ShutterMuse: Capture-Time Photography Guidance with MLLMs
Summary
Researchers introduce CaptureGuide-Bench, a benchmark for capture-time photography guidance, and ShutterMuse, a unified multimodal LLM trained to provide composition and pose recommendations, demonstrating improved performance over general-purpose models.
View Cached Full Text
Cached at: 06/25/26, 09:10 AM
Paper page - ShutterMuse: Capture-Time Photography Guidance with MLLMs
Source: https://huggingface.co/papers/2606.25763
Abstract
Researchers developed a new benchmark and dataset for photography assistance, along with a unified multimodal model that provides both composition guidance and pose recommendations during image capture.
Real-world photography requires capture-time guidance for both camera framing and subject pose. Yet existingaesthetic croppingbenchmarks mainly evaluate post-hoc crop prediction and overlook subject-side recommendations, leaving the capture-time guidance capabilities ofmultimodal large language models(MLLMs) underexplored. To address this gap, we introduce CaptureGuide-Bench, a benchmark with two complementary tasks:photographer-side compositiondecision and refinement, and subject-side scene-conditioned pose recommendation. Our evaluation reveals limitations: general-purpose MLLMs can make composition decisions but lack precise refinement localization, while specializedaesthetic croppingmodels localize crops effectively but are limited to refinement; neither provides actionable pose guidance. To support model development, we further construct CaptureGuide-Dataset, comprising 130K samples with textual rationales and structuredvisual annotations, and develop ShutterMuse, a unified MLLM trained with supervised andreinforcement fine-tuning. Experiments on CaptureGuide-Bench show that ShutterMuse achieves the best overall photographer-side performance among evaluated baselines and competitivesubject-side pose recommendationwith substantially lower inference cost, demonstrating the potential of MLLMs as interactive assistants for photography during image capture.
View arXiv pageView PDFProject pageGitHub11Add to collection
Get this paper in your agent:
hf papers read 2606\.25763
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper1
#### ShutterMuse/ShutterMuse Image-Text-to-Text• 9B• Updatedabout 3 hours ago
Datasets citing this paper1
#### ShutterMuse/CaptureGuide-Bench Updatedabout 3 hours ago • 5
Spaces citing this paper1
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
MuseBench: Benchmarking Intent-Level Audiovisual Arts Understanding in MLLMs
MuseBench is a comprehensive benchmark introduced to evaluate multimodal large language models on nuanced, intent-level understanding of audiovisual arts, revealing that even the best model achieves only 48.29% accuracy compared to 87.18% for human experts.
Mind's Eye: A Benchmark of Visual Abstraction, Transformation and Composition for Multimodal LLMs
Researchers introduce Mind’s Eye, a benchmark of eight visual-cognitive tasks that reveals top multimodal LLMs score under 50% while humans reach 80%, exposing major gaps in visual abstraction, relation mapping and mental transformation.
Thinking on Shots: Consistent Multi-Shot Video Editing with Agentic Reasoning
This paper introduces an agentic framework combining LLMs and VLMs for consistent multi-instruction video editing across multiple shots, and proposes the MMLVE task and benchmark to evaluate performance.
TimeLens2: Generalist Video Temporal Grounding with Multimodal LLMs
TimeLens2 introduces a generalist video temporal grounding method using multimodal LLMs, treating temporal evidence as an interval set and achieving state-of-the-art performance across multiple benchmarks.
Zero-Shot Mission-Level Evaluation for Aerial MLLM Agents
MissionBench is a new benchmark for evaluating multimodal large language models (MLLMs) on long-horizon embodied tasks in aerial 3D environments, revealing that even the best models succeed on fewer than 35% of missions compared to 84.4% human performance.