Agree with @pisum here wrt the issues at hand.
One more possible issue at hand here is that I believe the two images submitted for comparison were cropped for comparison of detail. I tried to duplicate the process and found when I took the submitted online observation original size and cropped it to a comparable size, I actually had to stretch it to get it to the size of the second image that was shown by the original poster - this further distorts the quality of the image. Maybe I’m missing something but that is the result I got.
Personally for speed of upload I tend to crop in close and submit my file size with a max of 1200 x 1200 and feel my images are fine and I retain a larger file size for future print purposes - 2048 x 2048 pixels is the max resize limit of the system. https://forum.inaturalist.org/t/notify-user-when-new-observation-matches-time-stamp-of-pre-existing-observation/253/10
@ tiwane stated: If it is larger than 2048 x 2048 pixels, it is resized so that its longest edge will be no longer than 2048 pixels. On mobile devices this is done before an image is uploaded (to save on data use) and on the web, this is done after the image is uploaded.