This paper is investigating an existing photogrammetry workflow to find a method for increasing its viability for scanning in smaller objects in a production scenario. We are taking a closer look at the different production steps within the scanning process, to examine their influences on the overall production speed & quality and investigate how these influences could be adjusted to positively influence the overall workflow. The gathered data is than used to build an automation toolkit to help us removing the bottlenecks, which found throughout our research. Afterwards, we are comparing the workflow of our automation toolkit with the previous manual workflow to draw a conclusion about its effectiveness and show an example use case.
This paper is no meant as step-by-step guide trough an ideal capturing process, but rather as overview about the remaining possibilities within existing workflows, regarding their optimization & automation to increase their viability. It lists the different steps of my research and explains the reasoning behind their implementation within the automation toolkit.
The research showed that the current process of photo scanning objects via photogrammetry leaves a lot of manual & repetitive labour for an artist within the workflow. Although the computer does all the complicated computations for the user, the workflow requires him to do all the simple boilerplate work to pipe data through different applications. Though this are small and simple tasks, their overall influence in a larger production can quickly add up to a big cost factor.
This is where the usage of an automation toolkit comes in handy, as it takes over many of these tasks from an artist, so that he/she must spend less time on the creation per asset. Though the viability of such an implementation depends on many factors, like the complexity and required flexibility of the underlaying workflow. Furthermore, in the case of photogrammetry, on the amount and type of assets which needs to be processed. As not every object is suited equally well for photogrammetry and too much manual clean-up per asset might eat up the benefit of the automation again.