← Back to assets

In the mirror world

allenai/objaverse

No image availableGLB

Asset information

Model is made using actual mirror in one scan from 263 photos normal+mirrored. Photoset consists of two different scenes in one environment: with a whole apple and a bitten one. Tape measure was set not only for the whole idea but to make it easier for Zephyr to distiguish the sides regarding the mirror - it was masked in all the mirror world. All the photos were taken in the real world, duh. But for the mirror world I had to mirror them. Basicly it was 4 sets of photos:with a whole apple normal and mirrored and two sets for bitten. Masking was a real pain! On close photos you just can't get were you taking this image through the mirror or not and what to mask. All 4 sets had their own masking pattern. After masking just it didn't align! Small test photoset was fine. But on the large photoset it always tried to be mirrored in the ground plane instead of mirror! Only using control points helped it. As a result I was able to scan not just the object but the idea!

Publisher
allenai
Format
GLB
License
CC BY
Dataset updated
Oct 8, 2026
Categories
art-abstract · food-drink
Tags
apple · mirror · 3dflowcup20
Animations
0

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace