Project Layer creation
3D structure creation¶
The 3D structure creation is a process when a sparse point cloud layer and camera positions are created from the photo set.¶
To begin creating layers and working with Pixpro offline we need to create a project first:
- Click on the Local Computing button in the home screen.
- In the New Project step window appears:
Here project name and location on the computer can be specified. More importantly the coordinate system of the created project must be selected. There are a few options in the dropdown menu:
Detect automatically - the software will automatically detect the appropriate coordinate system using the GPS information from input images. This will usually result in an UTM coordinate system.
Arbitrary - the project will remain in an arbitrary coordinate system for use without any initial GPS information.
Exact coordinate system - further down the list one can select the exact coordinate system to use in the project. Recommended for advanced users with clear referencing requirements.
After clicking NEXT, the Photo Set Manager window opens. Photo set manager can be used to add, remove, inspect photos that have been added. More advanced users can also inspect camera calibration data in this window. For more detailed information see Photo set manager page.
Click Add Photo or Add Folder and select the photos made for the 3D reconstruction.
When the photos are listed, click OK.
To execute the first step of creating a 3D model go to the 3D structure in the Workflow tab. The 3D Reconstruction Parameters dialogue window opens.
Advanced parameters can be expanded and then edited as necessary. If you do not want to delve into advanced settings deeper, just press start, default settings usually work well. The processing time will depend on photo count and resolution. After the process is complete, you should get a sparse point cloud and camera positions in the project.
3D Reconstruction parameters:¶
Main setting in this section is the Processing Speed dropdown.
Processing speed setting affects the reconstruction speed and quality.
Fast processing takes the least amount of time but it also requires the best possible image quality. This means that the 'Fast' setting is best for those who are confident that the photos taken for the reconstruction are more than suitable. This includes the overlap, general image quality and object coverage with room to spare. If these aspects are fulfilled, Fast processing setting can be used while saving time.
Medium setting, the default, is usually a good choice for most cases.
Slow setting allows using poorer quality pictures and provides more coverage while taking considerably more time to finish. If the Slow reconstruction setting combined with optimize cameras setting fail - there is a high probability that the pictures were taken inadequately for the 3D reconstruction.
Advanced properties button opens a new dialog with more reconstruction settings:
Camera Model: shows the camera model name.
Intrinsics source: shows where the camera intrinsic values came from.
Intrinsics may come from three different sources:
- Estimated - Intrinsics have been estimated from the image exif data. This means that no true intrinsic values have been found, optimization will be obligatory.
- Calibration DB - Intrinsic values have been found in the database for this camera. Only Final BA optimization method can be used as described above.
- DJI XMP data - Intrinsic values have been found in image file metadata. This applies to more advanced DJI RTK drones that save calibrated intrinsic values in the image metadata. Only Final BA optimization method can be used as described above. Camera Parameters section:
Image size: shows the image size in pixels.
Camera parameters section
In this section intrinsic camera parameters can be entered and inspected. Focal length, principal point and other values can be entered in pixels.
BA or Bundle Adjustment process settings can be changed as well. Local BA is a bundle adjustment process that is performed throughout the reconstruction process, while Final BA is a bundle adjustment process that will be performed only in the last step of the reconstruction. If the intrinsic values are present in the calibration database only Final BA will be used for the reconstruction. This usually yields the best measurement results. If the intrinsic values are not present in the calibration database, Local BA is used, and only the principal point is fixed while all other intrinsic values are optimized throughout the process. In this case measurements can be less reliable and less consistent.
If by any chance user will manually enter extreme intrinsic values, a warning will come up when trying to start reconstruction. If that is the case, just double check if the entered values are correct.
Camera intrisics are also managed through photo set manager, but here one can inspect and adjust those values right before reconstruction. This section is for advanced users looking to experiment. For more detailed information on camera intrinsics management and reuse see Photo set manager page.
