Thesis. NeRF-⁠based 3D reconstruction for second-⁠hand e-⁠commerce that lifts purchase intent.

3D Reconstruction, AI, Research, UX

2D plant image — Qualitative comparison of the reconstruction applications under the same conditions.

Polycam reconstruction — Qualitative comparison of the reconstruction applications under the same conditions.

2D binary image — 2D image

NeRF binary reconstruction — 2D image

Reconstruction process flow — Instant-NGP process

Professional camera output — Professional camera

iPhone 13 output — Professional camera

Initial camera positions — Camera-position improvement process.

Final camera positions — Camera-position improvement process.

2D source image — Successful test using iPhone 13 FullHD 60fps video.

NeRF 3D model — Successful test using iPhone 13 FullHD 60fps video.

2D source image — Output from plastic-covered objects.

NeRF 3D model — Output from plastic-covered objects.

Old back office

New back office — Old back office

BabyLoop e-commerce: 360° product view

Results vs. the control group

+16.2%Purchase intentionA statistically significant increase among users who experienced the 3D visualisation.

+23.7%ImmersionThe experimental group reported higher immersion scores.

+12%Product knowledgeA better understanding of product features.

+11.5%Utilitarian valueHigher perceived usefulness of the product.