MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202611108730 A) filed by Manipal University Jaipur on September 10, 2026, for A System And Method For Low Vram Automated Ai Video Generation Using Structured Workflow Management, Llms, And Diffusion Models For Generations.
Inventors include Vijay Prakash Sharma; and Siddhant Jaiswal.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: The present invention relates to a low-VRAM optimized automated video generation system that converts structured user input into complete cinematic videos. The system comprises Google Sheets to collect structured inputs, n8n for workflow automation, large language models for generating detailed prompts, and ComfyUI-based diffusion pipelines for text-to-image, image-to-image, and image-to-video generation. To keep the character consistent across different scenes, a base reference image is first generated and then reused with controlled conditioning. The system utilized a structured JSON-based storyboard format to maintain proper scene order and continuity. Unlike many traditional pipelines that require powerful GPUs, this system's entire visual generation workflow is specifically optimized for low-VRAM systems, so it can run on consumer-level GPUs. The present automated AI-based video generation system enables resource-efficient character-consistent cinematic video creation on 8GB VRAM and 32GB RAM, reducing hardware requirements and manual intervention.
Disclaimer: Curated by HT Syndication.
