ACM Multimedia 2026 Grand Challenge Proposal: Single-Image Guided Multi-Angle Image Synthesis

Jin Chen

Keywords: single-image guided multi-angle synthesis, artificial intelligence generated content, convolutional neural networks, dataset, grand challenge

TL;DR: We propose a Grand Challenge on Single-Image Guided Multi-Angle Image Synthesis supported by the MGMultiAngle dataset, aiming to enhance AIGC video controllability and accelerate industrial application of multi-view generation technologies.

Abstract: We propose a Grand Challenge on Single-Image Guided Multi-Angle Image Synthesis, a cutting-edge task that transforms a single reference image into a sequence of multi-angle images with consistent spatial logic and content integrity. This technology addresses critical pain points in digital content creation, product visualization, and architectural spatial presentation by automating the labor-intensive multi-angle image generation process, while serving as a precise control signal for AIGC video generation to enhance the controllability and predictability of video outputs. To support this challenge, we introduce MGMultiAngle, a high-quality benchmark dataset featuring high-resolution reference images (1080p), precise 3D spatial calibration for diverse poses, multi-angle paired annotations, and detailed structured captions. MGMultiAngle overcomes limitations of existing datasets (e.g., insufficient pose coverage, incomplete spatial consistency annotations) and provides a rigorous evaluation foundation for model development. Through this challenge, we aim to foster innovations in single-image-driven novel view synthesis, spatial geometric modeling, and visual consistency preservation, accelerating the industrial application of multi-angle generation technologies over the next 3–5 years.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 12