 Ladies and gentlemen, colleagues and friends, including those who cannot be with us today, to see that rock art is in danger is not an understatement, weathering processes and destruction by humans have already greatly reduced the amount of information available the condition of the art is continuing to deteriorate, therefore we must redouble our efforts documented if we are to give our past a future. As rock art researchers, we are fortunate to live in exciting times Lleidwadodd a Lleidwadodd i'r syniadau nôr, ond yn cael eu cymdeithasolio. Dwi'n ei gweiniad i ffodillwn mor wirio'r cyflwynt hwn yn hebodol. Mae ymddych chi yn gweithio'n ymddych â'i cyflwyntion nôl eich dd Wirio'r cymdeithasol? Mae'n hwn yn meddwl a'u gwylio'r cyflwynt ni, hwn yn meddwl i'r cyflwynt, a'i hollwn i'm gweithio i'r warchioneddau. However, as the application of the new technologies and their associated methodologies becomes more widespread, we need to remain critically aware. All of a sudden, it seems all archaeologists can document rock art. The specialisms possessed by a few are rendered redundant by the new technology available to the many. True, one can learn the essentials of a good surface-based documentation over a relatively short period of time, but this is not mutually exclusive to the digital methods. As the annual working seminar in the Swiss Museum demonstrates, it is very positive for the future of the discipline that more are taking interest in it. However, to obtain the best results from the technology still requires skill, arguably even more than before. To address all these heterogeneous elements and their articulations requires everything from an artistic eye to computer science and everything in between. Even though we are faced with as yet unknown possibilities, there are and will always be limits to how far one can push one's interpretation. These appear to be poorly understood by many, including experienced researchers. There are a number of false friends out there that can lead to some unlucky conclusions, to quote amateur archaeologist and finder of rock art, Martin Stolter. The perceived specialisms of the rock art research craft are seen as a black art by many outside the field. Although efforts are being made at Göteborg and Aarhus, the skills required are not widely taught at university level. Therefore, the methods appear sinister to many. For archaeologists and scientists, sinister is often read as being subjective or having dubious concrete foundations. As my colleague Enon Maier and I recently wrote, for many the model has become the reality and research is moving the focus of study further and further away from the rock. This leads to a loss of what has been termed by Gerhard Myrstwyth as the dialogue with the surface. This proposition is taken up and examined in greater depth in the researcher colleague Diodd Cwfford. In particular, Cwfford has highlighted the importance of fingertip inspection and its essential skill mastered only with experience but often overlooked. She writes, quote, Some people, although with the best intentions, argued that the tactile methods are becoming unnecessary with the developments and the digital techniques. The author entirely agrees with Cwfford that this position should be seen as preposterous. Equally preposterous is the notion that many colleagues claim they can see or illustrate things in the model that cannot be seen by the naked eye. This, I believe, to be fundamentally false. The resolution offered by the available technology is not yet sufficient to support such conclusions. Furthermore, there appears to be a fundamental misunderstanding concerning what compromises objectivity, particularly the difference between mechanical and true objectivity. There also seems to be a lack of awareness of the operation of the methods, that is to say how the result on the screen is derived. Objectivity is one of the key themes in these arguments, how we define it, how and if we are trying to understand it or if we are simply playing the ostrich. Increased or total objectivity is often used as a justification for the merits of new technology. However, few actually discuss what objectivity is, and no one, in rock art, is discussing what it actually means in the context of the new technologies and methodologies we are applying. We draw conclusions from study of the surface of the model, often without reference to the original, without understanding how that result has been obtained. We often fail to take enough control of settings or show awareness of the limitations of the technology. Looking at the texts of several recently published articles and from discussions with my colleagues, it has become apparent that this discipline has failed to understand the meaning of objectivity in the context of digital technologies. So, what is objectivity? There is more than one definition, depending on the context, and we need to remain mindful of which context we refer to. Structure from motion, structured light and laser scanning are mechanically objective. A computer is a deterministic machine. It cannot produce a truly random sequence. Given a set of input data it will produce accurate, that is, repeatable results. However, given a different set of input parameters, a different result will be produced. This is more so for SFM than with scanning, where low light, distance from the object and overlap between images introduce variation more regularly. The question is whether these differences are significant enough to influence the answers to the questions we are posing. What archaeologists tend to think of when they refer to objectivity in the context of the documentation of rock art is something that is independent, unbiased and a true representation of reality, devoid of any interpretation or operated influence. Despite the shifts in developments in technology, it seems that we are often not really dealing with a discussion of digital over analogue, or more versus less objective, but of surface-based versus bigger-based methods. The possibilities and limitations offered by existing and emerging techniques, and how we interpret the copies we have created, complete with their imperfections. The merits of surface-based documentation over an interpretive one is not a new one, although it has still been discussed in literature. The world's first large-scale surface-based documentation project was begun 40 years ago, following the decision taken in the late 1980s by Gerhard Mistrae, director of Tarnham Museum of Rock Art and Research Centre Underslufts in Sweden, and Ulf Bertelsen, head of the Swedish Heritage Board at that time, to move the fieldwork of Tarnham's Holistons Museum of Underslufts away from a figure-based documentation to a surface-based one. Let us now move on to consider operation. You will note that I mentioned structured light and laser scanning separately. Many users mistakenly believe that the leading product on the market, Creofon's Handyscan 700tm, is a laser scanner, and therefore independent of the photogrammetric system. The texture is built using on-the-fly photogrammetry by two stereoscopically moated calibrated precision optics operating under the range-finding principles. They measure the deformation