Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.25254 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916044456591360 |
|---|---|
| author | Cekinmez, Jasin Mitsuhashi, Ryo Wu, Addison J. Yin, Yida |
| author_facet | Cekinmez, Jasin Mitsuhashi, Ryo Wu, Addison J. Yin, Yida |
| contents | With unified model-generated images now widespread online, attributing their model of origin offers a path toward transparency and deeper insight into the characteristic behaviors of individual models. Prior work has explored provenance in LLM-generated text, diffusion model images, and datasets, but the separability of unified model-generated images remains an underexplored area. We address this gap by examining separability across corruption, domains, and prompt languages using images generated by seven unified models. We show that model attribution is highly feasible as our model achieves near-perfect accuracy with around 20K images per model. Corruptions and structural perturbations have only a modest effect on attribution performance, and cross-domain generalization reveals that semantic content contributes to separability but is not the dominant signal. Finally, we observe that for most models, prompt language attribution is around chance levels, suggesting minimal language-specific visual signatures. These findings highlight consistent model-specific visual characteristics in unified models outputs and open new directions for tracing and auditing generative image pipelines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_25254 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Guess the Unified Model: How Much Can We Recover from Generated Images? Cekinmez, Jasin Mitsuhashi, Ryo Wu, Addison J. Yin, Yida Computer Vision and Pattern Recognition Artificial Intelligence With unified model-generated images now widespread online, attributing their model of origin offers a path toward transparency and deeper insight into the characteristic behaviors of individual models. Prior work has explored provenance in LLM-generated text, diffusion model images, and datasets, but the separability of unified model-generated images remains an underexplored area. We address this gap by examining separability across corruption, domains, and prompt languages using images generated by seven unified models. We show that model attribution is highly feasible as our model achieves near-perfect accuracy with around 20K images per model. Corruptions and structural perturbations have only a modest effect on attribution performance, and cross-domain generalization reveals that semantic content contributes to separability but is not the dominant signal. Finally, we observe that for most models, prompt language attribution is around chance levels, suggesting minimal language-specific visual signatures. These findings highlight consistent model-specific visual characteristics in unified models outputs and open new directions for tracing and auditing generative image pipelines. |
| title | Guess the Unified Model: How Much Can We Recover from Generated Images? |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2605.25254 |