Hi @ioppermann .
"Better jpeg" mode? This mode acts like a standard jpeg-encoder when creating a new file. But if there is a file (overwriting), quant tables are borrowed from an existing file. And if the sizes of the existing and saved images match, the procedure changes. Instead of encoding a block, the corresponding block of the existing file is decoded, and if the decoded block matches the stored one, the dcts of this block are borrowed, otherwise standard encoding is performed.
Example:
modjpegover --im test.jpg --ow test-mod.png
test.jpg and test-mod.png is exist files.
See also: https://betterjpeg.com/
Hi @ioppermann .
"Better jpeg" mode? This mode acts like a standard jpeg-encoder when creating a new file. But if there is a file (overwriting), quant tables are borrowed from an existing file. And if the sizes of the existing and saved images match, the procedure changes. Instead of encoding a block, the corresponding block of the existing file is decoded, and if the decoded block matches the stored one, the dcts of this block are borrowed, otherwise standard encoding is performed.
Example:
test.jpgandtest-mod.pngis exist files.See also: https://betterjpeg.com/