Repository navigation
Releases: AbhayKale332/DW-Team_ENDRA
Release list
DepthWizard v1.0.1
DepthWizard Desktop v1.0.1
Download DepthWizard for Windows, macOS or Linux, with local model weights included or as a smaller application using hosted inference.
Changes
- Smaller installers and native runtime by excluding unused inference tooling, native test programs, duplicate sample assets, and default source maps.
- Linux native libraries discard unused local symbols while preserving their dynamic exports.
- Lower transient memory use for large raster overviews, tiled prediction, image textures, and worker inputs. Model weights, precision, and numerical outputs are unchanged.
- Offline samples now include only PNG Wankhede Stadium, Mumbai and the georeferenced Cartosat-2S TIFF project, with their previews.
- Both with-model and without-model installers are checked on Windows x64, Linux x64, Apple Silicon, and Intel Mac.
Installation
| Platform | Download | Install |
|---|---|---|
| Windows 10/11, x64 | .exe |
Run the installer. |
| macOS, Apple Silicon | arm64 .dmg |
Open the disk image and copy DepthWizard to Applications. |
| macOS, Intel | x64 .dmg |
Open the disk image and copy DepthWizard to Applications. |
| Linux, x64 | .AppImage |
Mark the file executable and run it. |
Choose a filename containing with-model for offline local inference, or without-model for the smaller hosted-inference download. Both include the application runtime; Python and Node.js are not required. Verify installers against SHA256SUMS.txt.
Notes
- Local inference uses the CPU. Large images can take several minutes.
- Basemaps, remote elevation data and live OSM features require internet even with a local model.
- Custom models must use the compatible DepthWizard V5 contract. Keep the
.onnx.jsonmetadata and any external weight files alongside the.onnxgraph when importing. - Unsigned installers can show an operating system security prompt.
- Automatic application updates are not enabled; use the download page for future releases.
Installer size improvements
Compared with desktop-v0.1.4. Sizes are MiB.
| Installer | Before | v1.0.1 | Reduction |
|---|---|---|---|
| DepthWizard-1.0.1-linux-x86_64-with-model.AppImage | 1702.2 | 1546.8 | 9.1% |
| DepthWizard-1.0.1-linux-x86_64-without-model.AppImage | 554.1 | 398.8 | 28.0% |
| DepthWizard-1.0.1-mac-arm64-with-model.dmg | 1547.0 | 1467.2 | 5.2% |
| DepthWizard-1.0.1-mac-arm64-with-model.zip | 1552.6 | 1468.3 | 5.4% |
| DepthWizard-1.0.1-mac-arm64-without-model.dmg | 399.9 | 320.1 | 20.0% |
| DepthWizard-1.0.1-mac-arm64-without-model.zip | 406.7 | 322.4 | 20.7% |
| DepthWizard-1.0.1-mac-x64-with-model.dmg | 1647.0 | 1567.2 | 4.8% |
| DepthWizard-1.0.1-mac-x64-with-model.zip | 1652.5 | 1568.2 | 5.1% |
| DepthWizard-1.0.1-mac-x64-without-model.dmg | 500.0 | 420.2 | 16.0% |
| DepthWizard-1.0.1-mac-x64-without-model.zip | 506.6 | 422.4 | 16.6% |
| DepthWizard-1.0.1-win-x64-with-model.exe | 1512.3 | 1435.7 | 5.1% |
| DepthWizard-1.0.1-win-x64-without-model.exe | 372.9 | 296.3 | 20.5% |
All four platform builds and packaged installer checks passed: https://github.com/AbhayKale332/DepthWizard/actions/runs/37202108940. Source commit: 7fba24dddeb2f6ff6c87f05ecd02e756d9a23078.
DepthWizard 0.1.4
DepthWizard Desktop v0.1.4
Download DepthWizard for Windows, macOS or Linux, with local model weights included or as a smaller application using hosted inference.
Changes
- Added a download page that follows published GitHub releases automatically, with Windows, Linux and Apple Silicon / Intel Mac choices.
