MODEST: Multi-Optics Depth-of-Field Stereo Images for Computational Photography

MODEST is the first high-resolution stereo DSLR dataset that systematically captures variations in focal length and aperture using professional camera optics. The dataset addresses key limitations of existing depth-of-field datasets for depth-of-field rendering, stereo vision, and optical scene understanding. It was created by independent-vision-lab and was last updated on the platform in June 2026.

Use Cases

• Training depth-of-field rendering models based on high-resolution stereo images with varied optical parameters.
• Developing stereo vision algorithms based on systematically captured focal length and aperture variations.
• Researching optical scene understanding based on a large-scale computational photography dataset.

Strengths

• First high-resolution stereo DSLR dataset with systematic optical parameter variations.
• Addresses key limitations of existing depth-of-field datasets.
• Created by a dedicated research lab (independent-vision-lab).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset size are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

independent-vision-lab

Collection Method

Systematically captured using professional camera optics.

Freshness

Last updated 2026-06-04 14:08:21; freshness should be verified.