**Magic Clothing: Controllable Garment-Driven Image Synthesis** This paper introduces Magic Clothing, a latent diffusion model (LDM)-based architecture for generating images of characters wearing target garments, conditioned on text prompts. The primary challenge is to preserve garment details and maintain faithfulness to the text prompts. To address this, the authors propose a garment extractor that captures detailed garment features and incorporates them into the LDMs using self-attention fusion. This ensures that the garment details remain unchanged on the target character. Additionally, the method employs joint classifier-free guidance to balance the control of garment features and text prompts. The garment extractor is a plug-in module compatible with various finetuned LDMs and extensions like ControlNet and IP-Adapter, enhancing the diversity and controllability of the generated characters. The authors also introduce MP-LPIPS, a robust metric for evaluating the consistency of the target image to the source garment. Extensive experiments demonstrate that Magic Clothing achieves state-of-the-art results under various conditional controls for garment-driven image synthesis.**Magic Clothing: Controllable Garment-Driven Image Synthesis** This paper introduces Magic Clothing, a latent diffusion model (LDM)-based architecture for generating images of characters wearing target garments, conditioned on text prompts. The primary challenge is to preserve garment details and maintain faithfulness to the text prompts. To address this, the authors propose a garment extractor that captures detailed garment features and incorporates them into the LDMs using self-attention fusion. This ensures that the garment details remain unchanged on the target character. Additionally, the method employs joint classifier-free guidance to balance the control of garment features and text prompts. The garment extractor is a plug-in module compatible with various finetuned LDMs and extensions like ControlNet and IP-Adapter, enhancing the diversity and controllability of the generated characters. The authors also introduce MP-LPIPS, a robust metric for evaluating the consistency of the target image to the source garment. Extensive experiments demonstrate that Magic Clothing achieves state-of-the-art results under various conditional controls for garment-driven image synthesis.