of the lasers projected onto the surface with reference to the reference values obtained during calibration. On-the-fly image recognition extracts the target points and segments these to create the coordinate system for the model. The precision optics may be affected by pressure, temperature and humidity. The scanner is neither as accurate or precise as we seem to believe. The figure often cited is an accuracy of 0.06mm at the point of initialisation. The latter part of that sentence is not cited by those using the technology, despite the Scandinavian agents emphasising it. Unpublished tests witnessed by the author show that when a circle measuring more than 1.5 metres in circumference is scanned, the 5mm error appears, and the measure does not meet in the same plane. Therefore, we can conclude that the greater the distance from the initialisation of the scan, the less precise the model is. There is also an issue with errors in the writing of the surface. Noise appears in the models at times. This is probably due to the precision of the machine, reflections from components of the rock, and the challenge that the system has when working with translucent materials, like plants, in our case lichen and bacterial growth. Lens distortion is not unexpectedly a problem in SFM. When scaling the models, the position of a point varies widely due to the distortion of the lens. Light levels affect this, leading to higher residual values. Residuals referring to the spread of values for the distribution of the points in the model. Intern leading to precision. I stated in my abstract that models can be manipulated. But so can other established forms of documentation. Manipulation occurs on two levels, intentional and unintentional, and sun transformations are more harmful than others. When it comes to the models, I find it more useful to think and refer to them as visualisations. It is also critical that people make their models available, and that the transformations applied are made clear as part of scientific transparency. The models are loaded with interpretation in the choice of settings, the area and amount of data included in the documentation. I note that many use the interpolated and extrapolated functions. On the one hand, in a well-meaning sense, to increase the number of features included in the point cloud and mesh. However, the very nature of their probability-based results, as Agisoft themselves state, in neither accurate or precise reconstruction. One might argue, so what? However, the question of how accurate the reconstruction is becomes paramount when assertions are made that we can see things in the model that cannot be seen with a human eye. It all boils down to a case of resolution, and if that resolution is adequate enough to answer the questions we pose. SFM is subject to variation, while structured light and laser scanning is consistent. In SFM, position areas are most evident during the scaling of the model by our manual correction. Agisoft and War Open are clear-form about the workings of their technology. The need for commercial secrecy of the software providers is understandable, but problematic for ascertaining the quality of what we are producing. This is detrimental from a scientific perspective, as academic convention now requires reproducible open science. I am of the opinion that there is no information that cannot be detected on the surface itself, as Maya often says, that surface contains more information than we have yet recognised, and this is why we need to document as many whole surfaces with the best means available to us. In this quest, there is no room for petty squabbling, politics or personal self-interest. Surface-based methods and evaluation of our data in consultation with the surface force us to look at the material with new eyes and new methods to give us new visualisations. If we are to believe we can see things in the model that we cannot detect with the human eye, it is generally accepted that the minimum size of an object could be resolved by the human eye is not 2mm, around the width of a human hair. Therefore, if we are to make this claim, the method we are using must be able to show an object at least the width of a human hair and moreover its structure. SFM and structured light scanning do not resolve objects the width of a human hair. I hate to break it to you, but in frontage, if a hair comes between the paper and the surface it shows in the rubbing. What we want when we create our copies is a one-one reproduction. The copies we have made are a great achievement and very useful, but they do not show as much as we sometimes think they do. Look at the information about the copy you have made and make your conclusions accordingly, according to the circumstances. We also need to be mindful that we are using technology that has not been designed with our needs in mind. The scanner is developed for use in reverse engineering. Different deviants from defined tolerances is most important. According to CREA form themselves, prediction of the surface is used to speed up and streamline modelling and smooth planar surfaces. This predicted surface is only replaced if the margins of tolerance defined by CREA form and therefore unknown are exceeded by the actual value. On the other hand, when we use the author function and photo scan, we are using the technology as intended. The alignment matching a mosaic in multiple aerial photos. The only differences are images are not from a satellite satellite. Let us now consider interpretation. As my colleague Ellen Meyer and I recently wrote, the model has become the reality. We study our computers and not the original rock surface. Yet the best model, the best representation of reality is not the model people fool themselves into creating thinking they have captured, but the original surface. So when the sun is shining, when the lamp is illuminated at night, there is no need for a model. The best model is the rock itself. All the resolution we strive for is there. It is the only true source of information. Again, as Crawford observes, nothing can replace study of the surface on site. To this, I add, nothing can replace the surface once it is weathered away, as construction will be in relation to limited understanding of a known geological distribution. Three paragraphs. Yr cwmgold hon yn iawn o'r blahau, both separately stated that a perfect documentation taking into account all the aspects that should be taken into account does not exist. For this reason, in my abstract, I call it a utopian elixir. We need again, as Crawford, Meyer, Millstor and myself have repeatedly stated, to use as many possible tools as we can from those at our disposal. For example, my own work in the field can call on artificial light, footage, fingertip inspection, favourable sunlight and image-based modelling. Models are processed in the field using cluster processing running on a supercomputer, and it's often possible to pull down the process model for study whilst on site. In the future, there remains an absolute and definite need for the specialist skills of the rock art specialist. We are still in dialogue with the surface, only now with even more powerful tools at our fingertips. Thank you.