3D Models

In the MRI virtual tour, we superimposed very simple 3D objects on the 360° background. Where did they come from?

Use a 3D modeling tool

It is possible to build such objects in tools like Blender (free) and A-Frame (also free). In the case of the MRI virtual tour, we created objects directly in A-frame. Several 3D modeling tools (Blender, Netfabb, Maya 2022, Rhino6) are available to U of A students and staff through Apporto. These allow you to build, view, and edit 3D models. Such tools are developing quickly, shifting between commercial and open-source offerings, and fast becoming easier to use (e.g., see the online brain2print which can create a printable 3D model from an MRI scan).

Formats and Modeling tools

Not all formats work with all modeling tools. Some tools allow you to convert among formats, but not all of them. And, new formats are being developed.

Download from a repository

More and more repositories of 3D models (many free) exist, and, depending on the model format are most suitable for 3D printing (e.g. STL and OBJ) and/or being added to VR environments (e.g., GLB, FBX, PLY).

Capture a Model

Modern mobile phones offer apps (e.g. the free scaniverse) that facilitate capturing real world objects and exporting them as 3D models that can be used in VR. There are several technologies (as of March 2025) for capturing a model: Photogrammetry, LiDAR, and splatting.

Photogrammetry The earliest attempts to build 3D models required clever recombination of regular photographs from many different angles. Newer technologies (e.g. LiDAR and splatting) have different benefits.
LiDAR short for "light (laser) detection and ranging" measures distances by bouncing lasers off objects and calculating the time it takes for the light to return. While long used for large-scale mapping, LiDAR can also scan smaller objects. LiDAR has been available in PRO iPhones since the iPhone 12, but iPhone LiDAR is less powerful than professional LiDAR systems, which offer higher precision (higher point density), greater range, and faster data capture. FBX is a good format for LiDAR scans.
Gaussian Splatting is the newest of the three techniques mentioned here. It is better than LiDAR at handling reflective and transparent surfaces. It is faster and more photorealistic than photogrammetry. Gaussian splatting uses millions of tiny ellipsoids (a.k.a. splats) to create a scene: these splats are assigned a position, transparency color and size in a projected 3D point cloud. Using a tool like scaniverse on your phone, you can capture and export splats (PLY is a good format for splats) You can see an example splat on antimatter 15 (embedded below). Splats are plagued with clouds (floaters) which can be removed with editing software like Supersplat. As of Spring 2025, splatting is promising, but there is a quality gulf between consumer-level captures (e.g., with an iphone) and professional capture devices.
Holograms As of Spring 2025, I don't know of any consumer-level hologram devices, but Bryan Carter at the U of A has been experimenting with broadcasting live holograms (using an app called formavision), so it is possible, but still bleeding edge.

You can view an example splat in the interactive "antimatter" site:

DSP 4 (5 pts): 3D models