Arindam Shiva Tripathi
← All work

project - M.S. Thesis

Reframed

A Quest 3 passthrough AR experience where classic artworks animate beyond their frames and become a treasure hunt.

Figure 1. Reframed transforms physical artworks into spatial mixed-reality experiences that animate and extend beyond the frame.
Role
Solo Developer & Designer
Team / Org
Solo; thesis advised by Prof. Ming Lin and Dr. Beth Bonsignore
Timeline
Mar 2026 – Present
Skills / Tools
  • Passthrough AR
  • Image tracking
  • Pose estimation
  • Playable experience

Overview

Developed independently during UMD’s New Works Incubator, Reframed recognizes printed artworks on a wall and anchors animated digital content to each one using on-device computer vision. A hidden-object search across five artworks turns it into a playable experience rather than a technical demo. It went from prototype to a public showcase and is the basis for my master’s thesis.

The experience

Figure 2. The virtual layer first aligns with the physical composition before extending beyond it.

The augmentation begins with close registration between the digital and physical imagery. This creates visual continuity between the original artwork and its mixed-reality extension, helping the viewer perceive the effect as emerging from the painting itself.

Figure 3. The Great Wave transitions from a static print into a layered spatial scene.

The physical print remains present as the foundation of the experience while digital elements animate and begin to occupy additional depth. The original work is not replaced; it becomes part of the final mixed-reality composition.

Figure 4. Elements of the painting separate from the flat image and emerge into the viewer’s space.

In this sequence, elements from Wheat Field with Cypresses begin to separate from the two-dimensional composition and extend outward into the room. The frame becomes a threshold between the original image and a spatial version of the scene.

Figure 5. Moving around the work reveals the depth between the physical painting and its virtual layers.

From the front, the physical and digital imagery can appear closely aligned. As the viewer moves sideways, parallax reveals the separation between virtual layers, making the three-dimensional construction of the scene visible through movement alone.

In the room

Arindam guides a visitor wearing a Quest 3 headset past framed prints of the five artworks on easels.
Figure 6. The installation at NextNOW Fest, with a visitor starting the hunt.

Reframed ran as a walk-up experience over two days and drew 50+ visitors. I hosted the station, getting each person into the headset and through their first artwork.

Arindam points a headset-wearing visitor toward a printed marker on the floor.
Figure 7. Every session starts at a printed floor marker, which the headset recognizes before the hunt begins.
Arindam watches a visitor in a Quest 3 headset move between the easels.
Figure 8. Watching a visitor work through the hunt.
The five printed artworks on easels in a row, with a small floor marker in front.
Figure 9. The five registered artworks, with the floor marker in front.
A tester in a Quest 3 headset looks at a print of The Great Wave taped to a wall.
Figure 10. Testing recognition on a single print on a wall.

What I did

  • Built a Quest 3 passthrough AR system that recognizes physical printed artworks and anchors animated content to each one, using on-device ORB feature matching, homography, and 6-DoF solvePnP pose estimation.

  • Wove a hidden-object search across five registered artworks so the piece works as a playable experience, not just a technical demo.

  • Took it from prototype to a public showcase at UMD’s NextNOW Fest, drawing 50+ visitors over two days.

  • Basis for my master’s thesis, advised by Prof. Ming Lin (VR/spatial computing) and Dr. Beth Bonsignore (HCI).

Outcomes

50+
Showcase visitors
5
Registered artworks