Here is a copy of the method statement we supply with our verified views for planning. It describes, in detail, how verified views are created. It's a process we developed in the early nineties and is always evolving, though the core principles of working with surveyed photography for perspective accuracy remains the same.
It is not common for visualisers to give this information away for free. However when we are asked to assess the methodologies of others it is surprising how often they are either vague, have become marketing documents or employ clever-sounding terminology to make the process seem more 'expert' than it is. We are sharing ours so we can move toward standardising the production of visuals for planning applications and thereby create a consistent standard of quality.
If any part of this methodology seems unclear or outmoded I would love to hear about it, so we can continue to make the process even simpler and more transparent.
RGCGI PLANNING VIEWS METHODOLOGY
The computer-generated architectural photomontages used in
planning applications are called ‘Verified Views’. These computer renders,
blended into site photography, can be independently verified as accurate by following
each part of the view’s production process, as documented in this methodology.
This methodology ensures
every part of the image creation process is accurate, clear and reproducible.
It complies with Appendix C of The London View Management Framework Supplementary
Planning Guidance March 2012 as well as The Landscape Institute’s Advice
Appendix C of the LVMF SPG classifies verified views as distinct types of ‘Accurate Visual Representations’, or AVRs. Types of AVR depend on the rendering treatment of the proposal and can be applied as appropriate in the planning application.
AVR0: A silhouette
AVR1: A wire outline
AVR2: A lit block model, untextured
AVR3: A lit and detailed model, textured with materials
There are three main stages to creating all AVRs:
- Obtain surveyed photography from agreed viewpoints.
- Simulate real world conditions with digital information
- Combine simulation with photography.
1. OBTAIN SURVEYED PHOTOGRAPHY
Accurate and consistent site information is crucial, from the resolution of photography to the fidelity of the survey, but the content of the view must also be carefully chosen. The nature of the viewpoint, weather conditions and the time of year can lead to misleading impressions of the development
To determine appropriate
viewpoints, RGCGI use a combination of site recce photography and rendered
views from a local area digital model. Sun lighting can also be simulated
digitally to ensure the photographer is sent to site at an appropriate time of day.
The resulting material is evaluated by the design team and presented to
planners to ensure it addresses their concerns before detailed work is
The photography requirement for an AVR is:
- High resolution (5000x4000 pixels minimum) to allow detailed features to be read clearly during survey and alignment.
- Of a high colour depth to aid the accurate matching of light and shade during the digital rendering process
- A simple two-point perspective, without the convergence of verticals caused by tilting the camera. If necessary, a ‘shift lens’ is used to lower the horizon and allow more into the top of the frame without tilting the camera.
- A natural view, with minimal perspective stretching the periphery of the scene, so architectural proportions and scenic depth are as expected.
- Thoroughly documented so the camera position, lens and time of day can be matched in the digital simulation, and for future verification
It is possible to achieve all this with almost any modern digital camera. However, most commonly available equipment will trade one aspect to meet others. Resolution may be high, but colour depth poor. They may offer adequate resolution and colour depth but no offer no attachments for a standard view angle shift lens. Also, some digital cameras perform in-camera image “improvements” that are not classifiable or repeatable by other cameras when endeavouring to verify the photograph.
For this reason, RGCGI use a high-end medium-format camera such as the Alpa Max with a Leaf Altus 75 digital back. This is coupled with a 47mm calibrated architectural shift lens. The resulting resolution of the images is 6658x4984 and the colours are high dynamic range. After a small amount of cropping, the lens’s horizontal angle of view is broadly equivalent to a standard 50mm lens or 40 to 45-degrees. This represents the least amount of perspective distortion.
Where the subject is far from the viewpoint, a 150mm zoom lens is used alongside the standard lens. This creates a roughly 13-degree view angle to mimic the eye’s ability to focus on detail.
Where the scene is too wide to fit within a single shot, and a contextual view is still required, a medium format, wide angle 35mm lens is used to provide a view angle of about 67-degrees. If this is not wide enough then a panorama is stitched together from multiple 47mm photographs. In this case the final image is marked clearly with where each tile of the panorama sits
Our photographer uses a levelling tripod to ensure the camera is level on all axes and positions it 1.6m above the ground, the average human eye-height. The camera is always located on a pavement or pathway accessible to pedestrians and its position is marked with a Hilti-nail or spray ready for the surveyors. The amount of shift and shutter settings are also recorded along with the date and time of day.
