Pix4D Desktop is designed to deliver accurate results automatically. However, in some cases, the outputs can be improved with manual intervention.
How to improve the results of step 1. Initial Processing
How to improve the absolute accuracy of the model
How to improve the Point Cloud
How to improve the 3D Textured Mesh
How to improve the Orthoplane
How to improve the DSM
How to improve the Orthomosaic
The outputs of step 1 are the camera internal and external orientation and the automatic tie points. These will be used for all the outputs of the following processing steps 2 and 3. Therefore, the results of step 1 should be improved in case the reconstruction is not optimal:
- One of the ways to verify that the reconstruction is good is to analyze the Quality Report.
- Another indicator of a good reconstruction is the reprojection error of the Automatic Tie Points or the Manual Tie Points on the images around the model.
- If some images are not calibrated, the cause can be that the images are not sharp, the overlap is low or the processing options are not optimal.
How to calibrate more images while processing for dense vegetation areas: How to improve the outputs of dense vegetation areas?.
How to manually calibrate images (time-consuming, manual workflow): How to manually calibrate uncalibrated Cameras in the rayCloud.
- If none of the images is calibrated and processing has failed, most probably the dataset is not optimal or the settings selected for processing are not correct.
For more information: Error e0046: "Processing failed.".
- If some images are blurry or badly illuminated or the take-off/landing images are included in the project, removing these images from processing may improve the results.
How to disable images: How to remove Images from a Project.
In order to improve the absolute accuracy of the model, Ground Control Points (GCPs) are needed. It is important to distribute the GCPs both horizontally and vertically all around the model to achieve a high accuracy globally.
For more information about GCPs: Using GCPs.
If the reconstruction (step 1) is not optimal, the point cloud may be noisy. Make sure that the results of step 1 are optimal before generating the point cloud. However, if the point cloud is still noisy, there are some actions to be taken:
- If the borders of the point cloud are less dense or they do not correspond to the area of interest, it is possible to restrict the generated point cloud by drawing a processing area.
- If the project is oblique and sky is also captured, the sky may be reconstructed around the model. To remove the sky it is possible to annotate the original images.
How to annotate the images to remove the sky: How to remove the Sky from the Point Cloud using the Annotation Tool .
- If the point cloud is noisy and there are some points to be removed, it is possible to edit the point cloud and assign the points to the Deleted point group. These points will not be exported when the point cloud is exported from the rayCloud. They will also not be used for all the output generated after the editing of the point cloud (e.g. DSM, Orthomosaic, 3D Textured Mesh, etc).
The Mesh receives as input the point cloud. If the point cloud is noisy, the mesh will also be noisy. Make sure that the results of step 1 and the point cloud are optimal before generating the 3D Textured Mesh. However, if there are planar surfaces that are distorted or there are areas with holes, draw a surface to improve the results.
How to improve the 3D Textured Mesh: How to improve the 3D textured mesh.
The Orthoplane receives as input the point cloud. If the point cloud is noisy, the Orthoplane may also be noisy. However, there might be other causes that can introduce noise in the Orthoplane:
- The orthoplane box is not perpendicular to the facade and the final output is not an ortho projection.
For step-by-step instructions on how to align the orthoplane to the facade using a surface object: How to draw a new orthoplane.
- The background noise is projected on the orthoplane.
For step-by-step instructions on how to remove the background: How to remove the Background of the Orthoplane using the Annotation Tool .
The DSM receives as input the point cloud. If the point cloud is noisy, the DSM may also be noisy. Make sure that the results of step 1 and the point cloud are optimal before generating the DSM. However, if the DSM is still noisy, there are some actions to be taken:
- If the edges of the building are noisy and the DSM is not sharp, draw a surface on the roof plane. All the points above and below the surface will not be used for the DSM generation.
How to improve the DSM drawing surfaces in the rayCloud: How to correct Building Artifacts in the DSM and Orthomosaic with the rayCloud.
- If the vegetation has artifacts and distortions, use the Triangulation method for the DSM generation.
How to improve the DSM for dense vegetation areas: How to improve the outputs of dense vegetation areas?.
The orthomosaic receives as input the DSM and the position of the original image. If the DSM is noisy or the reconstruction is not optimal, the orthomosaic will be distorted. Make sure that the results of step1 and the DSM are optimal before you generate the Orthomosaic.If the Orthomosaic is still noisy, it is possible to edit it in the Mosaid Editor:
- If there are moving cars in the scene, they will appear with transparency in many positions of the orthomosaic. This effect is called ghost effect. To improve the orthomosaic, it is possible to edit it in the Mosaic Editor, draw a region around the car and select an image in which the car does not appear.
How to edit the Orthomosaic in the Mosaic Editor: How to improve the Visual Aspect of Buildings in the Orthomosaic with the Mosaic Editor.
- If buildings or dense vegetation areas are distorted, it is possible to improve the visual aspect of the orthomosaic by selecting the planar projection for the distorted areas.
How to improve the visual aspect of buildings in the Mosaic Editor: How to improve the Visual Aspect of Buildings in the Orthomosaic with the Mosaic Editor.