ImagineTime Benchmark: 750 Cases for Evaluating Temporal Reasoning in Image Models

Xin-Rui released the ImagineTime benchmark in 2026 to evaluate image generation models. It contains 750 benchmark cases designed to test a model's ability to produce ordered 2x2 motion sheets with coherent entities and state transitions. The dataset was published with the paper 'Can Image Models Imagine Time?' and is hosted on Hugging Face.

Use Cases

• Benchmarking model performance on temporal coherence based on the described motion sheet task.
• Evaluating spatial relation understanding in generated image sequences based on the dataset's process specs.
• Testing a model's ability to follow task constraints and interaction prompts as defined in the benchmark cases.

Strengths

• Contains 750 benchmark cases for structured evaluation.
• Specifically designed to test multiple facets of temporal reasoning in image generation.

Limitations

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

Provenance

Source

Xin-Rui via Hugging Face.

Collection Method

Created as a public benchmark for the associated research paper.

Freshness

Last updated 2026-06-10 08:04:46; freshness should be verified.