By the way, this is also part of the reason why the OpenEXR (and MultiLayer OpenEXR) file formats were invented: to capture data, exactly, with none of the “manipulation” that is ordinarily applied to “an image file.”
Remember that “an image file” is designed to make life very easy for cheap display hardware. It has gamma-encoding and lots of other things. Plus, it’s sandwiched into a particular numeric range. Because it’s meant to be displayed. Per contra, the EXR file format is designed to capture a numeric data-set. Every floating-point number that the renderer produced, exactly as the renderer produced it. (Think of it as “a renderer’s RAW file …”)