Advanced Parameters
You can set advanced parameters in the SDK JSON configuration file for more flexible control.
Aerial Triangulation Module
| Parameter | Type | Default | Description |
|---|---|---|---|
num_features | int | 20000 | Number of feature points extracted per image |
grid_count | int | 200 | Pixel grid count for feature sampling; larger values produce more AT sparse points |
Reconstruction Module
| Parameter | Type | Default | Description |
|---|---|---|---|
min_avali_memory_size | double | 0 | Minimum available memory for the device; controls tile size and count for tiled reconstruction. 0 uses maximum available memory |
fill_water_area_with_AI | bool | false | Use machine learning for water-area inpainting [requires ML plugin] |
build_overview | bool | true | Whether to build image pyramids for 2D deliverables |
cut_frame_2d | bool | false | Whether to output 2D deliverables in tiles |
cut_frame_width | int | 5000 | Pixel width per 2D tile when cut_frame_2d is true |
dom_gsd | double | 0 | Ground resolution of true orthophoto; 0 means auto-compute. SDK applies a minimum guard; values below the limit are ignored |
dom_gsd_min_scale_to_original | double | 0.25 | Minimum guard: dom_gsd will not exceed original image GSD × this factor |
tile_extention | string | ".webp" | Image format for true orthophoto tiles: .png or .webp |
use_gcj02 | bool | false | Convert orthophoto tiles to GCJ-02 coordinate system |
fast_mode | bool | false | Faster processing at the cost of some mesh detail |
keep_undistort_images | bool | false | Keep undistorted images after reconstruction (default: do not keep) |
remove_small_islands | bool | false | Remove isolated floating components from the 3D model |
mesh_decimate_ratio | double | 1.0 | Mesh simplification ratio; smaller values mean more simplification |
use_draco_compression | bool | true | Enable Draco vertex compression for 3D Tiles |
tex_image_type_3dtiles | int | 0 | 3D Tiles texture format: 0=JPEG, 1=WebP |
roi_for_2d | roi | - | Region of interest for 2D reconstruction. See ROI structure |
roi_for_3d | roi | - | Region of interest for 3D reconstruction. See ROI structure |
roi_coordinate_system | JSON | WGS84 | Coordinate system for ROI. See coordinate system |
Important
On Windows, enabling fill_water_area_with_AI requires the machine learning plugin: ML plugin. Extract all DLL files into the same directory as reconstruct_full_engine.
Parameter Structures
roi
An ROI defines a spatial extent with a 2D boundary, min_z, and max_z.
| Field | Type | Description |
|---|---|---|
boundary | array | 2D polygon vertices of the ROI boundary |
min_z | double | Minimum elevation of the ROI |
max_z | double | Maximum elevation of the ROI |
Example:
json
{
"boundary": [
[x1, y1],
[x2, y2],
[x3, y3]
],
"min_z": 100.0,
"max_z": 500.0
}