Image Mapping

Photographing subjects using multiple images to create hi-res canvases

In recent years a lot of innovative work has been produced in the field of high-resolution image mapping, using the latest digital technology. We have all seen glorious images of paintings, that have large scale zoom capabilities using megapixel cameras and actionscript coding, allowing the online user to zoom into a feature of interest. Rarely however have these techniques been employed in producing broadcast television sequences, and with good reason as we have discovered on the new BBC 2 series ‘Renaissance Revolution’.

The average UK television viewer will no doubt miss any innovative clues or notice the influence of modern technology as they listen to Matt Collings narration and view the inherent beauty and hidden symbolism of the Raphael’s ‘Madonna of the Meadow’, Bosch’s ‘Garden of Earthly delights’ and Piero’s ‘Baptism of Christ’.

So! As the leader of the team behind the ‘graphic sequences’ I feel we should add subtitled questions as each graphic unfolds. Did you notice that you have entered a graphic sequence, which had been cleverly substituted on the end of a camera tracked hand held shot? Did you notice that as the camera zoomed, swooped and panned, there wasn’t a single pause which would have made our job infinitely simpler? And another thing, did anyone notice that we have been travelling from the gallery wall and have finished the move by examining a detail, which is less than 1 cm squared. This is all displayed for you at 1920 x 1080 pixels so that you can still experience the delights of seeing this painting in high resolution. Did anyone notice? Of course not!

How it was done

After being approached by ‘Blakeway Television’ we began the challenge of how to digitise great works of art which would allow us to develop a very flexible system to turnaround HD sequences so they could fit into a typically demanding and changeable television production process.

A professional photographer friend, Rupert Truman, was called on to help solve this puzzle. Rupert is more often seen in a large outdoor location shooting a surreal scene for a Muse or Biffy Clyro CD cover, but it was Rupert’s former experience in using older large format cameras to record architecture which was to assist ‘hitting the nail on the head’.

After some experimentation it was decided that we needed the most modern equipment to record the digital information, which was, the Phase One P65 camera ‘back’. This is the piece of kit, which fits onto the High definition famous Hassleblad camera, which can record an amazing 65,000,000 pixels in one image. This was only the first of many hurdles jumped as even a range of tiled images using this capture device still meant that the process of stitching everything together was complex, time consuming and potentially troublesome. There are a number of software stitching products on the market but each use some form of distortion to match the tiles and we all wanted to avoid this.

Our solution was to use what some would consider an old fashioned method. We used a large format camera, the ‘Sinar’ 5×4 but with a sliding back device which allowed us to capture 4 tiles from one camera position. We all remember the image of a photographer with his head under a cloth staring at a plate glass screen; well it was a similar camera but with added technology attached. The sliding back meant that by attaching the Phase one P65 to one of 4 positions we could in short achieve an image of 260 mega pixels without moving the camera lens. By extrapolating these numbers we could achieve a print resolution (300dpi is the glossy Sunday magazine standard), which was 12 metres in height.

The process of photographing the paintings themselves was a real honour. Sharing the National Gallery with no one other than the early morning cleaners was one thing but our trip to Vienna to capture Raphael’s masterpiece was certainly something to remember.

In Vienna’s grand ‘Kunsthistorisches Museum’, we were taken to the ‘Restorers’ wing where in a large institutional room, an array of European painters’ masterpieces were sitting around on industrial type trolleys waiting for attention. Our final destination in the restorers studio was a welcome relief as the studio was already kitted out with suitable lights and a remote controlled easel to rest any medium sized painting.

In readiness for our visit the gallery staff were quickly removing the ‘Raphael’ from the wall and then removing the gilt frame and glass. But the biggest surprise of all was that after careful securing on the easel we were left to our own devices for the rest of that day – just myself, the photographer, his assistant (Richard Round-Turner) and of course Raphael.

The process of capture.

It was quickly apparent that as Rupert said we were on the edge of what was possible; not due to available software, but more from the glass optics (lenses) available to us. The camera had to be completely parallel to the painting to keep within focal depth, so much so that in areas where there was more relief the focus was drifting off and stopping down further was our only re-course. After calibrating the colours and light using our ‘grey card’ and the ‘lightroom’ software, we were able to transfer the images onto my ‘macbook’ for a quick check of line-up. This process was gratifyingly simple and we could therefore proceed as planned but the day rapidly vanished beneath digital tiles and documentation.

Because the producers had been keen to avoid any sequences in the programme which looked like ‘graphics’ or ‘analytical’ sequences against the ubiquitous black background, we had to ask for the frame itself to be taken to the studio so that we could capture its features in almost as much detail. In following this methodology through, then we had to venture upstairs to the gallery to photograph the wall where it normally rested. This was essential so that we could also seamlessly carry a post produced graphic sequence onward from a camera shot earlier in the production process.

Painting by numbers

For once the term ‘painting by numbers’ has a new credence as even the most advanced computing power from CERN and NASA combined was unlikely to help us process all these digital tiles unless we applied the maths and a system we call ‘swapping out’.

All the images had to be grouped into at least 3 compositions of, wide proxy images, mid range tiles and close up details. When our virtual cameras started on the wide image of the painting in the gallery we used a proxy image, which was more than enough to retain high resolution information until the camera ventured closer. At this stage the mid tiles took over and the proxy faded off. But when the camera went ever closer we were forced to use an extreme close-up detail, which was usually the point when a software display pop-up appeared and became our ‘message of doom’, which read something like ‘unable to process the image caching’.

When people ask how you did something they usually expect to hear about some ground breaking new technological advance. We had that in the process of capturing onto 65megapixel camera backs; but the real trick in how it was done comes down to working out new methodologies to get around technological limitations or simply ‘swapping out’.

The sequences themselves were programmes in Adobe after effects using virtual cameras over the tiles in a virtual 3 dimensional space. It was simply the scale of this production, which moved it into a realm of complexity rarely seen on our television screens.

Hopefully, as I say, most of the graphics sequences should look more like carefully choreographed sequences based on Matthew’s commentary. If this is so then we have been able to move the technology into the background and concentrate on elegant moves and interrogation of a painting, which can be understood better because of the resulting visuals.