TemplateCraft: Agentic Visual Template Generation

2026-09-25

AI Summary

This study addresses the time-consuming nature of manual orchestration in short video template generation by proposing a Qwen3-VL-based multi-agent collaborative framework that automatically translates natural language instructions into executable templates. Methodologically, it designs a pipeline encompassing planning, asset and effect workflow generation, and protocol compilation. A novel Planner-Evaluator closed-loop mechanism is introduced, leveraging execution feedback for targeted rollback and integrating stage-level with long-term memory to enable parameter-free iterative revision, complemented by persistent asset storage for enhanced reusability. Evaluated on the TemplateBench benchmark, the framework increases success rates for image and video template generation to 66.7% and 50.0%, respectively, significantly improving template compliance and stylistic consistency.

Abstract

The growing popularity of short videos has driven demand for one-click content creation. Visual templates turn uploaded images into personalized content with preset effects, but reusable template generation still requires substantial manual effort in asset preparation and tool orchestration. We propose TemplateCraft, a multi-agent system that converts natural-language instructions into client-executable templates through planning, material generation, effect-workflow generation, and protocol compilation. Its Planner-Evaluator loop uses execution feedback for targeted rollback, while stage-level and long-term memory support revision without parameter updates. We evaluate TemplateCraft on TemplateBench, derived from 60 real-world templates. With the same Qwen3-VL backbone, TemplateCraft raises image/video generation success rates from 56.7%/30.0% to 66.7%/50.0% over Planner-only (best-of-three) and improves template adherence and style consistency. With additional evaluation and revision, it matches or exceeds a GPT-4o Planner-only baseline on selected metrics. Persistent assets further improve cross-input style consistency.

Problem

• visual template generation
• short video creation
• one-click content creation
• template automation

Innovation

• Multi-agent System
• Visual Template Generation
• Planner-Evaluator Loop
• Memory Mechanism
• Execution Feedback