Clear Temporary Reconstruction Results:
This section is dedicated to a few checkmarks that will appear if previous reconstructions have already been made in this project. If a reconstruction was made previously, there will be temporary data saved to potentially reduce time needed for a repeat reconstruction. By default all checkmarks that can be - are selected and data will not be used to speed up the repeat reconstruction. It is generally a best practice to have everything checked and in turn redo all the steps of the reconstruction, but if the user has experience and processing time must be as short as possible, one can choose to skip clearing old data and reuse it for the repeat reconstruction.
EXIF prior values
In this section we can choose whether the prior position and rotation values that come with the photo metadata are used. These options should be used if the said values are known to be precise and correct, I.E. using an RTK drone, or PPK enabled images.
Either Incremental or Hybrid mapper can be chosen. The default is the incremental mapper, hybrid mapper should be chosen if there are problems with reconstructing large amounts of photos (500 or more). Hybrid mapper will change how the 3D reconstruction begins which might result in better results in some cases.
Note 1: rerunning the reconstruction in the same project will result in all layers being deleted.
Note 2: The reconstruction can be done from only a part of the whole photo set. There are two possible ways (both can be used simultaneously) to select a bundle of photos for reconstruction:
- After uploading a photo set, select the photos from the list on the Layers Panel by checking / unchecking the relevant photos. Then click Start 3D Reconstruction. The reconstruction will be performed only from the checked photos.
- After uploading photo set, select the photos on the map by drawing the polygon around the cameras. To do this, right-click the Photos on the Layers Panel. Click Select workspace area, and start marking around the needed area on the map. When finished marking, right-click to close the perimeter of the selected area. The photos within only the selected area will be used for reconstruction. Please note that the workspace selection option can be done only when the photos come with the GPS data in order to display them on the map.
After the reconstruction is finished the Sparse point cloud will be displayed on the screen with camera positions in blue. Cameras that were not included in the reconstruction will not be displayed in the 3D view, and will be marked as red in the layers panel.
Remember that sparse point cloud is only the first stage of creating a workable model. In the sparse point cloud, as the name suggests, there are only few points in comparison to the dense point cloud, so the object itself can be hard to discern at this stage.
After sparse point cloud is created it can be Reoptimized. Reoptimization is an additional process that should be ran if GCP's were added to the project.
Dense Point Cloud Creation¶
Dense point cloud is the second stage of processing, and it is required to start measuring and creating digital elevation maps and orthophotos down the line. Dense point cloud creation is also the lengthiest step.
Point Cloud Density - High or Ultra High density point clouds will provide more points and better GSD values later as a consequence, but it will take longer to create and will require more video memory (especially in the case of Ultra High density). Low or Light density point clouds will take less time to process while providing less points. Ultra Light density is reserved for getting a result as fast as possible, without any focus on quality. Medium density is a happy medium.
Image Size - Shows the resolution of the depth maps that will be generated. The depth map size depends on the input image size and the Point Cloud Density setting.
Depth maps filtering - Mild option will reduce the number of points while attempting to maintain the overall 3D structure of the scene. This process may require a considerable amount of RAM.
When the generation is finished, the dense point cloud layer appears in the 3D view.
There is an alternative way to create a Dense point cloud by cloning the Sparse cloud. To achieve this press right mouse button on the Sparse point cloud layer in the layers panel and select Clone as Dense Point Cloud. Then a small dialog will appear asking whether the point projections should be taken from two or three pictures. Two means that more points will be cloned, but there may be more unreliable points. Press OK and the cloned Dense cloud will be available to use for creating a DEM and Orthophoto. This workflow is only recommended when Orthophoto is needed urgently.
Any existing point cloud can be duplicated by pressing right mouse button on the Dense point cloud layer and selecting Duplicate point cloud. A dialog will appear and will suggest duplicating the tree of layers associated with the point cloud.
With a Dense point cloud generated one can easily see the scanned object and proceed with creating other layers. If the vizualization of the point cloud seems unsatisfactory, it can be adjusted in the Dense point cloud properties panel. Point size slider, Point type and Vizualization quality slider all affect pointcloud vizualization. These can be adjusted to taste.
