Hola. Salut. Hi.
https://confluence.atlassian.com/confcloud/configuring-attachment-size-724765525.html
The default setting for pictures in Confluence is 100MB as defined from the Product Support team at Atlassian in above link.
The pixel calculation from any/all image editors will calculate your resolution in a variety of formats. The bigger and more pixels/dots per inch the heavier/larger the image.
The way I understand our current networking/DBs/programming is that it will shrink the image to a thumbnail for storage and then expand the image as needed. I believe that this sequencing on the math to make priorities in necessary bandwidth/storage/speed? It also might be that the method of inserting images could cause issues with accuracy in the long run.
Example: tool bar insert versus copy/paste versus link etc
The reason the copy/paste drop/drag method is still being used is the functionality worked with WORD before and most editors used that tool before moving to the luxurious Confluence. I found that copy/paste with the mouse is a tool for limited use because of accuracy with the parsing of the data. CTRL C & CTRL V seems to be more accurate.
Also, this issue seems to be relevant to users as images sometimes change direction.
Any developers want to chime in on any of this R & D on this for Confluence?