- Added with-model installers containing the verified V5 forest ONNX model. Local CPU inference is selected automatically on first launch; no separate model installation or settings step is needed.
- Added without-model installers that use hosted Hugging Face inference by default. An internet connection is required for these predictions, and uploaded images are sent to the hosted service.
- Import a compatible model through Model → Install ONNX model… to switch to local inference automatically. Explicit inference choices are saved across restarts.
- Updated the release workflow to verify the model checksum and test both packaged variants on every supported platform.
Installation
| Platform | Download | Install |
|---|---|---|
| Windows 10/11, x64 | .exe |
Run the installer. |
| macOS, Apple Silicon | arm64 .dmg |
Open the disk image and copy DepthWizard to Applications. |
| macOS, Intel | x64 .dmg |
Open the disk image and copy DepthWizard to Applications. |
| Linux, x64 | .AppImage |
Mark the file executable and run it. |
Choose a filename containing with-model for offline local inference, or without-model for the smaller hosted-inference download. Both include the application runtime; Python and Node.js are not required. Verify installers against SHA256SUMS.txt.
Notes
- Local inference uses the CPU. Large images can take several minutes.
- Basemaps, remote elevation data and live OSM features require internet even with a local model.
- Custom models must use the compatible DepthWizard V5 contract. Keep the
.onnx.jsonmetadata and any external weight files alongside the.onnxgraph when importing. - Unsigned installers can show an operating system security prompt.
- Automatic application updates are not enabled; use the download page for future releases.
v0.1.2-alpha
DepthWizard Desktop v0.1.2
DepthWizard brings single-image height estimation, 3D terrain viewing, analysis, and project exports to a desktop application for Windows, macOS, and Linux.
Features
- Run compatible DepthWizard V5 ONNX models locally on your CPU, without a Hugging Face account.
- View and explore 3D terrain, use analysis tools, and save or export projects.
- Open bundled sample scenes and existing projects without installing a model.
- Install or replace a model through Model → Install ONNX model….
- Keep imported models, settings, and inference results in the operating system's application data directory.
- Use the bundled inference runtime without installing Python or Node.js.
Changes in this release
- Moved Electron application code into the root
Desktop/directory, with shared runtime code and separate macOS, Windows, and Linux packaging configurations. - Updated the release workflow for the new source layout and platform output folders.
- Fixed macOS packaging failures caused by an empty signing certificate setting.
- Enabled Python UTF-8 mode in CI to prevent Windows console encoding failures during model staging.
- Fixed tagged builds failing with
Version not changedwhen the package version already matches the release tag.
Installation
| Platform | Download | Install |
|---|---|---|
| Windows 10/11, x64 | .exe |
Run the installer and follow the setup steps. |
| macOS, Apple Silicon | arm64 .dmg |
Open the disk image and copy DepthWizard to Applications. |
| macOS, Intel | x64 .dmg |
Open the disk image and copy DepthWizard to Applications. |
| Linux, x64 | .AppImage |
Mark the file executable and run it. |
Verify downloaded installers against the release's SHA256SUMS.txt.
Model setup
If model weights are not bundled with your installer, obtain a compatible DepthWizard V5 ONNX export. Keep the .onnx graph, its .onnx.json metadata, and all external weight files together. Extract model bundles before choosing the graph in Model → Install ONNX model….
After the model loads, select Local ONNX model in File → Settings, then use Test connection before running a prediction.
Notes
- Local inference uses the CPU; GPU acceleration is not configured.
- Local predictions, 3D viewing, analysis, and project exports work offline once a compatible model is installed. Online basemaps, remote elevation data, and live OSM lookups require internet access.
- Missing model weights do not prevent opening the application, existing projects, or bundled sample scenes.
- Unsigned installers may trigger Windows SmartScreen or macOS Gatekeeper. Check the release's signing status before distribution.
- Automatic application updates are not enabled; download future installers from GitHub Releases.