PortraitCraft: A Benchmark for Portrait Composition Understanding and Generation
Not on the map. Why?
Abstract (arXiv)
Portrait composition plays a central role in portrait aesthetics and visual communication, yet existing datasets and benchmarks mainly focus on coarse aesthetic scoring, generic image aesthetics, or unconstrained portrait generation. This limits systematic research on structured portrait composition analysis and controllable portrait generation under explicit composition requirements. In this paper, we introduce PortraitCraft, a unified benchmark for portrait composition understanding and generation. PortraitCraft is built on a dataset of approximately 50,000 curated real portrait images with structured multi-level supervision, including global composition scores, annotations over 13 composition attributes, attribute-level explanation texts, visual question answering pairs, and composition-oriented textual descriptions for generation. Based on this dataset, we establish two complementary benchmark tasks for composition understanding and composition-aware generation within a unified framework. The first evaluates portrait composition understanding through score prediction, fine-grained attribute reasoning, and image-grounded visual question answering, while the second evaluates portrait generation from structured composition descriptions under explicit composition constraints. We further define standardized evaluation protocols and provide reference baseline results with representative multimodal models. PortraitCraft provides a comprehensive benchmark for future research on fine-grained portrait understanding, interpretable aesthetic assessment, and controllable portrait generation.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 18 of its 27 references that are in the map. First posted on arXiv 4 Apr 2026.
Builds on
18 of 27 references.
References · 18
- 2025Qwen3-VL Technical Reportcited by 2,793
- 2023Vision-Language Models for Vision Tasks: A Surveycited by 839
- 2017AVA: A Video Dataset of Spatio-temporally Localized Atomic Visual Actionscited by 449
- 2016Photo Aesthetics Ranking Network with Attributes and Content Adaptationcited by 125
- 2024UniReal: Universal Image Generation and Editing via Learning Real-world Dynamicscited by 102
- 2022Personalized Image Aesthetics Assessment with Rich Attributescited by 58
- 2024ConsistentID: Portrait Generation with Multimodal Fine-Grained Identity Preservingcited by 43
- 2025ArtiMuse: Fine-Grained Image Aesthetics Assessment with Joint Scoring and Expert-Level Understandingcited by 41
- 2025EchoShot: Multi-Shot Portrait Video Generationcited by 29
- 2024FINECAPTION: Compositional Image Captioning Focusing on Wherever You Want at Any Granularitycited by 27
- 2025Argus: Vision-Centric Reasoning with Grounded Chain-of-Thoughtcited by 27
- 2024APDDv2: Aesthetics of Paintings and Drawings Dataset with Artist Labeled Scores and Commentscited by 18
6 more
- 2025LAPIS: A novel dataset for personalized image aesthetic assessmentcited by 16
- 2025Science-T2I: Addressing Scientific Illusions in Image Synthesiscited by 15
- 2023An Image Quality Assessment Dataset for Portraitscited by 14
- 2025HyperLoRA: Parameter-Efficient Adaptive Generation for Portrait Synthesiscited by 10
- 2025HiFi-Portrait: Zero-shot Identity-preserved Portrait Generation with High-fidelity Multi-face Fusioncited by 5
- 2025MDD-LLM: Towards Accuracy Large Language Models for Major Depressive Disorder Diagnosiscited by 3
Built on by
Citing papers per year: 2026: 1.
1 paper.