A tech startup specializing in high-resolution photo archiving uploads large image files to Amazon S3 using parallel multipart uploads via the S3 API. Files are occasionally reuploaded, replacing existing objects. In the first month, images are frequently accessed by customers. After that, access becomes sporadic and unpredictable, varying from object to object. The company wants to minimize S3 storage costs without sacrificing durability or accessibility. What combination of actions should a solutions architect suggest to fulfill these goals?