The invention addresses the lack of comprehensive and quantitative methods for measurements of unwanted visual “mura” effects in displays and image sensors. Mura is generated by errors that are significantly smaller than what is needed for the function of the device, and sometimes smaller than the random variations in the patterns or structures. Capturing essentially all mura defects in a workpiece in a short time requires a daunting combination of sensitivity, statistical data reduction and speed. The invention devices an inspection method, e.g. optical, which maximizes the sensitivity to mura effects and suppresses artifacts from the mura inspection hardware itself and from noise. It does so by scanning the sensor, e.g. a high-resolution camera, creating a region of high internal accuracy across the mura effects. One important example is for mura related to placement errors, where a stage with better than 10 nanometer precision within a 100 mm range is created. A sampling scheme reduces the data volume and separates between instrument errors and real defects based on their different geometrical signatures. The high-resolution camera scans sparse lines at an angle to the dominating directions of expected mura defects, creating extended sensor fields with high internal precision, and quantifying edge placements in small windows in said extended fields. The mura is classified and presented as type, location and severity.