XnView Software

With JPG lossless function, cropping a tiny number of pixels only from the right or bottom edges seems to work better than left or upper edges. So if you have a 1027x769 jpg, it can be chopped to exactly 1024x768 more easily from bottom or right edges. This seems like a feature and not a problem, so I hope this info is useful.

I guess that this has something to do with the IDCT algorythm of the JPEG-format, which allows to ignore some pixels on the bottom-right easier than at the top-left. I guess top-left pixels are processed first, but does this explain this behaviour??? Don't no.

While we are at, just some basics about the JPEG-format every moderate user of XnView should know:

-File is not saved in RGB, but in luminousity, saturation and color value (color vectors are first translated)
-Image is patterned in equally sized squares for luminousity, saturation and color value
-Okay, I spare you the details about DCT and IDCT (2 dimensional triangular wave functions (cosines))
-Saturations and colors are saved as IDCTs with size of 16x16 pixels, whereas only luminances are saved as more detailed IDCTs with size of 8x8 pixels
=> because you notice more details in luminance than in saturation or color
-Don't wanna explain Huffman table building...

Zooming in can also make area selection more precise. The new v.1.82 will have a very useful "Do nothing" option in 'Swithching' modes section which can prevent accidental double-clicks when making area selection. This was a nice v.1.74 feature. Its back under a different name.