Actually, I find the examples given for this to be unnecessarily complicated for the typical extract operation. Here's the entire entry:convert -size 16000x16000 -depth 8 image.rgb[640x480+1280+960]' image.rgb image.png
The examples above produce multiple images. How about something simpler, like this?-extract width
-extract widthxheight{{+-}offset}
extract the specified area from image. The option is most useful for extracting a subregion of a very large raw image. Note, these two command are equivalent:
convert -size 16000x16000 -depth 8 -extract 640x480+1280+960 image.rgb image.png
convert -size 16000x16000 -depth 8 image.rgb[640x480+1280+960]' image.rgb image.png
The option's form -extract width does not give an extracted image for me, just a resized one. I'm not sure what such a form should reasonably be expected to do, though.
convert -extract 640x480+1280+960 bigImage.rgb extractedImage.rgb
convert 'bigImage.rgb[640x480+1280+960]' extractedImage.png
Finally, I am confused by the results of the following two commands, which I guess means I am confused about -geometry, in general:
The first gives something I expect, a horizontal extraction of dimensions 640x55 from the top of the logo, the width coming from that of the input image (logo, 640x480). So I'd like the second to give me a vertical strip with dims 55x480, but it gives an error, "convert: geometry does not contain image `LOGO'."convert -extract x55+0+0 logo: horizontalWiz.png
convert -extract 55x+0+0 logo: verticalWiz.png
If I use "48" in place of "48x" I get an " invalid argument" error.
Rick