$ show TencentARC/StereoCrafter
StereoCrafter is an application that converts monocular videos to stereoscopic 3D videos using a two-stage diffusion-based pipeline. The first stage estimates depth via DepthCrafter and generates a right-eye view through forward warping. The second stage refines the stereo video using a fine-tuned Stable Video Diffusion model with spatial tiling for high-resolution processing. The system relies on external dependencies including DepthCrafter and Forward-Warp submodules, and requires manual download of model weights. The codebase demonstrates modular design with clear separation between depth estimation, warping, and inpainting stages.
The LLM identified 2 files worth highlighting for reuse:
Defines a reusable StableVideoDiffusionInpaintingPipeline class that extends Diffusers for video inpainting with spatial tiling. Handles high-resolution videos by splitting frames into overlapping tiles, processing them independently, and blending results to avoid seams. Isolated from application-specific logic and adaptable to other video inpainting tasks.
Contains blend_h, blend_v, and spatial_tiled_process functions that implement a generic spatial tiling strategy for processing large images/video frames in chunks. Includes horizontal/vertical blending to seamlessly recombine tiles. This pattern is reusable across various image/video processing tasks requiring fixed-input-size models.