My research methodology demands that I provide a detailed definition for each project in the case study. Rather than defining each of the three projects at the outset, I preferred to reflect on each project in turn, establish what had been learned and identify questions that were either fully or partially unresolved. So, by factoring these into the design of the next project, I have been able to create the conditions where these outstanding areas would arise and could therefore be tested in practice.
Project 2 identified clear practice which could deal effectively with loss of coverage, texture smearing and doubling. Moreover, the nature of the camera move did provide some loss of texture resolution and I was able to use the focal length of the projection camera to widen the field of view and there adjust the resolution of the image that would eventually be projected from that camera. Whilst this technique did overcome the problem, the project only provided limited scope. For the third project I wanted to explore these techniques much more by creating a camera move that would push extensively into/through an environment thereby creating a constant and incremental magnification of areas of the matte painting leading to loss of pixel resolution.
For this, I am drawn to the theoretical principles presented by Ray and Charles Eames work on “How to Deal with the Relative Size of Things in the Universe”, applied to filmmaking. The resulting concept became known as the ‘Powers of Ten’, which refers to any of the integer powers of the number ten; in other words, ten multiplied by itself a certain number of times (when the power is a positive integer). The number one is, by definition, a power (the zeroth power) of ten.
(Eames & Eames 1977) https://www.eamesoffice.com/the-work/powers-of-ten/ This work can be seen to simulate a camera being pulled up from earth and deep into outer space, and then pushed into an organic form at molecular level, all without optical trickery or excessively large images.
Earlier in my study, I had explored a projection technique briefly discussed by Garrett Fry in a training video for the Gnomon Workshop: https://www.thegnomonworkshop.com/tutorials/advanced-3d-matte-painting-techniques/ Chapter 11. Last accessed 11th June 2019.
My exploration of this technique is documented in my blog: https://pholmesphd.home.blog/2019/06/25/projection-tests/
I believe this highlights a degree of tacit knowledge insofar as I describe a workflow that can deliver the ‘powers of ten’ principles, albeit within sterile lab conditions and using only checkerboard textures as proxies for actual matte painted images.
The ‘Powers of Ten’ principles have been used many times, in both fiction and non-fiction, to create the illusion of extreme magnification of movement. As my interest in in digital environments, I would be able to cite many examples of the epic shot in which the camera pushes through the scene, but will focus on just two:
In summarising his approach to the DMP work on the opening sequence for Disney’s Maleficent, Dylan Cole explains that the scene is mocked up with simple 3D geometry and various renders of the wireframe are taken along the motion path of the shot camera and serve as the templates for the matte paintings.
Frames 00:00 – 00:45
https://www.artstation.com/artwork/nbnn6/ Accessed 19th Dec 2021
For the epic ‘Powers of Ten’ shot in Ghost in the Shell, Gottfried Eder describes a similar approach but with a combination of simple 3D geometry used as the basis for the projections and LIDAR scans for more distant elements.
Frames 00:00 – 00:25
https://www.artstation.com/artwork/Oo3Kyb/ Accessed 19th December 2021
The basis of the design for this project will be around a camera animation that pushes extensively through a built environment. The goal will be to investigate the techniques required to maintain image resolution using the general principles underpinning the ‘powers of ten’ concept. In designing the camera move, and general scene setup, I will look for ways to add complexities and variables that thoroughly test the techniques applied in project 2.
I will use a low-level camera angle, for at least part of the shot, to maximise the challenges of texture smearing associated with extreme perspectives. In addition to the extensive ‘push in’ camera movement, I will also add rotation to create coverage issues.
In many ways the actual style of the environment to be portrayed is arbitrary. However I have chosen to depict a ‘Steampunk’ town and there are several factors that influenced that choice, the first being that I had access to a high-quality geometry set. https://kitbash3d.com/products/steampunk/
I found the architecture to be interesting visually, a factor which cannot be overlooked given the amount of time spent facing the screen required to complete a project of this type. However, of more significance to the study, was the design of the buildings and the challenges likely to be presented to the camera projection process. This manifests in many areas but most evident is the non-uniform shape of the buildings; ornate topology, curves and protrusions that would cast and receive light, reflections and shadows and also create a lot of texture doubling scenarios, as would the street furniture such as lampposts and rails.
The second factor influencing the decision to use this asset set was that all the models were already textured. In an article in 3D World, Environment artist Tamas Gyerman discusses the merits of using a CG base for a matte painting. They argue that various structural elements can be tested in combinations and moved around until a satisfying composition is achieved. (2022, pp.48-53)
The premise here is that simple images of the respective buildings, or image sets, are rendered from the 3D application from each projection camera identified during the strategizing phase and these images are used to assemble the base projection. It is worthy of note that both examples of epic shots (provided above) utilised this technique, albeit in slightly different ways.
On my previous projects, I had built my own geometry to serve as the basis for camera projection but this would allow me to use renders from these computer-generated assets as the base for my projections and then add depth, greater realism, and artistic flair using a technique known as over-painting.
It seems that it would be straightforward enough to render the diffuse information for each image and then overpaint the light, shadows, and reflections once the camera projection system had been implemented. However, my intention is to build a lighting setup for the scene, the primary reason for this being to create challenges in blending projections together, where there is light change from one camera position to the next. A further benefit of this approach will be to leverage more from the geometry and the textures in terms of reference and possibly utilise the light, shadow and reflection AOV passes in the matte painting process.
I will be depicting a bright daytime scene and will therefore need a key light to provide a colour, intensity, and direction for the sun. However, I also want to look at the use of image-based lighting to add realism, especially to the reflective surfaces such as glass and metals.
The image (below) represents my first visualisation rendered from the 3D application following layout of the buildings and ground plane. A sky/mountains texture has been applied to a background object just to give an initial idea of colour and tonality for the scene.
Peter Holmes PhD Journal 