Micro-scale additive manufacturing has seen significant growth over the past years, where improving the accuracy of complex micro-scale geometries is seen as an important challenge. Using grayscale images rather than black and white images during production is an effective method to improve the fabrication quality. This thesis presents a model-based optimization method for improving the dimensional accuracy of parts using voxel-based grayscale dynamic optimization during continuous 3D printing. A detailed solidification model has been developed and used to estimate the curing dynamics of the resin used in 3D printing. The irradiance of the light beam projected for each pixel influences a larger volume on the resin than the targeted voxel. The proposed model-based method optimizes the images considering the light distribution from all closely related pixels to maintain the accuracy of the micro part. The results of this method have been applied to the printing of complex 3D parts to show that optimized grayscale images improve the areas with overcuring significantly. It is shown that the number of overcured voxels was reduced by 24.7% compared to the original images. Actual printing results from the experimental setup confirm the improve-ments in the accuracy and precision of the printing method. The optimization method has been further improved by allowing variable printing speed during pro-duction and optimizing the speed profile of the print alongside grayscaling. This approach allows for printing of certain geometries that would otherwise be challenging to produce accurately. Computational limitations of performing speed and grayscale optimization simultaneously has been overcome by utilizing the symmetry of certain special cases to reduce optimization variables.