The paper introduces TID2008, a new image database designed for evaluating full-reference visual quality assessment metrics. The database contains 1700 test images, each distorted in 17 different ways at four levels of distortion severity. Over 800 experiments were conducted with observers from Finland, Italy, and Ukraine, resulting in Mean Opinion Scores (MOS) that can be used to test and design new visual quality metrics. The database is freely available for scientific research and has been compared to the LIVE Database, showing superior performance in terms of the number of distortion types and accuracy of MOS estimates. The paper also analyzes the performance of various quality metrics using TID2008, highlighting the limitations of widely used metrics like PSNR and MSE, and recommending PSNR-HVS and PSNR-HVS-M for evaluating image filtering and lossy compression.The paper introduces TID2008, a new image database designed for evaluating full-reference visual quality assessment metrics. The database contains 1700 test images, each distorted in 17 different ways at four levels of distortion severity. Over 800 experiments were conducted with observers from Finland, Italy, and Ukraine, resulting in Mean Opinion Scores (MOS) that can be used to test and design new visual quality metrics. The database is freely available for scientific research and has been compared to the LIVE Database, showing superior performance in terms of the number of distortion types and accuracy of MOS estimates. The paper also analyzes the performance of various quality metrics using TID2008, highlighting the limitations of widely used metrics like PSNR and MSE, and recommending PSNR-HVS and PSNR-HVS-M for evaluating image filtering and lossy compression.