Initial considerations drew heavily on the findings from the period of reflection on my practice following completion of the first project in this study. The overarching issue from this was that, whilst the first project served to ground ‘camera projection’ as the most effective approach for applying textures to geometric objects in 3D space, the limited movement of the shot camera left unanswered questions in relation to this method. I was keen to establish whether, and/or to what extent, this approach would hold up when subjected to conditions where the camera was much more expansive in terms of lateral movement and rotation.
The Proposition
My concept sketch for this project (below) depicts a Victorian street scene. My intention is to present the final piece as a night shot with atmospheric lighting and volumetric fog.
The unfortunate consequence of these design considerations is that the topology of the scene is not especially clear in the sketch and some explanation is therefore necessary. The sketch shows the scene from the starting position of the camera. In the immediate foreground on the left side is a cobbled square with imposing steps and decorative stone pedestal pillar. I imagined these fronting a large public building such as a Town Hall or Library, which would be just out of shot on the extreme left. Adjacent to the square is a point at which two roads cross with buildings on each of the three remaining corners. My intention was to present each building with different styles whilst retaining characteristics that ground them in the period. To facilitate this, I gathered an extensive library of reference imagery, which served as useful research material on architectural features of the period, but also to provide texture sources for the photo-bash stage of the matte painting. Beyond the buildings I planned to add further structures to dress the shot. These will add scale by extending the scene and giving scope for distant light sources and volumetric effects. However, as these are peripheral elements, are never seen in detail and only in silhouette, the intention is to use stock CG assets rather than incorporate these into the matte painting and camera projection workflow. Consequently, these elements are peripheral to the study and not subject to significant discussion. I also planned to add street furniture such as streetlamps and bollards but was undecided at the outset whether to create these as CGI elements or to incorporate these into the camera projection.
In addition to the concept sketch, I had also constructed a simple 3D scene to facilitate the design the animation for my shot camera. The geometric objects used were simple cubes for the three primary buildings and a planar object for the ground. All were in the correct position, orientation and general scale and therefore suitable base objects for the modelling stage.
The video (below) shows the camera movement from three points in time over the duration of the shot. Simple geometric objects are used to represent the three primary buildings in the shot and a planar surface is used for the ground. The first (leftmost) image represents the starting position and therefore compositionally similar to the concept sketch. The third (rightmost) image represents the ending position of the camera, which is looking down the street and thereby presenting a view of the building façades that are not part of the concept sketch and will present coverage that will need to be added using additional projections. The second (middle) image simply represents a point midway along the camera animation, primarily to illustrate the motion path of the camera. This moves right and upwards, whilst rotating on both the hip and pitch of its axis to maintain framing.
there was no loss of resolution in the affected textures.
The composition of the scene and the design of the camera movement presents angled views of the façades and the ground surfaces that are quite extreme and the perspective of these planar surfaces change extensively over the course of the shot, factors that were built into the design to introduce coverage issues and projection artefacts such as texture smearing and doubling. A further design consideration is potential loss of quality in the projected textures, which impacted the decision to include the third (rightmost) building. Because the camera moves close to this building at the end of the shot, this will affect the resolution of the projected textures on its surface.
The composition of the scene and the design of the camera movement presents angled views of the façades and the ground surfaces that are quite extreme and the perspective of these planar surfaces change extensively over the course of the shot, factors that were built into the design to introduce coverage issues and projection artefacts such as texture smearing and doubling. A further design consideration is potential loss of quality in the projected textures, which impacted the decision to include the third (rightmost) building. Because the camera moves close to this building at the end of the shot, this will affect the resolution of the projected textures on its surface.
This project was framed to address several unresolved questions using the following set of localised objectives:
- Develop a workflow and software pipeline for efficient and effective shot realisation
- Explore methods of overcoming loss of texture coverage
- Investigate methods for identifying and fixing instances of texture smearing on textures projected onto surfaces that are subject to extreme perspective distortion
- Investigate methods for identifying and fixing instances of texture doubling on textures projected onto overlapping geometric surfaces
- Investigate methods for identifying and resolving the effects of image degradation arising from proximity of camera to the projection surface
Throughout this project I utilise the framework and instruments defined in my methodology. Reflective practice is used at every iterative stage and documented in my blog as a means of justifying the approaches taken and providing supplementary demonstrations from within the various software via annotated screen capture, and intermediary renders to provide clarification and context. For objective 1, I deploy the principles of ‘total production workflow’ to produce a clearly defined roadmap for completing the work, both in terms of the processes involved at each stage, and the software needed to realise these processes. This triggered investigation and utilisation of a previously unused methodology for approaching the digital painting aspects of the project. For objective 2 I build a diagnostic rig for testing and optimising the camera projections and use this to explore two different approaches for dealing with missing coverage. For objectives 3 and 4 I use the diagnostic rig to identify texture smearing and doubling and devise strategies for overcoming the degradation caused to the scene. Finally for objective 5 I again use the diagnostic rig to identify potential loss of resolution and implanted method to ensure
Peter Holmes PhD Journal 