In this article, a new Image-based enhanced steganography approach is proposed for color images and takes imperceptibility, robustness, and data capacity challenges into consideration. We then incorporate a loss analysis framework into our methods to evaluate the trade-off between image quality and embedding efficiency. Enhanced steganography methods encounter challenges regarding imperceptibility, robustness, and data capacity, offering in this paper a novel method solution for color images. Ranging from image quality to embedding efficiency, the proposed method enables us to analyze trade-offs by incorporating a loss analysis framework. A novel multi-format embedding approach is proposed, where various data types such as text, images, and compressed files can be embedded into color images without sacrificing the visual quality. The method utilizes advanced deep-learning techniques and an optimized encoder-decoder architecture to attain state-of-the-art results in terms of Peak Signal-to-Noise Ratio and Structural Similarity Index Measure. The robustness of the system is demonstrated through extensive experiments against common image distortions (e.g., compression and noise) while retaining high payload capacities. This work further represents a valuable advancement toward achieving secure and efficient data hiding as well as its usages in digital watermarking, secure transmissions, and copyright protection.