two images are collinear

Dear author, regarding the use of the DA3_NESTED_GIANT_LARGE model, if we input the intrinsic and extrinsic parameters of two images and these two images are collinear (the optical axes of the two cameras are parallel), can metric depth estimation be achieved? That is to say, can we use this model for stereo matching and measure the distance? We look forward to your reply. Because when we call the DA3_NESTED_GIANT_LARGE model, input two images and manually input the intrinsic and extrinsic parameters, there is a problem about collinearity.