Will I be able to do this with ImageMagick

[bobthebuilder]

Hello Everyone

I have discovered ImageMagick through my php development and would like to know if ImageMagick offers sufficient control to solve my problem.

I have a list of jpg files, together with a list of (x,y) coordinates for these files, and need to construct a master file that puts all these images in the file at the required coordinates. Reading through the documentation, I can't work out out yet whether montage (for example) would let me set the coordinates of each image, or whether it deals with just the overall image dimensions.

So my question is, can ImageMagick actually do what I need it to do, or is this asking too much and should I look elsewhere?

Many thanks

Bob

[fmw42]

You would need to write a script to overlay each image on a background at the correct location, if the number of images is large. IM can use a file with a list of images, but the commands you need have to have the coordinates manually specified unless you script it. Then the script can access the file of images and coordinates and loop over each one and place it at the desired location.

But if you only have a few images that can be specified in a simple list of images in one command line, then IM can flatten or mosaic a number of images into a larger image with each one specified to a given x,y coordinate in the composite image.

For new IM users good reading is at http://www.imagemagick.org/Usage/

For your specific problem, see

http://www.imagemagick.org/Usage/layers/

and especially

http://www.imagemagick.org/Usage/layers/#flatten
http://www.imagemagick.org/Usage/layers/#mosaic

[bobthebuilder]

Many thanks! I have had a play with the composite function and it seems I can do exactly what I want. I have become an overnight IM fan. I had been doing my thumbnail processing with php/GD and thought I would also combine my images in GD, but the memory requirements were ridiculous. Now it seems I can generate the final image off-line running these commands in a script.

Thanks again.

[anthony]

You may be interested in the examples shown at
http://www.imagemagick.org/Usage/layers/#example

this has a look generating images and piping it in a stream to a separate image machine 'layering' command.

[bobthebuilder]

Thanks Anthony. I am still learning my way around IM and all these tips are very useful. Initially I was doing stuff like this:

Code: Select all

convert -size 6733x4607 xc:white -fill white -stroke black^
 -strokewidth 2 -draw "rectangle 1,1,6732,4606" al1_main_design.gif
 
composite -geometry +295+340   small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +1559+340  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +2823+340  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +4087+340  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +5351+340  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +295+1377  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +5351+1377 small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +295+2414  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +5351+2414 small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +295+3451  small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +1559+3451 small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +2823+3451 small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +4087+3451 small_ollie.gif al1_main_design.gif al1_main_design.gif
composite -geometry +5351+3451 small_ollie.gif al1_main_design.gif al1_main_design.gif

and it was taking forever. Presummably that was very inefficient as it allocated an image at the start that was as big as I would need at the end of the process and kept having to process the big image. Now I am doing stuff like this:

Code: Select all

convert -page +295+340 small_ollie.gif -page +1559+340 small_ollie.gif^
 -page +295+340     small_ollie.gif^
 -page +1559+340    small_ollie.gif^
 -page +2823+340    small_ollie.gif^
 -page +4087+340    small_ollie.gif^
 -page +5351+340    small_ollie.gif^
 -page +295+1377    small_ollie.gif^
 -page +5351+1377   small_ollie.gif^
 -page +295+2414    small_ollie.gif^
 -page +5351+2414   small_ollie.gif^
 -page +295+3451    small_ollie.gif^
 -page +1559+3451   small_ollie.gif^
 -page +2823+3451   small_ollie.gif^
 -page +4087+3451   small_ollie.gif^
 -page +5351+3451   small_ollie.gif^
 -page +1+1         line.gif^
 -page +6234+4603   line.gif^
 -background white -layers merge  +repage seb_a1.gif

where the image is incrementally increased in size, so the execution time is manageable (still 5mins, but I can possibly live with that). With more experience I hope to reduce that further. I have read that running commands from PHP can be quicker than running from the command line so hope to ty that later. Thanks for your input.

[fmw42]

I have read that running commands from PHP can be quicker than running from the command line so hope to ty that later.

It is certainly worth trying, but you might want to see some timing tests at viewtopic.php?f=4&t=16779

[anthony]

Thanks Anthony.
it was taking forever. Presummably that was very inefficient as it allocated an image at the start that was as big as I would need at the end of the process and kept having to process the big image.

Yes and as it was saving to gif all teh IM IM had to so a lot of work and an lot of IO.

Your second method will avoid much of that.

where the image is incrementally increased in size, so the execution time is manageable (still 5mins, but I can possibly live with that).

Actually IM -layers option will scan though the image and calculate the final image size once, and create it (with the current background color). It them composes each image at the virtual offset (page) given onto that larger image, and junk everything.

However one speed up is to NOT read that one image over an over. (assuming you are only composing that single image). But read it once and clone it. However you will need to change the page offset of the clone (already in memory) rather tha using -page to set it for an image being read in.

Code: Select all

convert  small_ollie.gif \
  \( -clone 0 -repage +295+340 \) \
  \( -clone 0 -repage +1559+340 \) \
  \( -clone 0 -repage +2823+340 \) \
  ....
  -delete 0 \
  -background white -layers merge  +repage result.gif

You save IO on reading the image multiple times, and will have far reduced memory needs. Basically all the images will use the same data store, as clone does not 'copy' image data (just the meta-data) until something actually tries to modify it which does not happen in this case.

PS if you know what the final size, and want the 'origin' (0,0 pixel), then you can just set the size using the first image. No need for the final repage in that case.

Code: Select all

convert  -page 6235x4604 small_ollie.gif \
     \( -clone 0 -repage +295+340 \) \
     ...
     -background white -layers mosaic result.gif  

The -layers merge is more for low level or relative placement, where you don't care about the space no image was overlaid in, or you are dealing with negative offset placements. It generates a layered image that can be used for further layer image work.

Read the other parts of the layered image examples.
http://www.imagemagick.org/Usage/layers/