logo of Bee Arts CIC

logo for BRE4TH - a woman's face in profile breathing a stream of neon yellow, orange and pink breath logo of the Wellcome Trust

BEE ARTS COMMUNITY INTEREST COMPANY RESEARCHED AND DEVELOPED THE ARTISTIC POSSIBILITIES FOR A NEW WORK BASED ON THE NON-INVASIVE DIAGNOSIS OF BREATH. THIS PROJECT WAS FUNDED BY THE WELLCOME TRUST.

AUGUST 2009 - MARCH 2010

Inside of the PTR-MS machineO

HOW BRE4TH WAS DEVELOPED - VISUALISATION DEVELOPMENT


Initial Thoughts

There were two starting points for the development of our visualisations. One was to find ways of visualising the data because we felt that for both scientists and members of the public, visualising the Volatile Organic Compounds (VOCs) might be a clearer and more immediate way to inform people about the presence of VOCs in breath. Our second motivation was to try and visualise breath almost intuitively, non-scientifically, in a way that conformed to peoples' expectations of what breath would look like if they could see it magnified to the molecular level. Our aim was to eventually find ways of combining these two approaches so that we could not only accurately and scientifically visualise the data coming out of the PTR-MS machine, but we could do it in a way that allowed for an intuitive understanding by the general public.

We also researched visualisation software being developed that could be used to visualise data such as Processing - http://processing.org/ (see also our Visualisation Research page for more examples).

Point of View (POV) of Visualisations
We thought a lot about the point of view the breath analysis visualisations should utilise and felt that we had a choice between two different points of view, moving directly towards the audience or moving from one side of the frame to the other, for example left to right. Seeing the breath visualisation moving from a central point and coming forwards into one's field of view intuitively felt like the right POV to use but we also felt it was worth exploring a sideways POV. Interestingly, the sideways method is the one used by most realtime graphing animations, such as the software that accompanies the Proton Transfer Reaction Mass Spectrometer (PTR-MS) machine.

These stimuli prompted a broad range of explorations following our central questions about how to reveal the hidden information in breath. Below are brief descriptions and video examples showing the progression through the development of our own visualisations.


1. Vapour test

The Vapour test visualisation was set up to explore methods of visualising breath as a cloud or a vapour. This uses the POV of breath coming directly towards the audience (1st POV) and it is a simple visualisation of breath as a vapour or cloud of very small particles.




2. Colour vapour test

The problem with the grey vapour test is that it seems to suggest that there is only one type of information in the breath whereas a significant amount of VOCs are usually revealed in breath samples. We chose to indicate this using a range of different colours, with each colour representing a different VOC. However, the test also illustrates breath as a lot of very tiny particles, and we realised that we would either have to find a way of magnifying individual particles and/or using imagery like this as a background out of which more defined particles would emerge.




3. Cluster test

This test looked at ways of visually clustering VOCs together so that they weren't as evenly distributed as they might seem to be in the earlier vapour tests.




4. Wave test

This test was an exploration of a 3rd person POV of breath moving from left to right which also looked at the trace trajectories that could be formed by the movement of individual molecules.




5. Breath background test

This test explored the same POV as number 4 above, but visualising breath more as a vapour. What we noticed was that this approach didn't enable us to separate out the VOCs we were interested in emphasising from the rest of the breath data and so we abandoned the sideways point of view and concentrated on the POV of breath coming directly out towards the audience.




6. Larger particle test

This test explored representation of VOCs in breath as larger particles as opposed to the gaseous clouds in the first two tests, whilst still trying to maintain a sense of particles coming directly at the audience.




7. Cluster test 2

We were very interested in the shimmering quality of this test and the idea that the visualised breath can seem to be composed of individual elements clustered together. This background was eventually used as the background for Number 13 below.




8. Star test

This test combined radiating lines with the idea of particles emanating from somebody's mouth to give a sense of movement while also enhancing the directional feel (coming into the audiences' field of view).




9. Foreground detail

This test was an experiment in visualising individual VOCs. Although we felt that this was an interesting and exciting way to visualise a molecule, in testing, the 3D models it produced were too complex to animate in response to the data in realtime .




10. Foreground detail 2

This was a second attempt at visualising individual molecules using a lower polygon count and requiring less intensive rendering. We were interested in it as a potential alternative to the usual notion of molecules being made of tiny atoms, which tend to be visualised as spheres. The manner in which this visualisation seems to pulse with waves of energy travelling around it, struck us as being scientifically interesting without any attempt at being a precise rendering of a particular molecule.




11. Molecular background

This test was an attempt to combine the visualisation ideas that we felt were more successful, i.e. individual molecules forming a cloud that was expanding towards the audience out of somebody's breath. We felt that this could work well as the background breath out of which we would need to separate individual VOCs.




12. Background and foreground combination

This was one of our more successful tests in combing:
* the notion of a vapour,
* the front directionality/POV
* the separation of five individual VOCs
* with each VOC differentiated by colour.
We showed this visualisation to Jens Herbig, Chief Technology Officer of Ionimed Analytik (the organisation that loaned us the PTR-MS) who not only liked it aesthetically but commented that he felt it was a good visualisation that successfully separated key VOCs out of the complexity of breath data.




13. Foreground motion

One of the difficulties that we experienced was turning these designs into computer graphics that were fast and able to be rendered very quickly in realtime response to the PTR-MS data. This test was a very simple illustration of how three different VOCs could come out of the vapour of breath and into the foreground to be identified as specific VOCs. The test was part of the visual brief for the software developer who created the software link for our visualisations as used at the Science Museum Dana Centre events.



14. Final visualisation

This is an example of one of the final visualisation styles that we used at the Dana Centre events. It didn't achieve the visual look that we were hoping for but was low enough on polygons to be workable for the live visualisation. The top half of the screen shows the analyses of two mass scanned breath samples overlaid so that shared peaks of atomic mass units were easy to spot. The lower half of the screen shows highlighted VOCs from one of the breath samples on the left and the comparable VOCs from the other on the right. The ringed molecule shows the particular VOC being looked at in detail (also highlighted with circles on the top image) with text that provides the name of the VOC, its atomic mass unit and the concentration of the substances in parts per billion. To support the visualisations we programmed the functionality of cyclically rotating through the important foreground VOCs to allow for easy comparison of quantity in both breath samples.

We satisfied our brief in terms of this visualisation being clear and easily understood by a mixed audience (some very scientific and some not) however we were disappointed in not being able to achieve visualisations that were aesthetically closer to test 12, or more developed than that.




15. Photoshoot and video titles

During the press photoshoot for the Science Museum Dana Centre events, we hired models who we asked to act as if they were inhaling each others' breath and skin scents and set up visually striking opportunities for photographers. One of the more successful visualisations projected behind the silhouettes of the models was the Star Test, No. 8 above. We were very happy with this composite imagery includig the models and visualisation and decided to also use it as the title sequence for each of the five event videos.