A centuries-old stone arch. A chrome car part. A leather chair. Each one photographed from 200 angles, uploaded, and reconstructed as a 3D model that carries every crack, surface detail, and material imperfection. No laser scanner. No survey equipment. Just a camera and the right software. Photogrammetry turns photographs into geometry — and it's now fast enough and accurate enough to be a standard tool in commercial CGI production.
How Does Photogrammetry Work?
Take 200 photographs of a subject from every angle. Make sure each photo overlaps the next by at least 60 percent. Keep the lighting consistent throughout.
Now give those images to photogrammetry software.
The software identifies matching features across images (a scratch in the surface, a corner detail, a specific texture cluster) and uses those matches to triangulate where each camera was positioned. From hundreds of camera positions, it builds a point cloud. From the point cloud, a mesh. From the photographs themselves, a texture map that carries the real colour and surface appearance of the original subject.
The result is a 3D model that captures what was actually there — not an artist's interpretation of what it might look like, but real surface geometry and appearance. That's what separates photogrammetry from manually authored 3D visualization models, where materials are approximated by artists.
Where Is Photogrammetry Applied?
Drone over a construction site. 300 photographs. A point cloud accurate to centimetres, showing every structure, every grade change, every deviation from plan. No surveyor required.
That's photogrammetry in architecture and construction, but it's one use case among many.
In archaeology and heritage preservation, fragile or at-risk objects get photographed in place and reconstructed as permanent digital records. A thousand-year-old artefact that can't survive repeated handling gets scanned once and examined indefinitely.
In film VFX and game development, photogrammetry is the standard method for capturing photoreal environments and props. Scanning real-world materials captures the micro-geometry, wear patterns, and imperfection that would take artists weeks to recreate by hand.
In product work, scanned fabric, leather, stone, and wood carry accurate surface normals and reflectance data. Used in a render, those materials behave the way the real thing does under light — which is a core reason that high-quality CGI renders are increasingly indistinguishable from photography.
How Does Photogrammetry Compare to LiDAR?
LiDAR fires laser pulses at a surface and measures how long they take to return. The result is precise geometric data. But no colour: point clouds from LiDAR have to be colourised separately, usually from photography taken at the same time.
LiDAR works in the dark. It works on featureless surfaces — a blank white wall, a calm lake — where photogrammetry fails. It captures structural geometry with survey-grade precision.
Photogrammetry's advantage is accessibility. You need a camera — increasingly just a high-quality smartphone — and processing software. It captures colour and texture in the same pass as geometry. And modern software makes the workflow fast enough for commercial production timelines.
Most serious studios use both. LiDAR for structure and precision. Photogrammetry for surface colour and fine texture detail. The best captures combine them.
What Limitations Does Photogrammetry Have?
Chrome tap. Glass bottle. White sphere on a white background. All of them will break photogrammetry.
The algorithm works by finding matching features across images. Reflective, transparent, and uniformly coloured surfaces don't have stable features. The reflections change between shots. The clear glass shows a different background at each angle. The white sphere looks identical from every position. The reconstruction either fails or produces garbage geometry.
Movement is fatal. If the subject shifts even slightly between shots, the model corrupts. That rules out anything living, anything caught in wind, anything that vibrates. Outdoor shoots are also sensitive to changing cloud cover: if the lighting shifts mid-session, the shadows move across the image set and reconstruction quality drops.
Photogrammetry works best on static, matte, feature-rich subjects: buildings, landscapes, aged materials, matte-finish products, stone. Outside those conditions, expect to supplement it with other methods.
How Does Digital Bunch Use Photogrammetry?
We use photogrammetry as part of our 3D product visualization and environment capture workflow — scanning physical products, materials, and spaces to build assets that carry authentic surface data rather than artist-approximated textures.
For premium and luxury product work, that distinction matters. When a client's brand is built on the quality of a material (the way light moves through a woven textile, the sheen on powder-coated aluminium) we want to capture that, not approximate it.
We did exactly that for Brown Jordan. Scanning their furniture materials meant the rendered imagery accurately represents how the pieces look in person, the tactile quality comes through even in a still image. That accuracy closes the expectation gap between what a customer sees online and what they receive.
It's also part of why, when we get materials right, our renders are increasingly indistinguishable from photography.