vivo BlueImage Lab explores mobile imaging, computational photography, content generation, agents, and foundation models. The lab's public research presence is visible through the vivo Camera Research GitHub organization, which publishes projects spanning diffusion-based imaging, matting, bokeh rendering, visual grounding, photo editing, and video generation.
vivo's public BlueImage material describes an imaging stack that brings together sensing, algorithm, and imaging-chip technologies, while exploring AIGC and 3D imaging for future mobile imaging experiences.
Lab
vivo Camera Research
Public GitHub organization for vivo BlueImage Lab research projects.
Computational Photography
MagicBokeh
Unified high-zoom bokeh rendering research.
AI Photo Editing
SmartPhotoCrafter
Unified reasoning, generation, and optimization for automatic photographic image editing.
Video Bokeh
Any-to-Bokeh
Arbitrary-subject video refocusing with video diffusion models.
Matting
SDMatte
Diffusion-model grafting for interactive matting.
Aesthetic Recommendation
PPC
Photography perspective composition for aesthetic perspective recommendation.
Video Generation
MagicTryOn
Garment-preserving video virtual try-on with diffusion transformers.
Human Animation
Hyper-Motion
Pose-guided human image animation based on a large-scale video diffusion transformer.