For both photogrammetry and modeling, you need to anticipate spending time troubleshooting and correcting your models.
For both photogrammetry and modeling, you need to anticipate spending time troubleshooting and correcting your models. This can make it very frustrating to try to estimate a timeline, as it is difficult to predict where in the modeling process you will encounter difficulty. Both processes also often require a fair amount of hands-off time while projects render. During this time, you may not need to be present but you also may not be able to use your computer for other work.
The photogrammetry workflow usually looks something like this:
Set up your “studio” with camera, lights, necessary scaffolding and support materials. This can take anywhere from a few minutes to a few hours.
Take photographs of your subject from all angles. A good photogrammetric model can require anything from a few dozen to over a thousand photos, depending on its size and the complexity of its surface. For a small object without much deep relief, you can acquire these in a few minutes once you are familiar with the process. For a building, this might take several days, especially if you are using drone or other aerial footage, or if you encounter bad lighting or variable weather.
Load photos to your computer and preprocess. Usually this is pretty quick, especially if you know how to automate things like changing file formats.
In the software (like Metashape), you will align the photos, mask unneeded environmental elements, produce your point clouds, create a mesh, then texture the mesh. This part can be very time-consuming, both in terms of your active time as well as the time that the computer may need to work on a stage before you can do something else. Processing time is hugely variable depending on how powerful your computer is, how many photos you use and how consistent they are, how complex the model is, and other factors. Expect to spend a few hours on a small model, especially if you’ve practiced working with the software, and perhaps around a day if the structure is more complicated. You can break your model into pieces and stitch it back together to save processing time if you have a very complex model. At any of the stages in this part of the process, you may need to go back to stages 2 or 3: allow time to retake photographs if you discover you’re missing a section or that the images from one area won’t align properly!
Export your model for display or further modification in other programs.
Ideally, you want to start processing your photos while you still have access to your subject for photography so that you can retake photos if you need to. In practice, this may not be possible, especially if you are working on multiple models or working on location somewhere. Taking the time to make sure your photos are high quality and complete will save you a lot of time down the line.
If you’ve practiced and become familiar with the software and how to take photos for photogrammetry, you can process a model fairly quickly and do several small object models in a single day. This makes it reasonable to do this type of project with students in a class, perhaps with one class period devoted to acquiring images and another to processing them into a model, but in these cases you will want to be especially aware of how the computer speed can leave a lot of time when you can’t move on to the next step.
Your project will go much faster if you allow time before you start to practice photogrammetry with simpler objects or structures. The easiest objects to start with are those with a matte surface, without deep relief or cavity openings, and where every side or angle looks different. The easiest buildings to start with are simple facades without windows or shiny surfaces, and the best weather is overcast without cloud variability. These kinds of models will typically be faster to photograph and to render, with less likelihood of the types of errors where you have to go back and retake photos.
The workflow for modeling is very dependent on the thing you are modeling and the software you use. Typically, you might consider the following steps:
Acquiring information about your subject: measurements, materials, etc.
Building the structure of your model with its underlying geometric volumes and parts.
Assigning materials or surface finishes to parts of your model.
Adding elements like joints, moveable parts, skeletons, cloth structures, and other pieces that relate to how the subject responds to environmental conditions (optional, depends on what you’re modeling).
Creating an environment for your model, making animations or time lapses, and other features that show how your subject functions or responds to given conditions (optional).
Exporting your model for others to use or for you to put into other programs.
Each of these stages can take anything from a few hours to days or months, depending on how well you know the software and the object, how many details you want to include, how those details interact with each other, and what you are doing with the output. You could make a simple model of a contemporary five room house in SketchUp in an hour or so, projects with moving parts, surface detail, and specific lighting can easily take fifty times as long. While photogrammetry is a process of translation, where you are working to create a three-dimensional model of a subject that is in front of you, this sort of modeling is much more experimental and typically requires a lot of iterations as you figure out what processes and levels of detail create the model you need.
Continue Reading: Prototyping & Wireframing Dimensional Projects