Surveyors visit each viewpoint and perform a differential GPS survey of the camera position as well as a minimum of 12 clearly visible features in the view. These include points of high contrast in the foreground, middle ground and background. They avoid the edges of the photographic frame where there may be optical distortion in the lens that cannot be replicated in a virtual camera later. They also provide a surveyed position for the horizon so any shift used can be factored into later perspective alignment.
2. SIMULATE REAL WORLD CONDITIONS
The raw digital photography is adjusted in Adobe Photoshop to:
- Offer a balanced and natural exposure.
- Check the horizon is precisely level.
- Re-position a lens-shifted horizon back to the centre of the frame.
Surveyed information is supplied as a 3D CAD file for each photograph. The data is organised with reference to the Ordnance Survey’s National Grid and comprises the camera’s 3D ground position, and a set of 3D points of the visible features determined in the photography.
Using a 3D rendering application such as Autodesk 3DS Max, a virtual camera is positioned at the surveyed position. It is offset vertically by the recorded height of the tripod and its lens characteristics set to match the recorded lens. The digital photograph is used as a background in the simulation scene. Looking through the virtual camera, the 3D survey points are superimposed onto the background photograph. The scene is rendered at the resolution of the photograph, and the alignments of survey points to features checked. If necessary, minor adjustments are performed before rendering again. This iterative process continues until all points match their corresponding features.
A 3D digital model is either created from architect’s information or imported from their CAD or BIM format. The modelling detail is dependent on what type of AVR is to be produced, however the model is always accurately positioned according to OS both in plan and height above Ordnance Survey Datum. If the model was supplied by the architects, it is positioned and then cross-checked against the architects’ plans and sections.
If necessary, a local area digital model is included in the simulation to provide information on how the scheme is hidden by its context. In the case of AVR3 images this also increases the accuracy of shading and reflections.
A simulated daylight system is created to match the date and time of day as recorded on site. The virtual camera’s shutter and exposure settings are also set to match the real camera, and the 3D model is rendered onto the photograph. The resulting base image can be used for AVR0, AVR1 and AVR2 type images
For AVR3 renders, the detailed digital model is textured with simulations of the proposal’s materials. Digitally simulated materials can react in ways that real materials can’t; reflecting more light than is possible in the real world, for example. Care is taken to ensure that materials are physically based and react to the daylighting realistically. This ensures the correct levels of brightness and shade both on the proposal’s surface as well as its interior as viewed through glazing.
3. COMBINE SIMULATION WITH PHOTOGRAPHY
The result of processes 1 and 2 create a simulation of the proposal: a rendered 3D model that is correctly positioned in the background photograph. Using image editing software such as Adobe Photoshop the rendered model and the photograph are combined. At this point the render can be masked so foreground elements such as trees and other buildings obscure the proposal correctly.
In the case of AVR0 and AVR1 views the rendered model is used as a base for graphical treatments such as silhouetting or adding lines to describe the building’s massing. Where the building is partially obscured by greenery – or totally obscured by context – dotted lines are used to illustrate the hidden parts.
In the case of AVR3 views atmospherics such as mist can also be simulated, paying close attention to how the photographic context is affected so that the proposal is neither too hidden nor too prominent in relation to its environs.
The resulting image is distributed to the design team and client for comment. Necessary adjustments are made to the 3D model, the rendering or image editing. Sometimes it might be determined that another photograph needs to be taken in a different season or lighting condition. Once the team is confident that the image accurately represents the design it is prepared for printing.
An information border is added to the image, describing the photograph’s angle of view and the ‘optical axis’ or centre of the view. This is to highlight the forty-degree portion of least distortion. This is most important on a wide-angle view where subsequent analysis might require cropping to avoid distorted foreground elements distracting the evaluation, or otherwise creating a misleading depth to the photograph. The information border also includes a ‘range line’ describing the proposal’s position in the image, even if it is mostly occluded in the view.
Each proposed image is laid out in a document next to its corresponding existing image. Also included is a map of the viewpoint, a photograph of the camera position and its surveyed OS position.
Unbuilt buildings already approved for planning and that appear on a view are rendered as AVR1 wirelines and superimposed on the proposed image. This creates a third “cumulative” image set alongside the existing and proposed.
Where required, RGCGI will calculate the appropriate print size and viewing distance for a given view to match the real world. In the case of a public exhibition for example, A1 prints of standard lens photomontages can be best assessed from around 1.4m away.