Dense Point Cloud layer is the stage where Bounding Box usage is often highly recommended. A bb box is a cube shaped 3D area that can be adjusted by dragging the cube face to crop out dense point cloud points. This can be useful if the point cloud is unnecessarily large for the intended object. Using the bounding box may result in better looking models, shorter processing duration in further stages and overall tidier results.
To use the Bounding Box press the Bounding Box button in the toolbar . Bounding Box can be reset in the Dense point cloud properties panel. Below an example of Dense point cloud with and without Bounding Box.
A discrete Dense point can be made using only the points bounded by the bounding box. After using the bounding box as described, right mouse click on the current Dense point cloud and press Cut to bounding box.
This will not edit the current point cloud destructively. A new dense point cloud will be created using the designated bounding box and copied as an additional layer. This new Dense point cloud layer is identical to any dense point cloud in every aspect, but it will only contain the points previously bounded by the bounding box. The layer can be used for further data processing and can be exported.
If any problems with Dense point cloud visualisation occur Recreate Chunks function can be used to refresh the pointcloud visualisation. Press right mouse button on the point cloud in the layers panel and choose Recreate Chunks.
Sample dense point cloud¶
Mesh and Texture Creation¶
Mesh is a 3D surface created from the point cloud and textured using raster data from photos. To generate a mesh go to Workflow and choose mesh. In Generate Mesh window a few mesh generation parameters can be selected.
- Point cloud - allows selecting the point cloud used for mesh generation.
- Mesh Type - 3D and 2.5D meshes can be created. 2.5D mesh represents raster like digital elevation map surface stored in a mesh file format.
- Size Limit - Mesh can be Limited or Unlimited by the face count.
- Maximum Triangles - allows selecting the maximum amount of triangles to be generated in the mesh if the mesh size limit is set to Limited.
- Hole Filling - Select the type of hole filling. None will not fill any holes in the mesh. Quality preset attempts to fill holes while preserving overall structure. Aggresive option will try to fill even the largest holes.
Use Clipping Box - toggle the use of clipping box for the generated mesh. If the bounding box is not used, the mesh will be generated using the entire dense point cloud.
Generate texture - texture generation can be toggled. Poperties button nearby will bring up texture generation settings:
- Mesh - choose which mesh to texture. By default laste generated mesh is selected.
- Texture Size - size of a single atlas.
- Texture Atlas Limit - atlas limit can be chosen. Auto will determine the optimum amount of atlases.
Select photos button allows selecting which photos will be used for texturing.
Press start to begin mesh and texture generation.
Sample Mesh and Textured mesh¶
Digital Elevation Map Creation¶
DEM or Digital Elevation Map is a 2.5D representation of the scanned area and it is created from the dense point cloud. It can be shaded with various options and used for all measurements (volume cannot be measured on any other layers).
Before creating the DEM it is recommended to inspect and use the bounding box on the Dense point cloud. Bounding box is generated automatically after creating each dense point cloud and it can be changed or reset altogether at any time.
To generate the digital elevation map (DEM), go to Workflow in the top menu. Click Digital Elevation Map.
DEM generation dialog appears. Here you can see DEM parameters section and choose a few settings:
Point cloud: choose which dense point cloud should be used for the DEM creation.
Points Count: shows the amount of points in the selected dense point cloud.
Interpolation - standard height rasterization algorithm, will create an uniform triangle raster height map from the dense point cloud directly.
Triangulation - this option triangulates the surface between all point cloud points (as if creating a mesh or a TIN) and then the surface is rasterized to use as a Digital Elevation Map. This option is recommended in a few cases: firstly if the point cloud has extremely varied point density and secondly when the point cloud is imported from a CAD software and not generated using photogrammetry. When using this option always use a small GSD value to get the best results (for example 0.01 meter).
