Depends on the model. E.g. image enhancement or super resolution models are sensitive to the camera model and can be trained to fix artifacts introduced by specific cameras.