I have this problem with all jpg files:
At a 1200px long edge jpg file at 73% C1 11 estimates output size 139kb, but the actual output size is 795kb - more than six times as large.
Hi,
I don't know why there is such a discrepancy between the estimated and actual file sizes. The only way to get a somewhat reliable estimate is through iterative trial and error, which takes time. So I can only assume that C1 uses a heuristic instead of iterative trial and error. To reduce the final file size, you'd obviously have to reduce the quality setting further.
You can try to reduce the file size by starting with a 100% quality, and then reprocessing the JPEGs with a program like
JPEGmini.
What's wrong with jpg processing in C1?
Not much, but the size prediction could be improved. The JPEG quality is apparently pretty good, because only quality settings of 75% or more can be usably reduced (say 25% or more), but lower settings cannot be reduced much without visibly losing quality.
If other applications compress much more than e.g. JPEGmini or similar are willing to do, one could wonder about the resulting quality.
Cheers,
Bart
P.S. It is possible to further reduce JPEG sizes significantly but with a minor loss of quality. Google has developed a process called
Guetzli. But that process is
very very slow (up to 2 minutes per image in the implementation that I'm using, and it strips all metadata), so mostly usable for image resources on Websites which results in considerably faster loading pages and lower bandwidth requirements.