Fringe-pattern projection systems are capable of non-contacting 3D surface full-field measurement with high accuracy. However, the systems are prone to intensity saturation and low signal-to-noise ratio (SNR) when measuring objects with a large range of reflectivity across the surface. Intensity saturation occurs when the light intensity directed to the camera exceeds the maximum intensity quantization level. A low SNR occurs when there is a low intensity modulation compared to the amount of noise in the image. Saturation and low SNR can result in significant measurement error. This paper presents a method for saturation avoidance during object-surface measurement, by adaptively adjusting the projected fringe-pattern intensities, through the maximum input gray level (MIGL). A high SNR can be maintained while avoiding saturation by combining the intensities from phase-shifted images captured at different MIGL, into a set of composite phase-shifted images. In measurement of a black and white checkerboard at different depths, the newly developed method reduced errors by an average 0.25 mm compared to the highest accuracy measurement using a uniform MIGL.