simpleposter: a simple baseline for product poster generation

2026-05-09

AI Summary

This work addresses the challenges of low product appearance fidelity and imprecise multi-line text layout control in commercial poster generation by proposing an end-to-end method based on an image inpainting framework. The approach leverages full-parameter fine-tuning of a foundation model to faithfully preserve the main subject structure and introduces a zero-overhead, character-level positional encoding scheme to enable geometry-aware controllable text generation—eliminating the need for external modules such as ControlNet or OCR. By simplifying the overall architecture, the method substantially reduces computational overhead while achieving a subject preservation rate of 98.7%, significantly outperforming SeedEdit 3.0 (55.2%) and PosterMaker (85.3%), and simultaneously improving text rendering accuracy.

Abstract

Product poster generation poses distinct challenges beyond general poster design, requiring both faithful preservation of product appearance and precise control over dense, multi-line text layouts. Prior methods typically adopt inpainting frameworks augmented with auxiliary modules such as ControlNet and OCR encoders. However, these approaches introduce architectural complexity and computational overhead while still suffering from text errors and subject extension artifacts. We present SimplePoster, a simple yet effective inpainting-based framework that achieves faithful subject preservation and accurate, position-controllable text rendering without external controllers. Our approach builds on two observations: (1) full-parameter fine-tuning of the base model effectively suppresses subject extension, outperforming ControlNet-based alternatives; and (2) a zero-cost character-level position encoding enables geometry-aware text generation without dedicated layout modules. Experiments show that SimplePoster achieves a $98.7\%$ subject preservation rate, compared to $55.2\%$ for SeedEdit 3.0 and $85.3\%$ for PosterMaker, while also improving text rendering accuracy. Code, models, benchmark and a part of training data will be available at https://github.com/Alibaba-YuFeng/SIMPLEPOSTER

Problem

• product poster generation
• subject preservation
• text layout
• inpainting
• text rendering

Innovation

• inpainting
• subject preservation
• position-controllable text
• character-level encoding
• diffusion model fine-tuning