OmniVCus is a feedforward subject-driven video customization framework that uses a diffusion Transformer to generate videos of specific subjects under multimodal control conditions like depth and masks.

A video inpainting tool that automatically detects and erases fixed-pattern content like subtitles and logos from videos.

A Chrome/Firefox extension that uses LLM‑based subtitle or audio analysis to automatically detect and skip advertisement segments in Bilibili videos, with optional manual skip, user‑editable ad timestamps, and support for custom AI endpoints.

A Flask‑based web app that lets you upload videos, runs a deep‑learning watermark‑removal model on a GPU, and returns the cleaned video via a modern browser UI or a tiny REST API.

A VitePress plugin that automatically creates LLM‑friendly Markdown files (`llms.txt` and `llms‑full.txt`) and optional UI buttons for copying/downloading pages as Markdown, with support for custom labels, multilingual configs, and special `<llm-only>` / `<llm-exclude>` tags.

An agent-native video director skill that turns text prompts into reproducible video projects by automating asset sourcing, shot planning, and ffmpeg rendering.

A ComfyUI node for Wan2.2 that fixes slow-motion issues in image-to-video generation by increasing motion amplitude and improving camera movement responsiveness.

A family of Large Multimodal Models (LMMs) for video understanding and creation, enabling tasks like temporal retrieval, spatio-temporal grounding, and automated video editing.

VideoChat-Flash is a multimodal large language model designed for efficient long-context video understanding, capable of processing videos up to three hours long using hierarchical compression.

A tool for running and customizing real-time, interactive generative AI video pipelines using autoregressive diffusion models and a composable architecture.

A spatio-temporal video super-resolution project that uses Recurrent-Generative Back-Projection Networks and GANs to upscale low-resolution videos while maintaining temporal consistency and fine detail.

Astra is an interactive world model that uses an autoregressive diffusion transformer to generate realistic, action-conditioned long-horizon video predictions.

An open-source TypeScript SDK that allows developers and AI agents to create AI-generated videos using a declarative JSX syntax and a unified API for multiple AI providers.

JoyVASA is a diffusion-based framework that animates human and animal portraits using audio, decoupling facial identity from motion to enable high-quality, long-form video generation.

A training-time alignment framework for reference-to-video generation that improves identity consistency and reference fidelity without adding inference-time overhead.

Ditto is a framework for generating high-quality synthetic video editing data to train Editto, a state-of-the-art instruction-based video editing model.

ConsisID is a tuning-free, DiT-based text-to-video generation model that uses frequency decomposition to maintain consistent human identity across generated video frames.

MagicTime is a metamorphic video generation pipeline designed to create realistic time-lapse videos that depict significant physical transformations based on text prompts.

An image-to-video generation framework that transforms static images into high-quality videos using a multimodal LLM for semantic consistency.

A Metal-powered command-line tool for upscaling video files to higher resolutions on macOS.

A PyTorch implementation of EGVSR, an efficient video super-resolution framework that uses subpixel convolution to speed up inference for high-quality video upscaling.

An API implementation and example set for ByteDance's Seedance 2.0, enabling text, image, and multimodal reference-to-video generation with synchronized audio.

A concrete Claude Code skill pack for generating and publishing AI‑driven ad creatives (videos, images, thumbnails) via the Arcads API, with ready‑made prompts, multi‑step pipelines (Pixar‑style, claymation, UGC, etc.), cost tracking, and a Meta ad‑publishing helper.

A tool for creating hand-drawn explainer videos from scripts, featuring realistic stroke-by-stroke animation and programmatic motion graphics synced to AI-generated voiceovers.

llm.nvim is a Neovim plugin that adds AI‑driven code completion (ghost‑text) by calling large language models via the llm‑ls language‑server. It supports multiple backends (Hugging Face Inference API, Ollama, OpenAI‑compatible servers, and Hugging  Face TGI), automatically trims prompts to fit a model’s context window, and offers configurable per‑file activation, fill‑in‑the‑middle support, and custom request parameters.

Mysti is a VS Code extension that integrates 12 code‑generation AI providers (including Claude, Codex, Gemini, Copilot, Ollama, LocalAI, etc.) and adds collaboration features such as multi‑agent brainstorming, developer personas, autonomous execution with safety controls, @‑mention routing, and automatic context compaction. Install the extension, add at least one provider CLI (install a second for brainstorming), configure via JSON settings, and start coding with AI directly in the editor.

A ComfyUI plugin that adds a timeline‑based UI and a set of nodes for generating arbitrarily long videos with MiniMax H3. It splits the target duration into latent‑continuation segments, applies drift‑control masking, supports locked‑audio lip‑sync, and includes built‑in two‑stage SelfLift sampling for fast low‑res + high‑res generation. Install by cloning into `ComfyUI/custom_nodes`, then use the Material Planner to arrange media, set generation windows, and run the Finite Segment Sampling node to produce a seamless MP4.

A framework for generating infinite-length videos with high temporal consistency and controllable storylines using LoRA adapters on top of base video models.

An AI-assisted tool that generates storyboards and stitches them into coherent videos using Gemini and Vertex AI Veo.

A free, browser-based AI video upscaler that allows users to increase video resolution without sign-ups or software downloads.

VideoPrism is a Google‑DeepMind open‑source video foundation model (JAX/Flax) that provides pre‑trained video‑only and video‑text encoders. Trained on billions of image‑text and video‑text pairs, the frozen backbones achieve state‑of‑the‑art results on most video‑understanding benchmarks. The repo offers easy installation, checkpoint loading utilities, and three Colab demos (video encoding, cross‑modal retrieval, and classification fine‑tuning). Four model sizes (base/large, encoder‑only or encoder‑plus‑text) are hosted on Hugging Face, licensed under Apache 2.0/CC‑BY.

A local-first desktop environment for researching, planning, and rendering source-grounded educational videos using a mix of local and cloud AI models.