# Multi-image Composition on MICON-Bench uniformly sampled subset Source: [Multi-image Composition on MICON-Bench uniformly sampled subset]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:54:09.353Z Evaluation of multi-image composition capabilities using a uniformly sampled subset of the MICON-Bench dataset, reporting performance across detailed criteria including Object, Spatial, Attribute, Component, Foreground/Background interaction, and Story coherence. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Multi-image Composition | | Dataset (including any stated split) | MICON-Bench uniformly sampled subset | | Metric | Object Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Object Score](): Higher is better; unit: PERCENT. - [Spatial Score](): Higher is better; unit: PERCENT. - [Attribute Score](): Higher is better; unit: PERCENT. - [Component Score](): Higher is better; unit: PERCENT. - [FG/BG Interaction Score](): Higher is better; unit: PERCENT. - [Story Coherence Score](): Higher is better; unit: PERCENT. - [Average Composition Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Object Score | Result date | Report paper | | --- | --- | --- | --- | | GPT-Image | 92.9 | Not recorded | 699d30c46017c4460ffab7e4 | | Nano-Banana | 92.45 | Not recorded | 699d30c46017c4460ffab7e4 | | Nano-Banana-Pro | 90 | Not recorded | 699d30c46017c4460ffab7e4 | | DreamOmni2 | 90 | Not recorded | 699d30c46017c4460ffab7e4 | | BAGEL + DAR | 89.67 | Not recorded | 699d30c46017c4460ffab7e4 | | Qwen-Image-Edit-2507 | 89.4 | Not recorded | 699d30c46017c4460ffab7e4 | | OmniGen2 + DAR | 88.76 | Not recorded | 699d30c46017c4460ffab7e4 | | BAGEL | 88.24 | Not recorded | 699d30c46017c4460ffab7e4 | | OmniGen2 | 86.2 | Not recorded | 699d30c46017c4460ffab7e4 | | UNO | 56.03 | Not recorded | 699d30c46017c4460ffab7e4 | ### GPT-Image Object Score: 92.9. Result ID: 699e4eb25698665ec40d7d58. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Nano-Banana Object Score: 92.45. Result ID: 699e4eb25698665ec40d7d57. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Nano-Banana-Pro Object Score: 90. Result ID: 699e4eb25698665ec40d7d56. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DreamOmni2 Object Score: 90. Result ID: 699e4eb25698665ec40d7d5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### BAGEL + DAR Object Score: 89.67. Result ID: 699e4eb25698665ec40d7d5d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Dynamic Attention Rebalancing":true}. This record does not assert independent reproduction. ### Qwen-Image-Edit-2507 Object Score: 89.4. Result ID: 699e4eb25698665ec40d7d5b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### OmniGen2 + DAR Object Score: 88.76. Result ID: 699e4eb25698665ec40d7d5f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Dynamic Attention Rebalancing":true}. This record does not assert independent reproduction. ### BAGEL Object Score: 88.24. Result ID: 699e4eb25698665ec40d7d5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### OmniGen2 Object Score: 86.2. Result ID: 699e4eb25698665ec40d7d5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### UNO Object Score: 56.03. Result ID: 699e4eb25698665ec40d7d59. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.