Mirror in the Model: Ad Banner Image Generation via Reflective Multi-LLM and Multi-modal Agents

TL;DR

MIMO is a novel framework that combines multi-modal agents and iterative refinement to automatically generate high-quality, well-structured advertising banners from simple prompts and logos, outperforming existing methods.

Contribution

The paper introduces MIMO, a new agentic refinement framework that integrates hierarchical multi-modal agents and a coordination loop for improved ad banner generation.

Findings

MIMO outperforms existing diffusion and LLM-based baselines in real-world scenarios.

MIMO automatically detects and corrects errors during banner generation.

The framework requires only a natural language prompt and logo image as input.

Abstract

Recent generative models such as GPT-4o have shown strong capabilities in producing high-quality images with accurate text rendering. However, commercial design tasks like advertising banners demand more than visual fidelity -- they require structured layouts, precise typography, consistent branding, and more. In this paper, we introduce MIMO (Mirror In-the-Model), an agentic refinement framework for automatic ad banner generation. MIMO combines a hierarchical multi-modal agent system (MIMO-Core) with a coordination loop (MIMO-Loop) that explores multiple stylistic directions and iteratively improves design quality. Requiring only a simple natural language based prompt and logo image as input, MIMO automatically detects and corrects multiple types of errors during generation. Experiments show that MIMO significantly outperforms existing diffusion and LLM-based baselines in real-world banner design scenarios.