#CAAPerth Day Two – S1 – 3D recording, data capture and visualisation technologies for Rock Art

Chair(s): Geoff Avern, Jo McDonald

Discussant(s): Geoff Avern, Jo McDonald

Format: Long Paper Presentation with Roundtable

Schedule: Tuesday 26th 10:30 – 15:00

Room: Auditorium

Venue: University of Western Australia Club

Details from: http://caa2013.org/drupal/sessions

<live blog below>


Finished for lunch.


Q&A: noting that more data is not necessarily a good thing. You can end up with problems due to multiple errors. Also noting benefits of using targets for automated registration and hence improved accuracy. Again, back to the specific research questions. These also have an impact on data management. A: I would prefer a system where I had control over the stitching, registration and so on and where it was well documented, and transparent. (Reminded here of CHI work on RTI and empirical provenance and the digital lab notebook)


Quality measures of 3d data capture – accuracy of data points (range/ coordinate and colour), density of data points and coverage/ continuity of data e.g. influence of blind spots. So think carefully why we are recording? Is it a quick grab (of morphology)? Creating a replica? Is it a reference model i.e. aiming for best accuracy in all domains?

Anything working from a fixed position will have blind spots. Also the angle of incidence impacts accuracy so a single position does not allow this to be minimised. Colour data will vary, in addition to range data.

Also problem of laser scanners defining edges accurately.


Geoff Avern’s second mini-paper of the day – “What Information Are We Capturing? And What Are We NOT?”


Optical tracking systems and self-localising scanners e.g. NDI and Handy Scan. Also talking about structured light, and possible combination of this with triangulation beacons. Also talking about structure/shape form moton e.g. bundler, 123D, MicMac, Fit3D. Also Xapt – a CMOS camera – array of 16 creating a 3000 x 3000 pixels 50 frames per second. Suggesting mounting these on a drone for surveying rock art.


Geoff Avern talking now – “The Other” 3D Recording Technologies. Discussing laser scanning and photometric stereo and other photographic tools.


Follow up on my tweet about the 3d print.



Q&A: what option do Mirarr people have to edit and annotate? They have embraced it. Particularly the people who have been involved in the field. They want to bring in sacred sites information, walking routes through country, add video links about routes, schools getting more interested, etc.


Q&A: How deal with finding more image details afterwards, particularly given lots of superimposition? They find that they find most in the field – each image is individually recorded. DStretch is then used later and some additional ones found.


More details about the project on the Mirarr rock art website.


Very important to work out a system that allowed the Mirarr to have continuous, almost live action to the data as recorded. 251 sites recorded so far. First level was the basic record, second level more detailed, and third level very detailed recording. 8 days to record one site with 8068 surviving paintings.


Not sufficient detail from ipads – good for overview. Detail recording via DSLRs. Synchronised time stamps across the data. The system then used the synchronisation to combine the data on a secure online interface.


Collaboration with Environmental Systems Solutions


Aim to set up a world class rock art recording programme. Combining academic rock art experts with traditional owners, producing a catalogue, undertaking research – research aims landscape, people and environment, sequence, style and identity, heritage management and conservation.

Created a cultural data management system, undertooks survey and developed custom recording forms, looked at synchronising digital capture devices.



Mirarr Gunwarddebim – recording the Mirarr rock art. Area is in the Northern Territory technically within Kakadu National Park. Discussing Jabiluka protests 1990s-200s by Mirarr against mines.


Now starting – Recording rock art the Mirarr way (M. Marshall, S.May and G. Maclaren).


Discussion of practical application of RTI more generally for rock art e.g. extensive work by Cultural Heritage Imaging


Q&A: Were you able to capture RTIs during the day? Yes – done whenever the rock art sites were in shade where possible.


Martin Porr and Eleonora Gandolfi talking about UWA/ UoSouthampton collaboration on digital imaging of rock art in the Kimberley.


Connection collapsed for a while… Back on-line. My question related to what potential there was within their data structures for mediating access to material at a granular, contextual level e.g. only providing access to specific content within the panotour to specific groups. They have considered and implemented a broad data access structure that can respond to specific requirements. They also demonstrated the complexity of balancing different requirements. The question was prompted by our eMob project where we implemented very specific access controls as requested by communities, and subsequent discussions about the ways in which social medai sharing of images and other resources is subverting some access mediation.


They are using gigapixel imagery and 3d printing, and also tools to create digital annotations in three dimensions. Discussed 3d prints as alternative means of providing access to the digital record produced.


Now discussing contextual access. Now showing Manwanna dataset published via PanoTour at http://paulbourke.net/exhibition/Wanmanna/virtualtour.swf


Potential of using legacy data for processing via photogrammetric methods to derive 3d. (Note: We are exploring this at present on our archive of 30,000 photographs from the Portus Project to see if we can improve search and also identify components of buildings further to enhance our 3d record of the site).


Potential of relighting of models to show detail – shown via an animation but obvious potential for interactive. (No mention of RTI at the moment. Lots of discussion of rock art recording potential of RTI at our workshop yesterday.)


Meshlab examples shown with photogrammetric capture. Now discussing 2D vs 3d methods, via example of sites form India. He refers to 2.5D here to mean only capturing 3d data on one side of an object. Cape Labert recording examples shown next.

Limitations: baked on shadows, moving objects, fine structures such as grass


Opening paper J. McDonald – State of the Art: advances in Australian archaeological research and communication:

Laser scanning – cost and portability are an issue. Also complexity of use of point cloud data by non-survey users, Also the issue of replicability of results. Some difficulty in access to texture and colour. Finally the sheer amount of material.

As a result they have focused on photographic methods.