git-big-binary.sh

git init
# Create a big file of random data (50 MB)
dd if=/dev/random of=bigblob.dat bs=5000000 count=10

git add bigblob.dat
git commit -m "Add big binary"

du -sh .git # about 48M

# Should be one big file in here:
ls -lRh .git/objects/[0-9a-z][0-9a-z]

# Duplicate the blob and add to repo:
cp bigblob.dat bigblob2.dat
git add bigblob2.dat
git commit -m "Add second big binary"

du -sh .git # still about 48 MB

# Still only one large object
ls -lRh .git/objects/[0-9a-z][0-9a-z]

# Look at the object hashes in git whatchanged - the'll both be the same
git whatchanged

# Get the hashes directly this way
git hash-object bigblob.dat
git hash-object bigblob2.dat