Parameter

Description

Tiles

Tile Split Method

Select either of the following options:

§  Fixed 2D grid – The reconstruction area is divided into uniformly-sized tiles.

§  Adaptive 3D boxes – The reconstruction area is divided into 3D tiles in an adaptive manner, based on the memory requirements of each section of the reconstruction area. For example, sections of high resolution or close range, or areas with greater photo coverage, will be split into smaller tiles. The reconstruction area is adaptively divided along the horizontal, vertical and depth axes, allowing for complex three-dimensional data such as tall buildings or utility poles to be split into multiple tiles. When reconstruction tiles are created as adaptive 3D boxes, the display style option is automatically set to"3D Dynamic". In this display mode, the tiles and their boxes are displayed in absolute heights. Tiles that were created one on top of each other will appear as if they are floating in this mode. See "Setting a Reconstruction Tile’s Display Style" in the "Reviewing Your Build" chapter for more information.

Reconstruction Tile Size

Size of the reconstruction tiles. This can be set to Auto or to a specified size in meters. It is generally recommended to use the Auto option. When tile division results were non-optimal, setting a size in meters is recommended.

Note:    The size of the reconstruction tiles can only be set when using the 2D grid tile split method.

If you want to control tile size based on size in meters, three factors should be taken into account: RAM, resolution and overlap. For projects with standard resolution (5-10 cm photos) 100-150 meters is recommended on computers with 16 GB of RAM. Larger reconstruction tiles require more memory. On computers with over 64 GB of RAM, a tile size up to 400 meters can used be used. As a general rule, the reconstruction tile size should be the ground resolution of the data multiplied by 2000. For high overlap datasets, the ratio between the reconstruction tile size and the ground resolution should be lower.

Note:    If point cloud build steps are repeatedly failing, and you observe a very high RAM usage towards the end of the point cloud step, it is recommended to reduce memory requirements or reconstruction tile size.

Quality & Performance

Photo sampling interval

The resolution of the photos at which 3D points are extracted for the point cloud.

Mesh model resolution

Level of mesh complexity. The Low option will aggressively simplify the mesh.

Texture resolution

The resolution from the photos to use for texturing the model: Best will texture in the exact quality as the photos, while Medium and Low texture with downsampled textures for meshes with smaller textures.

Bands & Collections

Use for point cloud and model

Click the More Options  button to select the collections to include in the point cloud. See "Setting the Collections and Bands to Include in the Build Outputs" in this chapter for information.

Use for texture

Click the More Options  button to select the bands to assign to each RGB of the mesh model texture, and the collections to include in generating the texture. See "Setting the Collections and Bands to Include in the Build Outputs" in this chapter for information.

Point Cloud

Min. number of matches

The minimum number of matches required to generate each 3D point. When the point cloud generated is sparse with little overlap of photos, consider reducing the minimum number of matches. When there is noise in the point cloud output, it is recommended to increase the minimum number of matches, providing there is high overlap of the photos.

Correlation window size

Size in pixels of the window used to perform the correlation between photos and extraction of 3D points from them.

LiDAR

Merge LiDAR Points

The data to use in creating the point cloud: only LiDAR points, only points generated from photos, or Auto – supplement LiDAR with points generated from photos where LiDAR information is insufficient.

LiDAR collection mode

The direction of LiDAR collection assists in calculating surfaces from the point cloud. When there is no trajectory file and the LiDAR collection is known to be airborne, select Aerial. Otherwise, select Auto.

Use LiDAR colors

The data to use in texturing the model:

§  Auto – Supplement photo texture with color from LiDAR points where photo texturing information is insufficient.

§  Only LiDAR colors – Only color from LiDAR points.

§  Only photos – Only photo texture.

Model

Smooth surfaces

Determines whether surfaces are flattened.

Enhance edges

Determines whether 2D edges in the project’s photos are correlated and 3D edges extracted from them, to produce the 3D model. This setting produces better-defined 3D models with sharper edges, which is important for urban area reconstructions and thin structures such as poles and wires.

Fill in ground

Fills in ground gaps in areas with low coverage by incorporating calculated ground surface. Recommended to improve build results in low overlap, nadir projects.

Fill in walls

When PhotoMesh detects a vertical gap, it automatically fills in the missing sections to form a complete and continuous vertical surface. This option is only available when "Fill in Ground" is set to Yes.

Smooth walls

Smooths out vertical surfaces.

Remove floating components

Level of removal of small floating components in the mesh model (e.g. noise, electric lines, lamp poles).

Tile vertical clipping buffer

Vertical buffer to apply to the tile. Adding a buffer with a positive value is useful when there are objects that were cut off in a specific tile. A buffer with a negative value can be used to crop areas with small floating components.

Texture

Color saturation

Vividness of the colors of the textured model.

Texture blending

Determines whether to use data blended from multiple photos in texturing the model, or only data from the optimal photo. Select No if blending results in ghosting artifacts. Select Extra if Yes produces color patches in areas with strong color variance.

Sharpening

Level of image sharpening applied to the final result.

Advanced

Reconstruction flags

Flags that specify alternate behaviors and settings that PhotoMesh should apply. These flags can be obtained from Skyline support.