Video2X — ML video super-resolution & interpolation

Machine-learning video super-resolution and frame-interpolation framework (C/C++ v6) wrapping Real-ESRGAN, Real-CUGAN, RIFE and Anime4K via ncnn/Vulkan, with CLI and Qt6 GUI.

Link

• Visit resource

Related resources

• rife-ncnn-vulkan — RIFE interpolation via VulkanPyTorch-free, cross-platform RIFE frame-interpolation CLI using ncnn and Vulkan for concurrent CPU+GPU processing.
• Practical-RIFE — real-time frame interpolationActively maintained, deployment-focused fork of RIFE (real-time intermediate flow estimation) for arbitrary-timestep video frame interpolat…
• BasicSR — image/video restoration toolboxOpen-source PyTorch toolbox for image/video super-resolution and restoration implementing EDSR, RCAN, ESRGAN, EDVR, BasicVSR and SwinIR.
• vkdt — Vulkan node-graph media processingGLSL/Vulkan node-graph (DAG) processing engine for raw stills and video, running all processing as GPU shaders.
• openfx-misc — OpenFX Plugin SetCollection of OpenFX (OFX) plugins — Merge, Roto, keyers, generators, retiming — compatible with any OFX host.
• SimpleVQADeep-learning no-reference video quality assessment for user-generated content, in PyTorch with SlowFast feature extraction.
• butteraugli (Rust port)Pure Rust port of Butteraugli from libjxl with XYB color, visual masking, and multi-scale bands, validated to within 0.001% of the referenc…
• AutoShotCVPR-NAS 2023 shot boundary detection model with a NAS-optimized architecture and the new SHOT dataset (853 videos, 11,606 annotations).
• ITU-T P.1203 reference implementationOfficial Python reference implementation of the ITU-T P.1203 HTTP adaptive streaming QoE model (modes 0-3), outputting per-second and integ…