Free photogrammetric digital reconstruction

EDIT: I forgot to put this here: the post below is a author’s guest post by Peter Falkingham of the School of Earth, Atmospheric and Environmental Science,
University of Manchester. Enjoy!

This blog post coincides with my latest paper on open source photogrammetry software that appears in Palaeontologia Electronica yesterday.  I want to briefly summarize the paper, but also to tell you about where the technology seems to be heading, and try and communicate some of my excitement.

Using just a normal camera, free software, and a desktop computer,
you can generate 3D digital models of everything from hand specimens,
through mounted skeletons, and beyond buildings and outcrops.

In the paper, I present some freely available, open source software for producing high resolution 3D models of specimens ranging from centimetre scale to outcrop scale, using photographs taken with an ordinary camera and processed on a desktop computer. The software was developed by researchers at the Graphics and Imaging Laboratory of the University of Washington, principally to generate 3D models from the growing databases of photographs online (see, “Building Rome in a day”). My paper describes where to get the software and how to use it, and presents a number of examples to illustrate the quality of the models. Great models comparable to those acquired through laser scanning can be generated using any consumer camera, without manual point matching or calibration. The idea is that I wanted to make it explicit to people that the realm of 3D data acquisition is no longer restricted to expensive photogrammetric reconstruction software or laser scanning equipment.

Why do we want digital models?

Production of 3D digital models of palaeontological specimens is far more than just making pretty pictures for publications.  Previously, digital models have been used to look at body mass reconstructions (Bates et al., 2009a, 2009b; Hutchinson et al., 2011), ranges of motion (Mallison, 2010), conservation (Bates et al., 2008, Adams et al., 2010), or visualisation e.g. cross section/profiling (Falkingham et al., 2009), among many other uses.
Digital models provide an objective data source that can be distributed and shared among researchers as easily as sending an email. The key word there is ‘objective.’ It means that you or I can perform a study, or analysis, and then give the model to someone else to either repeat the analysis or build upon it using the same base data. Journals which are embracing the internet as a medium for publication, like Palaeontologia Electronica, provide a fantastic outlet for presenting this kind of data with publications.

Keeping pace with technology

Of course, the problem with trying to publish on rapidly advancing technology is that it does just that – it rapidly advances! When I first submitted the manuscript last year, the commercially available photogrammetry software I was aware of was all very expensive, and required camera calibration and/or specific markers in the images. Since then, the price of commercial software has really plummeted, and they’ve only got easier to use. $200 can get you software with a nice graphical user interface (GUI) to generate the 3D models with the click of a button. There have also been a number of online web services appearing that allow you to upload your photos, and then download the finished model, either for free or relatively small fees [though if using these you should read the terms and conditions carefully, as they might take copyright of the photos and digital models]. This all makes generation of 3D digital data even easier (while the programs described in my paper are easy to use, they are not always entirely easy to set up).

Nevertheless, I maintain that the open source community is where the bleeding edge of the technology lies. The programs used in my paper seem to handle larger data sets more capably than the affordable commercial software (undoubtedly because this is what they were originally intended for with sourcing thousands of images online). They are also being actively developed – take a look at VisualSFM, which provides a GUI for the previously mentioned programs, and importantly takes advantage of the graphics processing unit (GPU) which, in certain cases, can act as the equivalent of hundreds of processors. This enables the software to match the photos into a 3D model in real time (check out the video on the VisualSFM website), using a desktop computer [dense reconstruction still relies on PMVS and runs on the CPU, but this is actively being developed to run on the GPU instead].

Looking forwards

It’s very easy to get lost in the possibilities of photogrammetry in the immediate future. In my paper I allude to using photogrammetric software on smartphones. This year, we’ll be seeing smartphones with quad core processors and 1gb of RAM or more, making smaller models easily within their processing power. Already there are apps available for android and iphone that can generate [admittedly basic] 3D models from 3 or 4 photos taken with the phones built in camera.

But it’s not just making the current process easier or faster, it’s the novel possibilities that are really exciting. For instance, how about capturing the 4th dimension, time? Or how about creating accurate 3D models of specimens which were photographed, but are now lost or damaged? One could imagine a fossil tracksite visited by hundreds of tourists each year but exposed to weathering; by using google image search and filtering by date, one could generate 3D models year by year, retrospectively documenting the rate of erosion.
The production of digital models through both laser scanning and photogrammetry has been around for some years. But now that the affordability and usability have reached a level where anyone can produce digital models as needed, I look forward to seeing the ways in which palaeontology can grow, both through novel applications, and through dissemination of data.

References

Adams, T., C. Strganac, M. J. Polcyn & L. L. Jacobs (2010) High Resolution Three-Dimensional LaserScanning of the Type Specimen of Eubrontes (?) glenrosensis Shuler, 1935, from the Comanchean (Lower Cretaceous) of Texas: Implications for Digital Archiving and Preservation. Palaeontologia Electronica, 13, 1T:11p, http://palaeo-electronica.org/2010_3/226/index.html.

Bates, K. T., F. Rarity, P. L. Manning, D. Hodgetts, B. Vila, O. Oms, À. Galobart & R. Gawthorpe (2008) High-resolution LiDAR and photogrammetric survey of the Fumanya dinosaur tracksites (Catalonia): Implications for the conservation and interpretation of geological heritage sites. Journal of the Geological Society, London, 165, 115-127.

Bates, K. T., P. L. Falkingham, B. H. Breithaupt, D. Hodgetts, W. I. Sellers & P. L. Manning (2009a) How big was ‘Big Al’? Quantifying the effect of soft tissue and osteological unknowns on mass predictions for allosaurus (Dinosauria: Theropoda). Palaeontologia Electronica, 12, 33.

Bates, K. T., P. L. Falkingham, D. Hodgetts, J. O. Farlow, H. Breithaupt Brent, M. O’Brien, A. Matthews Neffra, W. I. Sellers & P. L. Manning (2009b) Digital imaging and public engangement in palaeontology. Geology Today, 25, 134-139.

Falkingham, P., L. Margetts, I. Smith & P. Manning (2009) Reinterpretation of palmate and semi-palmate (webbed) fossil tracks; insights from finite element modelling. Palaeogeography Palaeoclimatology Palaeoecology, 271, 69-76.

Hutchinson JR, Bates KT, Molnar J, Allen V, Makovicky PJ, 2011 A Computational Analysis of Limb and Body Dimensions in Tyrannosaurus rex with Implications for Locomotion, Ontogeny, and Growth. PLoS ONE 6(10): e26037. doi:10.1371/journal.pone.0026037

Mallison, Heinrich. 2010. The digital Plateosaurus II: An assessment of the range of motion of the limbs and vertebral column and of previous reconstructions using a digital skeletal mount. Acta Palaeontologica Polonica, 55:433-458