Meshroom is a photogrammetry software application that makes working with complex objects in digital environments much simpler. By taking multiple photographs and reconstructing them as 3D models, Meshroom makes digital work much more manageable.
Meshroom makes use of images by simply dragging them into its Images pane on the left-side, where it will analyze their metadata before starting work on reconstruction.
Photogrammetry
Photogrammetry is an efficient technology used to create three dimensional models from photographs taken from various angles of an object or place. Architects and surveyors alike rely on photogrammetry to quickly scan large areas of land.
Photogrammetric software is used to detect overlapping pixels in photographs and stitch them together into 3D models of objects, or it can also be used for making measurements and calculating dimensions. Although there are various forms of photogrammetry used in 3D scanning processes, one that stands out is multi-image photogrammetry.
Photogrammetry can be applied in many different areas, from architectural modeling and land surveying to forensics. One advantage of this form of photogrammetry is that it produces high-quality results without complex and costly equipment; its only drawback being it cannot scan objects with flat surfaces such as carpeting. Also, many photographs must be taken to produce an accurate model.
To get started with Meshroom Photogrammetry, start by uploading your source images to a folder on your computer, opening it, dragging them out into Meshroom software, then clicking the green start button at the top. From there you can monitor its progress via its log tab in the lower right corner.
Once the software has finished running, you will be presented with a list of images it has used – acceptable ones will have checkmarks next to them while any images it rejected will be marked by an X. From here, you can go back through and examine each individual image or examine the final model itself.
Meshroom also provides the Live Reconstruction option, which will gradually build a model as you add more pictures to the source folder. This feature is particularly helpful if you don’t know exactly how many are necessary for producing a quality model.
Augmented Reconstruction
Photogrammetry’s goal is to reconstruct a scene’s geometry from an unordered collection of photographs or videos taken without order, using “Structure-from-Motion” or multi view stereoscopy techniques. Once this data has been processed and organized into 3D model forms it can then be utilized for 3D modeling, augmented reality virtual reality applications and many others.
Meshroom offers two methods for performing SfM: global and hierarchical approaches. While global may be faster, its results may be less precise. Both approaches can be configured via checkboxes in the Configuration tab.
Selecting between global and hierarchical SfM is critical as its selection can have a substantial effect on reconstruction performance. For instance, choosing an appropriate calibration metric could drastically change its accuracy.
In order to accelerate computation times, SfM processes can be parallelized using options in the Configuration tab. For instance, an intermediate file format such as Alembic allows them to run faster on render farms and therefore reduces overall computation times. Furthermore, parallel thread counts can be adjusted as necessary in order to increase computational efficiency.
Users of SfM configuration tab can also select the minimum and maximum numbers of images to monitor per step and track matching sessions respectively, along with whether to enable Rig Constraint for automatic locking of intrinsic camera parameters such as focal length and distortion.
After setting up SfM, reconstruction can commence. The Live Reconstruction pane provides updates of its progress while calculations take place, and when complete the resulting 3D model can be seen in its own window. You may export the model in various formats like OBJ or PLY for exporting later. Meshroom also generates textures for this mesh and offers an option to reduce quality while improving performance for large projects that contain many images.
3D Modeling
Photogrammetry is an incredible powerful tool for 3D modeling objects. This technique involves taking multiple photographs of an object and stitching them together into one high-resolution mesh that can then be rendered or animated using other programs such as Meshroom (which has quickly become popular within photogrammetry world). In this guide we’ll show how Meshroom can help make this process even simpler!
Start by dragging your images into the Images pane on the left and allow the software to start processing them. Within several steps, it will have assessed all photos and marked those which meet its criteria as acceptable before beginning mapping and meshing operations, which could take hours or days depending on their complexity.
Once meshing is complete, all results can be seen in the 3D Viewer. In addition, double-clicking any file will give access to their contents; then import them directly into your favorite 3D modeling application.
Meshroom uses AliceVision photogrammetric computer vision framework to perform 3D reconstruction and other computational photography tasks, with its node-based design that allows you to customize individual processing pipelines according to your needs. AliceVision can also be used for stitching multiple images together into panoramas, converting low dynamic range to high dynamic range (HDR), tracking moving objects or stitching multiple images into panoramas.
When using this program, be sure to select high-quality images. Image quality has an immense effect on results – low-quality cameras will produce less detailed models and may have noticeable distortions; motion blur should be avoided by using tripod or fast shutter speed; natural lighting or controlled light sources should also be utilized when shooting scenes for best results; use Camera Calibration node for accurate and high-quality results.
Texturing
Photogrammetry software enables you to infer the geometry of a scene from unorganized photographs or videos taken in sequence – this method is known as Structure-from-Motion (SfM).
Once reconstruction has completed, you can view your 3D model by dragging and dropping its files onto Meshroom’s Images pane on the left side of the screen. JPEG, PNG, BMP GIF and AVIF file types are supported for viewing your model.
All data produced by Meshroom will be saved in a “MeshroomCache” folder associated with your project file – though you may choose to move this folder and its contents wherever is appropriate, as projects are easily portable.
To produce better textures from your reconstructed model, export/import the mesh back into one of several 3D modelling software programs like Blender, Maya or ZBrush and use their UV mapping feature. Meshroom offers some basic UV mapping as a result of reconstructing models; for optimal results however, retopologizing your mesh in your preferred software before importing back into Meshroom for texture editing is recommended.
An alternative approach would be to conduct a revopoint scan, import the mesh into meshroom and use its calculated point cloud to project textures onto it. Aligning both must be accurate; sometimes this requires trial-and-error, however Meshroom does not support use of NVIDIA GPUs for computed shaders.
