In their tests, shouldn’t the input be some kind of RAW file instead of a jpeg?
Otherwise, how do we know the compression gains don’t come from the artifacts?
It probably does. But, regular lossless compression tools, like gzip, don’t compress compressed images, like jpegs, very well. This tool is explicitly for losslessly compressing lossily-compressed images, and doing it very quickly.
I tried to discuss this with the developer. I’m not sure what the takeaway is.
https://github.com/catid/Zpng/issues/3
In their tests, shouldn’t the input be some kind of RAW file instead of a jpeg?
Otherwise, how do we know the compression gains don’t come from the artifacts?
It probably does. But, regular lossless compression tools, like gzip, don’t compress compressed images, like jpegs, very well. This tool is explicitly for losslessly compressing lossily-compressed images, and doing it very quickly.
I tried to discuss this with the developer. I’m not sure what the takeaway is.
https://github.com/catid/Zpng/issues/3