I am looking for a method to align a set of real-camera images of a nose with their corresponding rendered images of the same 3D head in identical poses. The goal is precise alignment of key features like nostrils and edges to ensure accuracy for model tracking purposes. Its challanging as sometime key features like nostrils are hidden, and shading and ligtening is not exactly the same. I am open to alignment methods, whether feature-based (e.g., key points or edges), template matching, or other innovative solutions you might suggest. The deliverable should include a clear and functional method or algorithm that can achieve this alignment effectively (we need to recrated it so code and methods need to be suppled). If you have experience in image processing or computer vision, I would love to hear your approach and see how you can help with this project. Ultiamtely we'd like this to work on edge hardware - but for now getting it working is more important. Example images attached (pairs of real images and rendered images). We need to rendered image to be aligned as best as possible with the real image. The solution needs to work on all these images.