Improve the accuracy of your photo project by specifying common points in two images.
You have the option to refine your project by defining matching points to "stitch" two images at a time. The term "stitching" is specific to ReCap 360's photogrammetry projects and is essentially synonymous with registration. Like registration, ReCap 360's stitching process entails the user manually specifying target points within the interface in order to refine stitching across 2 images at a time (compare to Manual Registration for scans). This feature conjoins with the automatic Photo-to-3D service and allows user input for correction (when some images were not stitched correctly) or general refinement.
In this part of photo project workflow, you will be using Survey Settings to define target points. Begin by choosing a "left" and a "right" photo from your collection that share at least 3 points that can be seen across both images. You will then choose a target point in either picture, and match the same point as closely as possible in the other picture. X, Y, and Z coordinates in meters are required for each point
Use the mouse scroll to zoom and choose accurate points. You should see a small circle with a number matching your selection in both images. When a point is selected in both images, a yellow outline around half of the circumference of the selection circle will appear. The example below shows the same point selected in both images:
To define multiple points in one image before matching them in the other, press Ctrl to change this number and create a point. This is per user preference and will not affect the reconstruction outcome. Be aware, however, that if you have a point defined on only one image, a pink outline around one quarter of the perimeter will appear, signifying that the point will not be included as a survey point unless it is matched in the other image. You can define multiple targets for accuracy, but none are required. While there are no firm or universal recommendations, ReCap 360 Pro typically relies on vantage points and quality, and requires at least 3 observations of a feature in order for the images to be reconstructed into meshes, rasters, or point clouds. Targets can be added before submitting a project or after the project has been submitted. If it's after, the project will have to be resubmitted for the changes to take effect.