Western Australia from Space

The brand new $396-million-dollar New Museum for Western Australia, opening in 2020, approached me for a visualisation of Western Australia from space, having seen a fulldome sequence from my 4k planetarium movie, Dark. They needed something high-resolution and not too expensive. And not just a screenshot from Google/Bing/etc Earth – the Big Tech clichés of how we see our planet.

Of course, I jumped at the chance – what a privilege – but the details were a bit daunting. Their reference image was this shot from Dark (which one of their designers had seen):

The original is a 72dpi 4k fulldome frame – so 4096 x 4096px (~16.8 megapixel).

But, it’s a quite different request, when a client approaches you and asks to use a frame from a movie and then asks whether they can print it @400dpi to put on a wall. The two images the Museum requested were (1) 639 x 1050mm and (2) 2270 x 2300mm. A0 is 841mm x 1189mm. But to print at 400dpi with a 72dpi source at this scale would be disastrous: soft, pixellated, just useless – especially with complex recognisable textures. I know exactly what the problems are having worked at a high-end desktop-publishing company in Melbourne in the mid-1990’s, operating Iris (giclée) printers. Beautiful, complex machines that you need to know a great deal about to get fine-art archive-quality results, let alone wall-scale prints that stand up to scrutiny.

What the New WA Museum needed were not simply interpolated, scaled images but newly rendered ones with entirely new ultra-high-resolution satellite image maps, star maps, atmospheric blends, Raleigh scattering, cloud maps and so forth – not simply a reproduction of that single shot from Dark.

This required a re-created re-textured model, shot with a controllable virtual wide-angle lens, not a fisheye. Calculating the pixel resolution for 400pdi, it approaches something like (1)10062 x 16535px (~166 megapixel) and (2) 35748 x 36220px (~1.3 gigapixel). These are big images, with huge amounts of detail. Nevertheless, I decided it would be best to provide them in 1:1 aspect ratio, so the designers could move around a bit: it’s important to have a bit of flexibility.

It’s a lot of data, but mostly it’s a question of what sort of data, where to get the data and how to process it – a mixture of processed non-visual data sources, some obscure scientific file formats, publicly available high-resolution sources, machine-learning and custom routines.

So, to cut a long story short, the New Western Australian Museum got their gigapixel images and I ended up building a 3D animated Earth model with adjustable, compositable layers that enable me to create reasonably accurate illustrative gigapixel images of the Earth from any angle, anywhere on the Earth, at any time of the year – for a defined snapshot. And that serves the purpose. To make it concurrent and updatable is a big job, but do-able – just time and money. It’s work-in-progress that I hope will become a more accurate representation over time – enabling a far closer view than you see here. Of course, how much more extensive non-public data is, one can only imagine – sometimes this stuff can be hard-to-find, if not inaccessible, other times there are obscure treasures hidden away in public repositories – if you know where and how to look.

I look forward to extending it beyond the surface of the Earth, both within and without, working out a methodology and pipeline to semi-automate how one might approach other planets in our solar system and even exoplanets: given manageable data recorded by instruments designed by brilliant and enquiring minds, and made available for exploration – let’s imagine, to infinity and beyond!

We live in amazing times, where there is an extraordinary variety of publicly available scientific data that enables us to envision our planet in ways that were never before possible. This is part of the commonwealth of scientific knowledge – for all humanity, our custodianship of the Earth we live on, and our responsibility as part of nature. Our unique planet that no-one owns.