Compression — common mistakes

When an exam question asks you to recommend a compression type, justify your answer based on the data. If the data is a program, financial records, or a medical scan, you must choose lossless and explain that data integrity is paramount. If it's for streaming a movie or sharing a holiday photo on social media, lossy is more appropriate because file size and transmission speed are more important than perfect fidelity.

Pay close attention to units! Questions will specify bits, bytes, kilobytes (KB), kibibytes (KiB), megabytes (MB), or mebibytes (MiB). Remember: $1 byte = 8 bits$, $1 KiB = 1024 bytes$, $1 MiB = 1024 KiB$. Always show your conversion steps in your working.

Because for certain types of data, any loss of information is unacceptable. For example, a compressed text document must be perfectly reconstructable, as changing even one character could alter the meaning. The same applies to software executables, financial data, and scientific measurements where absolute precision is required.

Yes, but the results vary. Applying lossless compression (like ZIP) to a file that is already losslessly compressed will yield very little or no extra size reduction. Applying further lossy compression to an already lossy file (e.g., re-saving a JPEG) will reduce the file size further but will also cause additional, cumulative quality degradation.

Yes. For lossless compression, the limit is determined by the amount of redundancy in the data. A file with completely random data cannot be compressed losslessly and may even increase in size. For lossy compression, you can theoretically keep compressing until the file is tiny, but it will become completely unusable as all meaningful data is destroyed.

Consider the purpose. Is it for archival? Use lossless (e.g., FLAC for audio, PNG for images). Is it for streaming or sharing online where file size is critical? Use lossy (e.g., MP3/AAC for audio, JPEG/WEBP for images, H.264/H.265 for video). The key is balancing the need for quality against the constraints of storage and bandwidth.