I'm working with millions of small images (~100x100) of different sizes.
If I store them as jpgs on a harddisk, they would exceed my disk's inode limit.
If I store them as binary files like HDF5, they would take up >100GB even when I apply compression (h5py's gzip lossless compression is nowhere as compact as jpeg's compression).
Are there any standard ways to store these images as a single file with jpeg compression so that it'll neither take up lots of inode or harddisk space? I'd also like to read these images easily through python.