Today's Image & Video Generation: Fastest-Growing Projects — August 20, 2026
Today's the Image & Video Generation space on GitHub, there's a noticeable trend towards innovative image manipulation and generation techniques that cater to various creative needs such as artistic illustration and photo restoration. Developers are also leveraging AI tools for video editing, indicating an expanding interest in automating complex visual content creation tasks.
The repository shanliuling/dsh-image-gen has seen significant growth this week with a score of 31.17. This project allows users to generate images directly within DeepSeek Harness chats, providing a streamlined experience for image generation without the need to leave their current workflow. Its rapid growth likely stems from its unique approach and integration capabilities.
dacnay816y62-hub/photo-revival, with 454 stars, transforms everyday photos into poetic white-paper hand-drawn illustrations using Codex skills. This repository's popularity may be attributed to the creative appeal of converting mundane images into artistic masterpieces, combined with its ease of use and distinctive output style.
yokel1121/muyang-illustration-skills offers 25 fixed illustration styles for Codex, including prompt recipes and visual references. With a growth score of 16.28, it appears to be growing due to the comprehensive set of resources provided for users looking to enhance their image generation capabilities with predefined styles.
The Sateezg/codex-bridge repository bridges Claude Code's Codex CLI login to enable image generation and GPT-5 subagents without requiring an OpenAI API key. This tool, boasting 252 stars, is gaining traction as it offers developers a versatile platform for integrating AI-driven image creation into their projects without dependency on external APIs.
The shixinnt/codex-image-context-runtime repository provides a Codex plugin and local MCP runtime for context-bounded image generation and inspection. With a growth score of 11.56, its popularity could be due to the robustness it offers in handling complex image contexts locally, making it easier to manage and inspect generated images.
six-nut/PocketMen-with-you enables users to create high-fidelity Codex companions from two or more photos using local open-weight neural editing techniques. This project's growth score of 10.33 highlights its appeal in generating personalized digital companions without relying on external APIs, appealing to a niche audience interested in custom AI-driven characters.
The oil-oil/dsh-vision repository focuses on near-native image understanding for DeepSeek Harness with a growth score of 9.83. Its growth is likely due to the comprehensive approach it takes towards image comprehension and its potential integration points within existing workflows that require advanced visual analysis capabilities.
moskoo/xxg-portrait-rebuild-light offers an image-edit skill for portrait photos, reconstructing lighting elements and restoring skin texture with a director-led workflow. This repository's growth score of 7.78 suggests it is growing due to the precision and control offered in enhancing portrait photography with AI-driven tools.
The Tencent-Hunyuan/Hunyuan3D-Buffalo1.0 project presents a unified multimodal model for scalable 3D generation, understanding, and editing, garnering 212 stars. Its growth score of 7.03 indicates increasing interest in comprehensive solutions that address multiple aspects of 3D content creation and manipulation.
Lastly, mtsee/unicut is an AI-driven open-source video editing tool with a growth score of 6.81 and 185 stars. The repository's popularity may be due to its ability to streamline the complex process of video editing through automation and AI integration, appealing to both professionals and enthusiasts looking for efficient content creation tools.
These projects reflect an exciting period in the development of AI-driven image and video generation technologies, showcasing a range of innovative approaches aimed at enhancing creativity and efficiency across various applications.
The repository shanliuling/dsh-image-gen has seen significant growth this week with a score of 31.17. This project allows users to generate images directly within DeepSeek Harness chats, providing a streamlined experience for image generation without the need to leave their current workflow. Its rapid growth likely stems from its unique approach and integration capabilities.
dacnay816y62-hub/photo-revival, with 454 stars, transforms everyday photos into poetic white-paper hand-drawn illustrations using Codex skills. This repository's popularity may be attributed to the creative appeal of converting mundane images into artistic masterpieces, combined with its ease of use and distinctive output style.
yokel1121/muyang-illustration-skills offers 25 fixed illustration styles for Codex, including prompt recipes and visual references. With a growth score of 16.28, it appears to be growing due to the comprehensive set of resources provided for users looking to enhance their image generation capabilities with predefined styles.
The Sateezg/codex-bridge repository bridges Claude Code's Codex CLI login to enable image generation and GPT-5 subagents without requiring an OpenAI API key. This tool, boasting 252 stars, is gaining traction as it offers developers a versatile platform for integrating AI-driven image creation into their projects without dependency on external APIs.
The shixinnt/codex-image-context-runtime repository provides a Codex plugin and local MCP runtime for context-bounded image generation and inspection. With a growth score of 11.56, its popularity could be due to the robustness it offers in handling complex image contexts locally, making it easier to manage and inspect generated images.
six-nut/PocketMen-with-you enables users to create high-fidelity Codex companions from two or more photos using local open-weight neural editing techniques. This project's growth score of 10.33 highlights its appeal in generating personalized digital companions without relying on external APIs, appealing to a niche audience interested in custom AI-driven characters.
The oil-oil/dsh-vision repository focuses on near-native image understanding for DeepSeek Harness with a growth score of 9.83. Its growth is likely due to the comprehensive approach it takes towards image comprehension and its potential integration points within existing workflows that require advanced visual analysis capabilities.
moskoo/xxg-portrait-rebuild-light offers an image-edit skill for portrait photos, reconstructing lighting elements and restoring skin texture with a director-led workflow. This repository's growth score of 7.78 suggests it is growing due to the precision and control offered in enhancing portrait photography with AI-driven tools.
The Tencent-Hunyuan/Hunyuan3D-Buffalo1.0 project presents a unified multimodal model for scalable 3D generation, understanding, and editing, garnering 212 stars. Its growth score of 7.03 indicates increasing interest in comprehensive solutions that address multiple aspects of 3D content creation and manipulation.
Lastly, mtsee/unicut is an AI-driven open-source video editing tool with a growth score of 6.81 and 185 stars. The repository's popularity may be due to its ability to streamline the complex process of video editing through automation and AI integration, appealing to both professionals and enthusiasts looking for efficient content creation tools.
These projects reflect an exciting period in the development of AI-driven image and video generation technologies, showcasing a range of innovative approaches aimed at enhancing creativity and efficiency across various applications.