GSD: ground sampling distance value. This value is initially suggested according to the current Dense point cloud and the bounding box (if the bounding box is used it does affect GSD calculation). The smaller the value the more detailed the DEM will be. Entering a smaller value should be done with care because that might result in holes where the dense point cloud is not dense enough. Holes can be filled, but excessive hole filling distorts volume data.
Smoothing: smoothing value represents how much smoothing is applied when generating the DEM. If none is selected from the dropdown menu - the DEM will have no smoothing applied, if a value of 3 is selected, heavy smoothing will be applied to the surface. Smoothing will result in a more smoother and neater looking surface, but applying too much smoothing can reduce overall volume accuracy and smooth out vertical structures. Default value of 2 is a good compromise.
Hole filling: there are 4 options here:
Internal: Fills holes that have limited area. If the hole is enclosed and within the main scan area it is considered internal. All holes that are open to the peripheral area will remain as such.
External: Only non enclosed holes will be filled which will result in better looking DEM edges. Every hole that has a definite area will be left as such. Note: this process is much more demanding than internal only filling so it may take much longer to complete. Using bounding box with this option is highly recommended.
Internal and External: All holes will be filled while attempting to create a non-infinite project area. Using bounding box with this option is highly recommended.
Smoothing: smoothing radius value represents how much smoothing is applied when generating the DEM. If none is selected from the dropdown menu - the DEM will have no smoothing applied, if Radius 6 is selected, heavy smoothing will be applied to the surface. Smoothing will result in a more smoother and neater looking surface, but applying too much smoothing can reduce overall volume accuracy and smooth out vertical structures. Default value of Radius 3 is a good compromise.
Use bounding box: if checked the DEM will be created using only the portion of dense point cloud which is within the set bounding box.
After all parameters are selected press start and the DEM will be generated. This process is usually a lot faster than 3D structure or dense point cloud steps.
Note: the project should be georeferenced either to the photos location or the ground control points. If it was not georeferenced before, go to Georeferencing in the top menu. Select available or preferred way of georeferencing from the menu.
When the generation is finished, the layer appears. NOTE: If other layers are enabled simultaneously with the DEM, all graphic object drawing will be done on the uppermost layer. It is always recommended to have only one layer enabled when adding new objects.
DEM shading can be changed in the Properties panel. A Legend should be visible that allows changing the visual shading appearance:
Vertical slider is used to set the low and high points on the histogram. Adjust the vertical slider until desired visualization is achieved.
Interval value represents the thickness of a single color interval. The smaller the value the smoother the DEM will appear.
Colors value defines the amount of colors desired on the DEM. The thickness of a single color interval will be calculated accordingly.
If any problems with DEM vizualization occur Recreate Chunks function can be used to refresh the Digital elevation map visualisation. Press right mouse button on the Digital elevation map in the layers panel and choose Recreate Chunks. This will also refresh the Orthophoto shading.
After DEM layer is created, a Dense point cloud can be recreated using the DEM surface vertices only. To create a Dense point cloud from a DEM, right mouse click any DEM layer and choose Create Dense Point Cloud. This will create a dense cloud that contains the DEM surface as points. The points will not have any color information and will be arranged in a single layer, identically to the vertices of a 2.5D (height map) type DEM surface. All edits made to the DEM surface will be represented. This point cloud can be exported using the standard point cloud export workflow.
Sample digital elevation map shaded with a gradient¶
Orthophoto is an orthorectified image representing the scanned area. DEM is needed to make a true orthophoto.
To generate the orthophoto, go to Workflow in the top menu. Click Orthophoto.
In the Orthophoto dialogue window, select the digital elevation map for the basis. Only currently georeferenced DEM's can be used for orhtophoto generation (currently georeferenced DEM's are marked as blue in the layers panel). Below the selection dropdown menu selected DEMs' GSD is displayed.
Further only two parameters are present: GSD and Color Correction checkmark. The smaller the GSD value is, the higher resolution orthophoto is created. Suggested value is 4 times smaller than the used digital elevation map. The value can be even smaller but expect longer processing times for high-resolution imagery. Color correction is recommended if the lighting conditions were changing during the scan, or the scan is combined from photos taken at different times. Color correction further increases the processing duration.
