Does
-evaluate-sequence actually have
multiply as a option?
What is definitely a bug is that
-list currently does not list the options for
evaluate-sequence That is
-list list does not list
EvaluateSequence as a possibile option to
-list.
But enough of this side topic... Is
jkop happy with the solution to use
-flatten with a
-compose multiply as the solution to his blending problem.
jkop wrote:The batch of images are black and white and depict the shadow-shapes cast by a group of buildings: one shadow every 6th minute, from 8 am to 6 pm, which means 100 images of shadows, each slightly different from the other because of the rotation of the earth. Now if I superimpose them in such a way that each image has only 1% opacity (i.e. almost transparent), plus disregard the white in each image, then the areas which are overlapped by all 100 shadows will be 100% dark, whereas other areas will be grey in various degrees depending on the amount of overlaps. The resulting superimposition can give me a hint on where to put plants, or trees, and where I can minimize areas which would recieve no sun light.
Actually looking at his later description. What is really asking for is a 'subtract 1%' for every black areas in each image, which is exactly a
linear_burn using negated images does! Note that the shadow images, as he described them, are already negated in that we are subtracting the black component from the initial white canvas, so it is a straight forward solution.
Code: Select all
convert shadow_*.png +level 99%,100% \
-compose linear_burn -background white -flatten \
shadow_merge_result.png
Of course the only real test is with the actual images involved.
The other way to look at this, would be to negate and scale all the shadow images (white on black), add them together, and negate the result.