Pandas warns during concat because it's changing how it handles object columns containing only True and False values they were previously treated as booleans automatically, but this is being deprecated. You can fix this by either using df.infer_objects() which won't handle NaN values well or explicitly converting with df[col].astype('boolean') (better when you have True, False, and NaN). For non-numeric data like image thumbnails, size mainly impacts memory usage rather than operation performance.