- Click Start to create the orthophoto. When the generation is finished, the layer will appear in the Map tab and the layer will be available to use as an orthophoto shading option in the DEM properties panel.
Contour Lines creation¶
Contour lines represent the terrain using the classic isohypse lines.
- To generate the contour lines, go to Workflow in the top menu. Click Contour Lines.
- In the Contour Lines dialogue window, select the digital elevation map for the basis. Set the lines interval value and offset value if needed. The smaller the interval value the denser the contour lines will be. The default is 1 meter. Height Offset just offsets the lines up or down by a defined amount.
- Click Start.
- As soon as the generation finishes, the contour lines can be viewed in the map and in the 3D view.
If the contour line generation failed with a message stating the digital elevation map is too hollow inspect couple of things:
See if the DEM is visually solid and without holes. If the DEM has many holes recreate it with hole filling or a smaller GSD value.
Crop the DEM using DEM crop tool to eliminate irregular edges, which might be causing the generation failure. After cropping the DEM, save the project for changes to take effect.
Sample contour lines¶
Early prototype of an automatic breakline generation is available. Breakline algorithm draws breaklines using a selected DEM surface.
To generate breaklines go to Workflow and choose Breaklines. There are three settings that can be chosen:
Digital Elevation Map - choose the digital elevation map which is to be used for breakline generation. Important note: DEM must be of 0.5 meter GSD or more. Digital elevation maps with a smaller GSD value will not work.
Radius - radius value that represents the sensitivity of the algorithm. The smaller the value, the more sensitive to surface height changes will the algorithm be. The larger the value - the less breaklines will be generated.
Line length - line length filtering value. The smaller the value the lower the threshold for breakline filtration will be. If a small value is selected - there will be more of short, unconnected breaklines.
After selecting these settings press start and the generation will begin. After generation is complete breaklines will be displayed. They can be viewed as a single visible layer, or combined with any other available layer.
When a breakline layer is selected, alternating line colors can be chosen in the properties panel on the right side of the screen. Default colors are blue and red.
Keep in mind that current functionality is in early prototype stage.
Full workflow is a single function that lets you generate all the layers sequentially without user presence. To launch full workflow window press button in the toolbar or F10 key on the keyboard. Full workflow dialog appears:
Check whichever layers you would like to generate. CLick Advanced Properties which are identical as described in this chapter for each layer accordingly except for Use the last generated (item) checkmark. Use this checkmark in the case of new layers being generated on top of existing layers. For example, if there is a dense point cloud already present in the project, you can generate a new one and use the newly generated point cloud for the DEM creation. In this case enable the Use last generated (item) setting.
After choosing all the settings and being sure that they best fit your needs you can save them as a preset, and then load these settings so there would be no need to enter them manually again. To use this function use Save as Preset and Load Preset buttons accordingly.
To save time even further enable the Load Generated Layers checkmark so after the generation, all layers will also be loaded without any user input eliminating any waiting times before doing measurements or other work.
With Pixpro you can process your projects on our servers. In the home screen of the software, choose Cloud computing:
Then enter the name of your project and choose the coordinate system.
Upload the images to our servers by pressing Upload. The Photo Manager will open. Here you can choose photos from your computer, inspect them. When ready - press ok.
Additionally you can add ground control points if your project requires. Once the photos and GCPs (optional) are added, click Next.
Upload and processing will begin, here you can track the progress. Hover on these icons with your mouse to get additional information. You can click Close, to close the software while our servers are processing. The current processing tasks will appear in the current cloud tasks list.
Once the processing on our servers is complete, you can download the entire project to your computer and start working on it.
Click download, choose the location for the project and the project. After download is complete you can click Open. The project now will become local and open.
There are no limitations on working with downloaded projects, they behave identically to those created locally. The downloaded project will appear among Recent projects list.