060109 VU Image-based modelling for archaeology (2018W)

Why this course?
Archaeology has always been in search of cost-effective approaches that accurately acquire three-dimensional (3D) geometry. Ideally, these workflows should also be straightforward to implement. Until a few years ago, acquiring 3D geometry using images (a process called Image-Based Modelling or IBM) was a task that only photogrammetric experts could fulfil. However, this situation has dramatically changed in the past two decades, mainly due to the many new insights obtained in the field of computer vision. As a result, powerful image orientation techniques such as Structure from Motion (SfM) and dense image matching techniques like Multi-View Stereo (MVS) have started to emerge and gradually complement proven photogrammetric concepts. The combination of these new concepts with the increasing computing power that became available to average users has resulted in a new mapping and 3D modelling paradigm that currently affects various research fields.

This success story can largely be attributed to the ease-of-use of current IBM applications, the seemingly limited knowledge necessary to create a geometrical 3D model and the wide variety of frame imagery that can be used: old and new, colour and greyscale, airborne and terrestrial, large- and small-scale. Although many users consider such IBM-pipelines as an ideal means to yield visually-pleasing, photo-realistic 3D models in a fast and straightforward way, many applications (such as archaeology) often rely on it to deliver (highly) accurate digital representations of real-world objects and scenes. However, this easiness of use makes many archaeologists unaware of the various pitfalls and technical difficulties that can still arise when using IBM. In addition, many 3D models are only merely created to ‘have a 3D model’, rather than to solve specific archaeological research questions.

Aim of this course
This course wants to remedy both issues by offering a very thorough introduction into the complete IBM pipeline: from image acquisition to the archaeological use of the generated 3D models, whether it is to research small artefacts, monitor excavations or survey complete landscapes with high-end orthophotographs. The course will, therefore, start with the essence of scientific photography and thoroughly explore (both in theory and practice) all the important photographic aspects (such as exposure, focal length, geometrical lens calibration, white balance, depth of field, image overlap) which are needed to obtain proper photographs for the complete 3D reconstruction pipeline. During this initial stage, students will also have to acquire their own photographs as to master all these concepts. Using a variety of case studies, the practice of IBM will be taught in Agisoft PhotoScan. The image sets and exercises will gradually increase in complexity so that students can become very familiar with the software while learning how to solve real-world IBM problems. If time permits, the usage and dissemination of these 3D models will be discussed: ‘How can they be shared?’; ‘How can new (geometrical) information be derived from them?’.

Methods used in this course
This course will consist of some theoretical lectures (supported by two pieces of literature), but most time will be devoted to the practice of photography and the IBM exercises. Students will learn how to critically undertake any archaeological IBM task and assess what information can be obtained from 3D surface models.

Expectations
Students are expected to bring a decent camera along (i.e. a camera that enables to manually dial in exposure settings). Moreover, students are expected to be familiar with standard Windows software and master English sufficiently well (hearing and writing).