For those who don’t know the tool: DroneSiteBuilder is a script for Autodesk 3ds Max that reads drone photo metadata to assemble terrain, map context, and matched cameras, bundling GeoBuild and CameraMatch and supporting handoff to Blender, SketchUp, and Autodesk Maya.

What DroneSiteBuilder is trying to fix

DroneSiteBuilder Enterprise turns a folder of drone photos into a full site context scene inside the 3ds Max ecosystem: terrain, aerial drape, buildings, roads, land cover features, points of interest, and one matched camera per photo.

The pipeline reads XMP metadata per image and uses it to place helpers at ground level and flight altitude, then creates a renderer-aware camera per photo and refines orientation using vanishing point detection. The tool targets 3ds Max 2020 or newer on Windows and uses Python 3.9 through 3.12 with NumPy and OpenCV, plus optional modules for the Google path.

First use, abbreviated

Install a standalone Python build and make sure Add Python to PATH is enabled during install, because the tool does not rely on the 3ds Max built-in Python. Place DroneSiteBuilder_Enterprise.ms and vp_detector.py in the same folder and do not install under Program Files, since Windows blocks writes there.

Launch 3ds Max and load the script via drag-and-drop into a viewport or via Scripting > Run Script. Paste your licence key into the activation dialogue and activate, with support for an Enterprise key format and existing CameraMatch Studio keys. After activation the key caches locally for thirty days and the rollout opens.

For the first build, keep original DNG or JPEG drone files in a single folder, keep filenames ending in digits, and keep an internet connection available for the auto fetch steps. Pick a terrain DEM source, keep NOAA GEOID18 enabled for continental US sites and disabled outside the US, then run Process Folder of Drone Photos and accept the summary dialogue. Inspect the result by comparing a matched camera view against the source photo.

The new checkbox: Google Photorealistic 3D Tiles

Enterprise v1.2 adds an opt in mode that fetches Google Photorealistic 3D Tiles and imports them into 3ds Max as a single merged glb, covering the photo bounding box. The intention is simple: pull a photogrammetric backdrop mesh with terrain, textures, and buildings baked into one continuous model, using the same dataset shown in Google Earth.

The Google path requires a Google Maps Platform project with billing enabled and the Map Tiles API enabled, plus a saved API key in the rollout. The manual describes 1,000 root tileset requests per month free and a per 1,000 price after that, with one DroneSiteBuilder run treated as one root tileset request. If the fetch fails, the tool reports common causes like missing API enablement, low coverage, EEA billed accounts being blocked for this endpoint, missing Python modules, and older 3ds Max builds needing a glTF importer.

This is also where the gaudrrails get real: the Google import produces mesh only content with baked textures, so you do not get separate terrain, buildings, roads, and features objects to isolate, re material, or hide by category.

The standard path is still the controllable one

If you leave the Google option off, the pipeline builds a layered context using public services. Terrain comes from USGS 3DEP in the US or OpenTopography worldwide, aerial imagery comes from Esri World Imagery as a single bounding box JPEG, and vector context comes from OpenStreetMap via Overpass.

The scene builds under a single root helper so you can move, rotate, or scale the entire site as one unit, while cameras intentionally sit outside that parent chain to support camera transplant between scenes. If buildings and roads fail to appear on a run, the manual provides a manual Overpass refetch button that rebuilds buildings, roads, and land features without disturbing terrain, drape, helpers, or cameras.

OpenStreetMap accuracy varies by location and that the resulting geometry accuracy for this context layer sits in the one to six meter range. It contrasts that with paid city model providers such as AccuCities when millimeter level detail matters.

Export and handoff from 3ds Max

DroneSiteBuilder does not change how 3ds Max exports, but it does influence what you export. The standard path creates distinct terrain, buildings, roads, and feature objects you can selectively include in deliverables, while the Google path imports a tile based mesh group that is already baked.

For broad DCC and engine handoffs, 3ds Max includes a built in exporter for glTF and supports export workflows for USD, with export options available through File export and USD and USDZ format selection when the USD tooling is installed. For USD pipelines, the official OpenUSD plugin for 3ds Max is published on GitHub, and Autodesk documentation describes supported USD content types including meshes and cameras.

If your downstream is realtime, glTF targets web and interactive use cases, and you can also push common interchange paths like FBX into engines such as Unity and Epic Unreal Engine, depending on your studio conventions. For camera and layout continuity between scenes, the bundle includes a camera transplant workflow that exports camera data to JSON and recreates it with full precision in another scene.

