chop with percentage affects other dimension

[muccigrosso]

If I use -chop with a percentage in only one dimension, the second dimension loses a pixel too. For example (and here I'm chopping the width, but the results are the same if I use the height):

Code: Select all

convert -format "%w x %h" rose: info:

yields: 70 x 46

Code: Select all

convert -format "%w x %h" -gravity east -chop 1x0 rose: info:

yields the expected 69 x 46, but

Code: Select all

convert -format "%w x %h" -gravity east -chop 1%x0 rose: info:

yields 69 x 45.

Why is the height getting chopped too?

It looks like even with a 0% chop, 1 pixel is always removed, as long as there's a percentage and not a number of pixels:

Code: Select all

convert rose: -format "%w x %h" -gravity east -chop 0%x0 info:

yields the same 69 x 45.

Version: ImageMagick 7.0.7-5 Q16 x86_64 2017-10-03 http://www.imagemagick.org
Mac OS X 10.11.6

[snibgo]

What version of IM? Please ALWAYS state the version of IM.

I can't find the thread, but a change was made recently, so that when a "%" was used in geometry, any result of zero was changed to one. This was to prevent "-resize 1%" etc from failing by trying to create an image of zero pixels.

I suspect that change has caused this problem.

[GeeMack]

I suspect that change has caused this problem.

Using ImageMagick 7.0.7-5 on Windows and running this command with the 640x480 built-in logo:...

Code: Select all

magick logo: -chop 0x0% info:

The output shows a single row of pixels has been chopped from both the width and height...

Code: Select all

logo: GIF 639x479 639x479+0+0 8-bit sRGB 256c 0.016u 0:00.004

Running the same command with a bit older ImageMagick 6.9.8-10 and "convert" produces the expected output showing "logo:" with the original 640x480 dimensions.

[muccigrosso]

What version of IM? Please ALWAYS state the version of IM.

I can't find the thread, but a change was made recently, so that when a "%" was used in geometry, any result of zero was changed to one. This was to prevent "-resize 1%" etc from failing by trying to create an image of zero pixels.

I suspect that change has caused this problem.

Sorry. Version added above.

And yes, that would do it! Guess I'll have to calculate the number of pixels then.