IMAGGarment-1 introduces a framework designed for fine-grained garment generation, allowing users to synthesize detailed clothing designs while controlling aspects like silhouette, color, and logo placement. This approach tackles the limitations of existing methods that typically rely on single-condition inputs, enhancing flexibility in personalized fashion design and digital apparel. The framework operates through a two-stage training process. The first stage focuses on global appearance, utilizing a mixed attention module to integrate silhouette and color. The second stage emphasizes local details, refining the garment by adding user-defined logos and ensuring spatial accuracy. A key element of this work is GarmentBench, a dataset containing over 180,000 garment samples that pair with various design conditions such as sketches, color references, and logo placements. This dataset is significant because it combines multiple data formats, making it easier for researchers and developers to experiment with garment generation. The dataset’s structure includes folders for different types of data and provides annotations in a specific Parquet format, ensuring consistency and usability. The experimental results show that IMAGGarment-1 surpasses existing models in several areas, including structural stability and color fidelity. The framework's design allows for precise control over garment features, making it a valuable tool for fashion designers and digital artists. The research team has made both the code and model publicly available, encouraging further exploration and development in this area of fashion technology.