Practical reality check

This toolkit aims to comperss a traditional drone to 3D context workflow into a short unattended run. The promise stands or falls on your input photos keeping XMP metadata intact, on your site area having decent public map coverage, and on you verifying the result against the plate before trusting it.

We don’t know the price yet, but if you already own the camera match tool in the Studio version, you get it for free, apparently. Quoting: Existing CameraMatch Studio licence keys activate the full DroneSiteBuilder bundle. Studio holders do not pay twice. 

Try both context modes on a test job: the standard layered build when you need controllable components and re-lighting freedom, the Google mesh mode when you need a fast photogrammetry backdrop and can live with mesh-only content and licensing constraints.

https://jamiecardoso-mentalray.blogspot.com/2026/03/dronesitebuilder-enterprise-v12-one.html

https://jamiecardoso.gumroad.com/

The post DroneSiteBuilder adds Google 3D Tiles option first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

Photomontage lives or dies on perspective. If the camera does not match the photograph, everything downstream becomes a negotiation with angles, scale, and taste. CameraMatch AutoSetup v6 targets that first domino by generating an aligned 3D camera from a single photo, then setting up supporting scene items that normally take a few manual passes.

What CameraMatch AutoSetup v6 does

CameraMatch AutoSetup v6 is an automated camera-matching tool for photomontage work. It detects vanishing points and lines in photographs and uses that information to create a 3D camera aligned to the plate, aimed at architectural visualisation and general 3D scene setup.

Once the camera alignment happens, the tool also covers a set of setup chores that tend to get skipped until they break a comp. It can calculate and display the horizon line. It can create a physical camera with an accurate field of view. It can match the render output resolution to the background image resolution. It can set a viewport background overlay that shows vanishing point lines and the horizon. It can also export cameras using the FBX format so the camera can travel to other applications.

The overall pitch reads like a checklist of common photomontage necessities, bundled into a single automated pass. That pitch also includes marketing claims about ease and straightforwardness, which remain marketing claims.

Host apps today, and what is promised next

The currently available versions work with 3ds Max and Blender. A version for Maya is described as planned, and additional host versions are described as coming next week.

Camera matching tools often stumble on the unglamorous details of units and output sizing, so it is notable that the material calls out unit setup dialogs and scene unit contexts as part of the broader setup conversation. That framing keeps the tool squarely in production reality rather than demo land.

Why this matters for post and VFX

A good plate match is not just about placing geometry. It is about making every downstream decision cheaper. When the camera and horizon are consistent, the layout behaves. Lighting behaves. Shadow direction behaves. Reflections behave. Even simple matte work becomes more predictable because perspective lines agree across layers.

For photomontage-heavy teams, the time sink is usually not the final render. It is the first 20 minutes of making sure the camera is believable, the frame matches the plate, and the viewport does not lie. CameraMatch AutoSetup v6 positions itself as a shortcut through exactly that part of the day, with its strongest promise being camera alignment plus the surrounding setup steps in one go. If you have ever watched a project drift because two artists picked two slightly different horizons, you already know whcih argument this tool wants to prevent.

Availability, docs, and pricing reality

Documentation exists as a free manual hosted on Google Drive. The manual is covering the script for 3ds Max, Blender, and Maya. Distribution is offered through two storefront channels: Gumroad and Patreon. A one off purchase option for the Gumroad of 45USD, or through a

The download description also references a full script with files for 3ds Max and Blender, and it mentions direct contact as an option. The complete contents of the downloadable package are not listed in the provided material beyond that.

Practical notes before you point it at a deadline

Automation shines when you can trust the inputs. This tool relies on reading vanishing points and lines from photographs, so images with heavy lens distortion, weak straight line cues, or ambiguous architecture may still need human judgment and cleanup.

As always, new tools and innovations should be tested before use in production. Run it against known plates, compare results to your established camera match method, and confirm the exported camera behaves as expected in your pipeline before you bet a delivery on it. And yes, it can be both fast and correct, but only after teh boring validation pass that everyone wishes they could skip.

https://jamiecardoso.gumroad.com/l/boqjtv?layout=profile

The post CameraMatch AutoSetup v6 speeds photo matches first appeared on DIGITAL PRODUCTION and was written by Bela Beier.