and produces significantly sharper depth maps than the small V2 model. On Apple Silicon (M-series), inference takes roughly 10–30 seconds thanks to the Neural Engine.
Save SBS… (⌘S): Saves a side-by-side stereo JPEG using your current Depth slider value. Default filename: <original>_sbs.jpg.
Save RGBD (⇧⌘S): Saves the RGBD JPEG (left=color, right=depth map). Reload it later with RGBD Image… for instant 3D viewing without re-running AI.
Focus Value: Controls the convergence point (zero parallax) for SBS and Spherical 3D saves. You can adjust it intuitively using the slider in the sidebar or enter a value from 0–255 directly. For precise adjustment, click Pick then click on the depth map to select a value visually. A red cursor will indicate the selected point; click Confirm to set the value. This ensures the selected depth level appears on the screen plane.
Save 360° 3D… (⇧⌘E): Saves a 360-degree 3D image in T/B (Top/Bottom) format from a 2D equirectangular image and its depth map. Unlike standard saving, it uses a special process to ensure the left and right eye images connect seamlessly at the edges. The current Depth and Focus settings are applied during generation.
Exported 360° 3D images can be viewed interactively by dragging and dropping them onto the following site:
3D 360 Viewer
Paste into apps like StereoPhoto Maker for advanced 3D editing.
| Mode | Description |
|---|---|
| Side-by-Side (SBS) | Full-width parallel stereo pair |
| Half SBS | Half-width SBS for 3D TVs |
| LR/RL | Left/Right swapped |
| Anaglyph Mono | Grayscale red-cyan glasses |
| Anaglyph Color | Color red-cyan glasses |
| Dubois Anaglyph | High-quality red-cyan (ghost reduction) |
Drag image files onto the viewer area to load them for AI processing.
If multiple images are in the same folder, you can switch between them using the < / > buttons in the sidebar or the keyboard shortcuts (⌘← / ⌘→). This is particularly useful for quickly reviewing a series of RGBD images or continuously converting 2D images to 3D.
| Key | Action |
|---|---|
⌘O | Open 2D Image |
⇧⌘O | Open RGBD Image |
⌘S | Save SBS Image |
⇧⌘S | Save RGBD Image |
⇧⌘C | Copy SBS to Clipboard |
⌥⌘C | Copy RGBD to Clipboard |
X | Swap L/R |
R | Reset viewer settings |
⌘B | Toggle sidebar |
⌘← | Previous Image |
⌘→ | Next Image |
↩ | Generate 3D |
⇧⌘B | Batch Conversion |
⇧⌘V | Video 3D Conversion |
Click Batch Conversion in the sidebar (or ⇧⌘B). Select an input folder and output folder, choose the export format (SBS, Spherical 3D, RGBD, SBS + RGBD, or 360 + RGBD), and set the Focus and Depth values to be used for the batch. All supported images will be converted sequentially. System sleep is disabled while processing.
Click Video 3D Conversion in the sidebar (or ⇧⌘V). Load a 2D video and convert it to 3D (SBS or Spherical 3D) using AI depth estimation. The current Depth and Focus settings are applied to the video. You can choose from MP4 (H.264/H.265) and MOV (HEVC) formats. Audio track copying is supported. The process utilizes hardware encoders and disables system sleep during conversion.
Resolution Limits: Due to hardware encoder specifications, the maximum supported output resolutions are: