Coordinate System Preservation with XMPs

This tutorial shows how to preserve a coordinate system while processing many datasets with reuse of a fixed-camera setup before a computation of camera parameters.

In the tutorial, a workflow with XMP metadata is used, which is suitable for processing of datasets obtained using a completely fixed-camera setup, i.e. fixed-cameras' position, orientation, focal length and everything else. Camera movement of just 0.001 millimeters or camera rotation of 0.00001 degrees is visible in an image. Therefore the camera setup has to be fixed really firmly. In such case, the application uses the XMP metadata (exported as locked) in the alignment process and computes (really) the same alignment as was the one from which the XMP metadata was exported.

The XMP workflow is used to process scans in the user interface (UI) or in the command-line interface (CLI) when preservation of model position and distances in scans is required and usage of exact camera positions is needed.

On this page can be found step-by-step videos with the XMP workflow and also the links to the datasets used, metadata, scripts and final models which are presented in the tutorial.

Step-by-Step Video Tutorial

Creating a model step by step: add images, align the images, set ground control points, export XMP metadata, set and export a reconstruction region, create a model, simplify the model, replace the source images, texture the model, export the model.


Creating a model using the XMP workflow: add images with XMP metadata, align the images, import a reconstruction region, create a model, simplify the model, replace the source images, texture the model, export the model.


Model comparison: a comparison of the reconstructed model with the one computed using the XMP workflow.


Command-line processing of 4 datasets using the XMP workflow: an illustration of command-line processing of many datasets being reused before a computation of camera parameters.

Data Used and Models Computed

Datasets, metadata and scripts used in this tutorial and also the final models computed in this tutorial may be downloaded here:
Datasets: Set2-a (click to download), Set2-b (click to download), Set2-c (click to download), Set-logo (click to download)
Metadata: click to download
Scripts: click to download
Models: Set2-a (click to download), Set2-b (click to download), Set2-c (click to download), Set-logo (click to download)

Epic Games Slovakia s.r.o. would like to thank the company PI3DSCAN (www.pi3dscan.com) for the opportunity to use their datasets Set2-a, Set2-b, Set2-c and Set-logo in this tutorial and also for the permission to distribute them.

www.3dmij.nl

Full-body Scans and Coordinate System Preservation

How to use RealityCapture with different full-body scanners

Continue

Watch step by step how to preserve a coordinate system reusing camera poses

See also: