Get original image size after decoder hinting
Swiss army knife of image processing
Brought to you by:
bfriesen
Menu
▾
▴
#29 Get original image size after decoder hinting
Hello
I'm using decoder hinting while reading image to make resizing faster:
image.read(osize, inputUri)
then I apply crop
image.crop(cropGeom)
and
image.resize(osize)
The problem is that my input image is 2560x1600 and after reading with hinting it becomes 640x400 but the cropGeom that I pass is built against 2560x1600 and it failes with error: "Magick: geometry does not contain image (unable to crop image)".
How can I resolve my issue? By getting rid of decoder hint? Or maybe I could access original image size (2560x1600) i respectively modify cropGeom?
Thank you for your help.
Related
On Fri, 4 Mar 2016, Przemysław Sobala wrote:
Are you using the transform() method? This is the only method I can
see which accepts both a size and a crop geometry. Otherwise, it
seems that the problem is that you are using arguments which don't
match the image which was actually read.
After the image has been read using the Image read() method, you can
now obtain its current size using the Image columns() and rows()
methods. Build any necessary crop() arguments on the current image
size. If the crop arguments are larger than both of the image
dimensions, then you will obtain the error you reported.
Bob
--
Bob Friesenhahn
bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
Related
Support Requests:
#29I have a mechanism to detect faces in images. I'm using it to get coordinates and size of human faces in the image. Then I'm using those coordinates and size to cut them out of the image using a crop method and then resize the cut parts.
I'm building crop() geometry based on original image dimensions (eg. 2560x1600) but after
image.read(Geometry(400,300), inputUri)Magick++ API reports that input image has size 640x400 and my crop geometry is wrong, because it's based on orignial size 2560x1600.To sum up: my workflow is:
1) detect face in input image of size 2560x1600
2) read input image with hiting (400,300) - GM reports image.columns() x image.rows() = 640x480
3) crop(225x225+295+487) - error! "geometry does not contain image (unable to crop image)" and i can agree with that
4) resize(400,300)
5) write
On Fri, 4 Mar 2016, "Przemysław Sobala" wrote:
I am not completely understanding what you are looking for, but if you
need the original dimensions of the JPEG input file, you should be
able to obtain it from the Image baseColumns() and baseRows() methods
immediately after reading the file. This should match your original
size.
Bob
--
Bob Friesenhahn
bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
It's the opposite; they need the crop geometry in terms of the reduced image. In this case, the hint (400x300) results in a 640x400 image (original scaled by 1/4). So the cropping size and offsets need to also be divided by 4. The scale factor is predictable; it's the smallest 1/power-of-two scale that is large enough to contain the "hint" size.
Either of the solutions mentioned in the original post should work (don't "hint", or record the original size and recompute the crop geometry.
Last edit: Glenn Randers-Pehrson 2016-03-04
Thanks to both of you. I scaled the cropping geometry, like Glenn said, based on
image.base{Columns,Rows}(), like Bob suggested.That solved my problem.
Have a nice day.