If you "tar" + "xz" a directory of files the large(r than one image) memory region xz will look at to find repeating byte sequences kind of does what you ask for, but repetitions of precisely the same bytes are rarely found in raw images, so don't expect much from that.

No general-purpose non-lossy compression (like zip, gzip, xz) will get you as good compression for images than compression algorithms specifically designed for a certain kind of media files.

You could theoretically encode all your images into one H.264 file in "lossless" mode, and decompress by extracting single frames from the encoded files.

But I'm not sure that's worth the hassle in comparison to e.g. just using webP lossless compression on each image, independently.