Pixel precision of the "Make Montage" command ?Bug ?
I'm using ImageJ to generate a panel for a figure in a manuscript. I have a
RVB stack of three channels, 1200x1200 pixel (no scale) images and I use the
"Make Montage" with 1 row, 3 columns, scale 1, First Slice 1, Last Slice 1,
Increment 1, Border Width 20, both options (Label Slices and Use Foreground
Color) unchecked (OSX 10.4.11, ImageJ 1.39u, Java 1.6dp).
I get a 1210x3610 panel, so I assume that some borders have been added to
the images because otherwise the size would be 1200x3600 if the borders were
just applied by "drawing" on the images. But if borders were simply added as
spaces between images, final should be 1200*3+4*20 and 1200+2*20, which
gives 3680x1260. So I started measuring exactly what had been cropped and
what had been added. At the left border and in the two middle borders, 20
pixels have been cropped from the corresponding images. But on the right
border, only 10 pixels of the right image have been cropped, and 10 pixels
have been added, so the final width of the image is 1200*3+10=3610. For the
height, it is the same : 20 pixels have been cropped from the top of the
images, but only 10 from the bottom, plus 10 pixels added, so the final
height is 1210 pixels.
This does not seem to be coherent, as it seems that the last (bottom and
right) borders are not correctly handled. The right image has 10 pixels more
than the left and center images, which is not very good to make a
multi-channel panel in a figure... Is it possible to check the source of the
Make Montage command to see if there is a bug there ? In my opinion, the
best would be not to crop the images, but to add the extra border spaces.