How to match a proposed building to a site photograph.

Decide what the photomontage is meant to establish.

A photomontage places a proposed building into a photograph of its surroundings. That sounds straightforward until a slightly different camera height, a shifted building origin or an unrecorded crop changes how large the proposal appears. A persuasive image and an auditable representation are not automatically the same deliverable.

Before opening the 3D model, write one sentence describing the decision this view should support: understanding a street entrance, comparing a proposed roofline with neighbouring buildings, or showing the relationship between a new volume and an existing courtyard. This determines the viewpoint, the background evidence and how much precision is necessary.

For planning or other formal submissions, ask which local methodology and evidence standard applies. This guide is a production checklist, not a substitute for a jurisdiction-specific verified-view protocol or a qualified surveyor.

Choose the viewpoint before choosing the rendering style.

Start from places a person can realistically stand and from the spatial relationship the audience needs to judge. A location across the street might reveal the whole façade; a view from the pedestrian route may show whether an entrance is visible. Do not quietly move a viewpoint just because a tree blocks the building. The obstruction may be part of the information.

Record a stable viewpoint ID, accessible position, camera height above the standing surface, approximate direction, date and time, and a marked location on a plan. For repeatable work, record surveyed coordinates and levels where the scope requires them. An approximate map pin is useful for finding the spot again, but it is not survey control.

Capture a photograph that can be matched, not merely admired.

Keep the original photograph, including its metadata and full frame. Record the actual lens or focal length used, whether digital zoom was applied, and whether the device corrected perspective or combined multiple frames. A phone's reported focal length may refer to a physical lens while its interface shows a 35 mm-equivalent value; these are not interchangeable without sensor information.

Where practical, use a level camera, a consistent height and a focal length suited to the agreed view. Avoid panorama stitching for a simple single-camera match unless the projection method is documented. Keep stable features visible: kerbs, window corners, parapets and other edges that can be compared against geometry or surveyed references.

Do not erase foreground trees, parked cars or fences from the source to make the proposal easier to see. If a clean background is separately produced for presentation, retain the unedited original and label the alteration. Confirm the photographer's permission for the intended use and check whether people or private information require additional care.

Match the virtual camera to the photograph.

Begin with the image projection, not the colour grade. Align the horizon and verticals, then work with focal length or field of view, sensor or film-gate settings, camera height, camera position and orientation. A virtual camera that uses the wrong sensor size can appear to match a stated focal length while producing a different field of view.

Use multiple stable references at different depths where available. Matching one roof corner is insufficient if the street edge and a distant façade disagree. Avoid stretching or warping the rendered building in post-production to conceal a camera mismatch. Record any lens-distortion correction or perspective correction applied to the background before alignment.

A quick overlay with contrasting edges can expose a mismatch before detailed rendering. If the original image has been cropped, store both the original and the crop dimensions. Changing the crop or using a different image aspect ratio after camera approval can silently invalidate the apparent match.

Connect the proposal to reliable site geometry.

A detailed BIM model is not automatically in the right real-world position. Verify the building origin, units, orientation, ground level and relationship to the site survey. Distinguish a local project coordinate system from geographic coordinates; importing a model at the wrong origin can produce a visually plausible but spatially false composite.

Check a few critical dimensions against the issued drawings: ridge or parapet height, building footprint, entrance level and separation from neighbouring structures. If the model contains design alternatives, isolate the approved revision and record its issue date. Where context geometry is approximate, identify it as such.

For a separate guide to units, origins, IFC/DWG exchange and revision packaging, see the BIM and CAD handover checklist. Camera matching is a second task: good source geometry is necessary, but not sufficient, for a trustworthy site composite.

Preserve foreground obstructions and depth.

The proposed building may sit behind an existing tree, wall or neighbouring structure. That foreground needs to occlude the rendered model correctly. Use careful masks or depth-based compositing, and check the transition at branches, railings and roof edges. A simple cut-out that places the new building in front of everything can exaggerate its visibility.

Separate what was photographed from what is proposed. Existing vegetation should not be silently replaced by an idealised landscape. Proposed planting can be shown, but its status and assumed growth should be clear. If the source image was taken in winter and the render adds full summer foliage, explain that the composite is illustrative rather than a direct seasonal record.

Pay attention to reflections and transparency: glass façades can reveal inconsistencies even when the building outline is correct. Do not use retouching to hide a known geometric conflict with an existing element.

Treat light and colour as interpretation, not measurement.

Match the photographed scene's general lighting direction, sky conditions and exposure so the inserted volume reads coherently. Shadows should respond to the proposal's orientation and the stated date and time when these are material to the decision. A pleasing golden-hour treatment may be appropriate for marketing, but it should not be confused with a time-specific daylight study.

Keep two distinct questions in review: does the building occupy the correct place and apparent scale, and does it look visually integrated? Resolve the first before the second. Colour grading, atmospheric haze and contrast can make an image feel natural while also making the proposal appear less prominent. Where visual prominence is being assessed, avoid unrecorded aesthetic manipulation.

If the brief requires a formal shadow, daylight or visual-impact assessment, commission the relevant specialist methodology. A photomontage alone does not establish compliance or quantify an impact.

Make the result reproducible and the assumptions visible.

Give the photograph, camera, model and final composite a common viewpoint ID and separate revision identifiers. Keep the unedited photograph and a record of alignment settings. State whether the image is a concept, marketing visual or formal assessed view, and who has checked the source geometry and camera.

Record limitations near the image or in its accompanying documentation, not only in an email thread. A statement such as “illustrative architectural visualisation; planting and finishes subject to design development” is appropriate when those elements are provisional, but it does not replace a required formal methodology.

Review geometry first, then approve the published image.

• The viewpoint ID and capture position are recorded and the photograph is the agreed original.
• Camera height, projection, crop and any lens corrections are documented.
• Model units, orientation, site origin and vertical levels have been checked against the latest issued references.
• Existing foreground obstructions are preserved and composited in the correct depth order.
• Roofline, footprint and key reference edges align with the agreed control evidence.
• Image edits that alter apparent scale or visibility are either absent or explicitly disclosed.
• Design revision and unresolved assumptions are visible to the reviewer.
• Any formal visual assessment follows its separately agreed local methodology and review process.
• Source-photo use rights, privacy considerations and approval for publication have been checked.
• The final exported file, not just a low-resolution preview, is approved with its viewpoint and revision ID.

Check the finished composite at its actual presentation size, including on a phone if it will appear on a website. Essential context must not disappear in a mobile crop. When reviewers request changes, use the rendering feedback checklist

For a project website, explain what the visual represents alongside verified project facts. The residential development content checklist