JPEG与PNG是否支持元数据/辅助数据压缩？

Great question—let’s break this down clearly, since metadata compression is a bit of a hidden corner of image formats!

原生格式对元数据压缩的支持情况 #

First, let’s cover what the JPEG and PNG standards natively support:

• PNG: This format actually has built-in support for compressed metadata. The standard defines three text metadata chunks:
  • tEXt: Plain, uncompressed text (the default for many tools)
  • zTXt: Text metadata compressed with the DEFLATE algorithm (same as PNG’s image data compression)
  • iTXt: Internationalized text metadata that also supports DEFLATE compression
    So if your PNG uses zTXt or iTXt chunks instead of tEXt, your metadata is already being compressed.
• JPEG: Traditional JPEG standards (like baseline JPEG) don’t include native compression for standard metadata fields (Exif, XMP, IPTC). These are stored in uncompressed APP segments (e.g., APP1 for Exif, which wraps a TIFF structure—while TIFF supports compression, Exif implementations almost never use it). That said, newer variants like JPEG XL do support compressed metadata, but that’s a separate format from classic JPEG.

Why most compression tools remove metadata instead of compressing it #

You’re right that most tools strip metadata by default—here’s why:

• Compatibility risk: Older image viewers, editors, or embedded systems might not properly parse compressed metadata (e.g., PNG’s zTXt or custom compressed JPEG APP segments). Stripping metadata avoids any chance of parsing errors or broken images.
• Implementation simplicity: Removing metadata is a trivial operation (just skip over the metadata segments/chunks). Compressing it requires identifying valid metadata fields, applying the right compression algorithm, and rewriting the file structure correctly—more work for tool developers.
• Perceived minimal gain: For many users, stripping metadata already cuts file size enough, and the extra savings from compressing metadata (even if it’s 16% of the file) isn’t worth the effort or compatibility tradeoffs.

Can you compress metadata instead of removing it? Absolutely! #

If you want to keep metadata while reducing its footprint, here’s how to do it:

• PNG-specific tools: Use utilities like optipng or pngcrush with flags that preserve and compress metadata. For example:

  # Optipng: Keep all metadata and use maximum compression for zTXt chunks
  optipng -keep -z9 input.png -o output.png
  

  These tools will automatically convert uncompressed tEXt chunks to compressed zTXt where possible.
• JPEG workarounds: Since classic JPEG doesn’t support native metadata compression, you have two options:
  1. Use modern tools like jpegoptim with flags to preserve metadata (some versions will attempt to compress the underlying TIFF structure in Exif, though results vary).
  2. Build a custom workflow: Extract metadata (e.g., using Python’s piexif library), compress it with DEFLATE or another algorithm, then embed it back into a custom JPEG APP segment. Note that this will require specialized tools to read the metadata later, so it’s only viable for controlled environments.
• Custom scripts: For full control, write a small script to handle metadata extraction, compression, and reinsertion. This works well if you’re dealing with a batch of files and need consistent behavior.

内容的提问来源于stack exchange，提问作者matt.baker