Today's Image & Video Generation: Fastest-Growing Projects — November 02, 2026
This week, the Image & Video Generation space on GitHub continues to see a flurry of activity, with projects ranging from innovative video storyboard generators to comprehensive AI-driven design agents and static image galleries. The most prominent project in this category is dsh-video-storyboard by SAOS-3624, which offers a unique way to encapsulate complex video content into single-image storyboards packed with detailed metadata.
dsh-video-storyboard provides scene-aware storyboard generation for the DeepSeek Harness platform, creating images that include timestamps, transcripts, and audio envelopes. With a growth score of 73.21 and over 350 stars, its popularity is likely driven by its ability to simplify complex video content into digestible storyboards that can be easily shared or analyzed.
ai-costume-gallery-cn, developed by Brilliant666, offers an extensive collection of AI-generated costume reference images in Chinese. The project has received a growth score of 40.38 and accumulated 119 stars. Its growth is likely due to the niche yet valuable resource it provides for artists and designers looking for inspiration or references.
design-agent, created by langchain-ai, is an open-source design agent that can generate on-brand visuals such as stills, decks, diagrams, charts, and motion graphics directly from chat threads. With 544 stars and a growth score of 35.33, the project's traction likely stems from its ability to democratize AI-driven design tools for broader audiences.
nvzhuang, developed by paipiange, offers an extensive collection of AI-generated female clothing reference images and a virtual try-on website, with support for Chinese and English language searches as well as various image generation APIs. The project has amassed 1,861 stars and a growth score of 28.94, indicating its significant appeal to fashion enthusiasts and designers seeking inspiration or virtual fitting experiences.
manga-generator, created by stockedge, is a comprehensive tool for manga creation that includes scriptwriting, image generation, layout design, PDF export, and more, all tailored for Codex users. With 160 stars and a growth score of 22.27, the project's growing popularity may be attributed to its unique approach in streamlining the entire manga production process.
MeerCut, developed by CodeWithAmeer, is a free and open-source video editor for Windows 11 that supports multitrack editing, titles, transitions, effects, color grading, audio mixing, and local AI tools. Despite having fewer stars (92) and a lower growth score of 17.33, MeerCut's development pace remains steady, suggesting it caters to a niche but dedicated user base.
pixel-loom, created by yangsir2077, is an open-source pixel art and animation editor that supports AI-generated content through MCP. With 100 stars and a growth score of 8.94, the project's modest yet consistent growth may be due to its niche appeal among pixel artists who value both traditional and AI-assisted creation tools.
shisei, developed by wquguru, is a free CC0 mannequin pose reference library designed for AI image generation tools like Nano Banana, ChatGPT, ComfyUI, and Midjourney. The project has accumulated 219 stars and a growth score of 7.96, indicating steady interest from developers working with these platforms.
CityDiffusion2, created by bart1259, is a project for generating and editing stylized top-down city maps using conditional diffusion models trained on OpenStreetMap data. With 23 stars and a growth score of 6.62, the project's steady development suggests it caters to a specific audience interested in urban planning and map generation.
iris-3b, developed by speridlabs, is a pixel-space generation model that can generate images directly without using VAEs, with fine-tuning capabilities for monocular depth estimation and image restoration. The project has 212 stars and a growth score of 6.14, indicating steady interest from researchers and developers in the field of generative models.
These projects collectively showcase the diversity and innovative spirit within the Image & Video Generation space on GitHub this week, with each tool addressing unique challenges and requirements for AI-driven content creation.
dsh-video-storyboard provides scene-aware storyboard generation for the DeepSeek Harness platform, creating images that include timestamps, transcripts, and audio envelopes. With a growth score of 73.21 and over 350 stars, its popularity is likely driven by its ability to simplify complex video content into digestible storyboards that can be easily shared or analyzed.
ai-costume-gallery-cn, developed by Brilliant666, offers an extensive collection of AI-generated costume reference images in Chinese. The project has received a growth score of 40.38 and accumulated 119 stars. Its growth is likely due to the niche yet valuable resource it provides for artists and designers looking for inspiration or references.
design-agent, created by langchain-ai, is an open-source design agent that can generate on-brand visuals such as stills, decks, diagrams, charts, and motion graphics directly from chat threads. With 544 stars and a growth score of 35.33, the project's traction likely stems from its ability to democratize AI-driven design tools for broader audiences.
nvzhuang, developed by paipiange, offers an extensive collection of AI-generated female clothing reference images and a virtual try-on website, with support for Chinese and English language searches as well as various image generation APIs. The project has amassed 1,861 stars and a growth score of 28.94, indicating its significant appeal to fashion enthusiasts and designers seeking inspiration or virtual fitting experiences.
manga-generator, created by stockedge, is a comprehensive tool for manga creation that includes scriptwriting, image generation, layout design, PDF export, and more, all tailored for Codex users. With 160 stars and a growth score of 22.27, the project's growing popularity may be attributed to its unique approach in streamlining the entire manga production process.
MeerCut, developed by CodeWithAmeer, is a free and open-source video editor for Windows 11 that supports multitrack editing, titles, transitions, effects, color grading, audio mixing, and local AI tools. Despite having fewer stars (92) and a lower growth score of 17.33, MeerCut's development pace remains steady, suggesting it caters to a niche but dedicated user base.
pixel-loom, created by yangsir2077, is an open-source pixel art and animation editor that supports AI-generated content through MCP. With 100 stars and a growth score of 8.94, the project's modest yet consistent growth may be due to its niche appeal among pixel artists who value both traditional and AI-assisted creation tools.
shisei, developed by wquguru, is a free CC0 mannequin pose reference library designed for AI image generation tools like Nano Banana, ChatGPT, ComfyUI, and Midjourney. The project has accumulated 219 stars and a growth score of 7.96, indicating steady interest from developers working with these platforms.
CityDiffusion2, created by bart1259, is a project for generating and editing stylized top-down city maps using conditional diffusion models trained on OpenStreetMap data. With 23 stars and a growth score of 6.62, the project's steady development suggests it caters to a specific audience interested in urban planning and map generation.
iris-3b, developed by speridlabs, is a pixel-space generation model that can generate images directly without using VAEs, with fine-tuning capabilities for monocular depth estimation and image restoration. The project has 212 stars and a growth score of 6.14, indicating steady interest from researchers and developers in the field of generative models.
These projects collectively showcase the diversity and innovative spirit within the Image & Video Generation space on GitHub this week, with each tool addressing unique challenges and requirements for AI-driven content creation.