Built with Unity (AR Foundation)

Challenge

At the time of development, Unity’s AR features were still in an early, beta stage, with AR Foundation being relatively new and lacking comprehensive documentation. This posed several challenges:

Much of the AR workflow required trial and error, as official resources and best practices were limited.
Achieving reliable plane detection and accurate placement across a variety of mobile devices required deep experimentation and problem-solving.
The project had a platform-specific requirement for iOS, which introduced its own set of hurdles:
Apple’s development policies require a paid developer account.
An app targeting iOS typically needs to be compiled on macOS using Xcode, which can be a barrier for Windows-based developers.

Solution

To overcome these challenges and streamline the iOS development workflow, I implemented the following approaches:

With Unity’s AR Foundation still evolving, I gathered knowledge from community forums, developer blogs, and experimental testing. This iterative process helped me:

Implement plane detection for floors and surfaces.
Enable object anchoring and scaling, allowing users to place 3D models naturally within their space.
Optimize AR tracking performance, ensuring smooth and stable model rendering.

Instead of relying on macOS for iOS builds, I took advantage of Unity’s Cloud Build system:

This allowed me to build the app in the cloud, targeting the iOS platform without needing direct access to a Mac.
After configuring iOS build targets and connecting the Apple Developer account, I was able to deploy builds directly to test devices, accelerating the development and testing cycle.

Result

The project delivered a simple, effective AR previewer that worked smoothly on iOS devices. Users could easily place and interact with 3D models in their real environment, gaining a better understanding of scale, placement, and design. Additionally, by utilizing Unity Cloud Build, I streamlined the iOS development workflow, eliminating platform constraints and ensuring faster deployment and testing.

Surface Detection & Placement Areas

Challenge

Although Unity handled much of the plane detection and tracking, there were several challenges to overcome:

Identifying and visualizing available placement areas in a way that was intuitive and clear to the user.
Providing visual feedback to indicate where objects could be placed.
Differentiating between horizontal surfaces at different heights—such as distinguishing between the floor and a tabletop—so users could place objects logically and not accidentally position items incorrectly. For example, placing a chair on the floor should behave differently than placing a vase on a table. Without additional logic, all horizontal planes detected by AR Foundation were treated equally, making it difficult to establish context-sensitive placement zones.

Solution

To improve placement precision and user feedback, I implemented a surface classification and visualization system, layered on top of Unity’s AR Foundation plane detection.

Detected surfaces were visually represented in the AR scene:

Yellow dots were used to mark detected planar surfaces, providing immediate feedback that AR Foundation had found a trackable area.
A blue strip or border was displayed around valid placement zones, clearly indicating where objects could be placed.
This helped guide users to align their device correctly and understand where they could interact with the environment.

To distinguish between different surface types (e.g., a floor vs. a table), I implemented logic based on the height of the detected planes relative to the user’s device and the environment:

Planes at ground level were categorized as floor surfaces.
Planes detected at elevated heights were categorized as tables or countertops. This allowed the system to enforce context-aware placement rules, preventing users from accidentally placing large furniture items on small surfaces like tables, and ensuring smaller objects like decorations could be correctly placed on elevated planes.

Once a surface was classified, placement constraints were applied based on the object type.

For example, sofas and chairs were restricted to floor surfaces, while vases and lamps could be placed on tables.
Users were given clear visual cues when objects were in valid positions, and snap-to-plane functionality ensured objects stayed anchored to the chosen surface during interaction.

Result

This system delivered an intuitive and reliable AR placement experience, giving users precise control over where objects could be positioned within their space. The combination of visual feedback, surface classification, and placement logic helped create a smooth and frustration-free interaction, enhancing both the usability and realism of the AR previewer.

Distance Placement & Object Anchoring

Challenge

By default, surface detection in AR Foundation continuously updates in real time, adjusting planes as the device moves. While useful for dynamic environments, this behavior made it difficult to hold an object steady at a fixed distance:

Objects would dangle, float, or drift, as the surface detection recalculated, causing them to shift unexpectedly.
There was no out-of-the-box solution in AR Foundation for locking an object’s position in space, especially when plane tracking was unstable or constantly updating.
Users needed a way to place an object “in space”—whether on a detected surface or at an arbitrary distance—and have it stay precisely where they left it, regardless of tracking updates.

Solution

To solve this, I developed a distance placement and anchoring system, providing users with the ability to position objects at a fixed distance from the camera and anchor them reliably in world space.

Implemented a raycast system from the device’s camera into the scene, allowing users to select a point in space at a specific distance.

The user could then** place an object at that point**, using UI controls to fine-tune the distance, rotation, and orientation.
This gave users precise control over object placement, independent of detected planes.

Once placed, objects were anchored using AR Foundation’s anchor system, ensuring they stayed in world space and did not shift with ongoing surface detection recalculations.

This stabilized objects and prevented them from floating or drifting, even if plane detection changed or was lost.

After placement, I disabled real-time updates for the plane associated with the object to avoid unnecessary recalculations.

This ensured the object would remain static, providing a smooth and consistent user experience, especially in scenarios where users wanted to view the object from multiple angles without it moving.

Added visual cues and UI indicators to confirm when an object was successfully anchored.

Provided users with manual adjustments after initial placement, allowing them to fine-tune the position or release the object to re-place it elsewhere.

Result

This feature significantly enhanced the flexibility and usability of the AR previewer. Users could confidently place objects at a distance, whether on a floor, table, or floating in mid-air, and trust that they would remain stable and correctly positioned. The system delivered a smooth and intuitive interaction, improving both the practicality of the AR tool and the overall user experience.

Case study

AR Model Previewer

AR model previewer is a IOS application that allow you to load any 3d model and view them in augmented reality. Application got such features:

Unity
IOS
AR
C#

AR Model Previewer

Start of development:: 2016
I am in the project:: 2016

Contribution to project

Loading models of format .fbx and .obj from internet
Different transformation of models, such as scaling, zooming, rotating etc
Placing models on different surface

Case impression

The task to create this up was a client’s idea. He wanted to create a model previewer for himself. He also wanted it to use unity and all its capabilities. The format of the 3d object was vast but cause of time limitation and inaccessibility of assets was limited to 2.

A new technology - a new knowledge

The project was created at that time when Unity AR was still experimental feature, and a lots of things needed to be learned a hard way

A usage of new Technology stackDue to not having an experience of working with AR and IOS development, lots of things needed to be learnedRead more

Finding surface to landDue to experimental functionality, surface finding was complicatedRead more

Drop the anchorLogic of determining distance and where anchor should goRead more

And the Client was satisfied with the results

Further changes and improvements were in discussion but never came to fruition

All projectsLet’s talk