Drag Your GAN
b2cFree
Interactive Point-based Manipulation on the Generative Image Manifold
Drag Your GAN is a research project enabling interactive point-based manipulation on generative image manifolds. Users can intuitively drag points on an image to target locations, synthesizing new content while preserving photorealistic quality and object structure. The project was presented at SIGGRAPH 2023 and accepted at CVPR 2021.
Supplemental document availableResearch projectInteractive image manipulationGAN-based image synthesis
ImageBind by Meta AI
b2bFree
A multimodal model that binds images with other modalities.
ImageBind is a foundational AI model developed by Meta AI. It can understand and generate data across six different modalities: image, text, audio, depth, thermal, and IMU motion. This allows for powerful cross-modal understanding and generation capabilities.
Multimodal understanding across 6 modalitiesCross-modal generationFoundation for AI researchOpen-source model
Precise Image Editing via Recognition and Generation Tasks
Emu Edit offers precise image editing by combining recognition and generation tasks. It supports various manipulations, including region-based and free-form editing. The tool also handles core computer vision tasks like detection and segmentation.
Region-based editingFree-form editingImage detectionImage segmentation
