Can someone share the parameters to tweak for the output to be just a high quality ortho without any attempt to build a 3D model?
The best you can do is use --fast-orthophoto, which creates a 3D model using only the sparse point cloud and skips the dense reconstruction step.
We don’t have a way to stitch images. See https://support.pix4d.com/hc/en-us/articles/202558869-Photo-Stitching-vs-Orthomosaic-Generation for a good explanation of stitching vs orthomosaic creation.