|
|
|
|
This filter uses a “displace-map”
to displace corresponding pixels of the image. This filter displaces
the content of the specified drawable (active layer or selection) by
the amounts specified in X and Y Displacement multiplied by the
intensity of the corresponding pixel in the 'displace map' drawables.
Both X and Y displace maps should be gray-scale images and have the
same size as the drawable
. This filter allows interesting distortion effects.
12.3.2. Activate the filter
You can find this filter through
→ →
12.3.3. Options (General)
-
Preview
-
Uncheck this option if your processor is slow.
-
Displacement Mode
-
You can choose working in
cartesian
coordinates, where pixels are displaced in X or Y direction, or
working in
polar coordinates,
where the image is pinched and whirled by displacing pixels in
radial or tangent direction.
Please see the next sections for details about these options.
-
Edge Behavior
-
These options allows you to set displacement behaviour on active
layer or selection edges:
-
Wrap:
With this option, what disappears on one edge reappears on
the opposite edge.
-
Smear:
With this option, pixels vacated by displacement are
replaced with pixels stretched from the adjacent part of the
image.
-
Black:
With this option, pixels vacated by displacement are replaced
with black.
12.3.3.1. Cartesian Displacement Mode
In both modes, direction and amount of displacement depend on the
intensity of the corresponding pixel in the displacement map.
The map, that should be a grayscale image, has 256 gray levels
(0-255), the (theoretical) average value is 127.5. The filter
displaces image pixels corresponding to pixels with values less than
127.5 (0 to 127) in map to one direction, corresponding to pixels with
values from 128 to 255 to the opposite direction.
-
X/Y Displacements
-
If the respective option is activated, image pixels
corresponding to pixels from 0 to 127 will be displaced to the
right for X, downwards for Y, image pixels corresponding to
pixels from 128 to 255 will be displaced to the left for X,
upwards for Y.
What you enter in input boxes, directly or by using arrow-head
buttons, is not the actual displacement. It's a coefficient used
in a
formula, which gives the pixel actual displacement according to
the scaled intensity
[]
of the corresponding pixel in map, modulated by the coefficient
you enter. Introducing intensity into formula is important: this
allows progressive displacement by using a gradient map.
This value may be positive or negative. A negative displacement
is reverse of a positive one. The value varies in limits equal
to the double of image dimensions.
When you click on the drop-down list button, a list appears
where you can select a displacement map. To be present in this
list, an image must respect two conditions. First, this image
must be present on your screen when you call filter. Then, this
image must have the same dimensions as the original image. Most
often, it will be a duplicate original image, which is
transformed to grey scale and modified appropriately, with a
gradient. It may be possible to use RGB images, but color
luminosity is used making result prevision difficult. Map may be
different in horizontal and vertical directions.
12.3.3.2. Polar Displacement Mode
-
Pinch
-
If this option is activated, the radial coordinates (i.e. the
distance to the image's midpoint, the “pole”)
of the pixels will be changed. Image pixels corresponding to
map pixels from 0 to 127 will be displaced outwards, image
pixels corresponding to pixels from 128 to 255 will be displaced
towards center.
For the values and the displacement map see above
(“X/Y Displacements”).
The displacement is independent from the polar distance, all
pixels are displaced by the same amount. So the image will not
only be stretched or compressed, but also distorted:
A 160x120 pixel image, plain white displacement map, and
displacement coefficient 20.0: this results in a 20 pixels
displacement towards center. This is a horizontal reduction in
size by 25%, vertical by 33%, and diagonal by 20%, so the image
will be distorted.
-
Whirl
-
If this option is activated, the angular coordinates of the
image pixels will be “displaced” by a map pixel
dependent amount. For a plain displacement map, the image will
be rotated, otherwise it will be whirled.
Image pixels corresponding to pixels from 0 to 127 in the map
will be displaced counterclockwise, image pixels corresponding
to pixels from 128 to 255 will be displaced clockwise.
For the values and the displacement map see above.
|
Note |
For a plain, non neutral map, if displace mode “Polar”
is enabled, this filter works like
Whirl and Pinch.
|
12.3.4. Using gradient to bend a text
Follow following steps:
-
Start with opening your image.
-
Duplicate this image. Activate this duplicate and make it
gray-scaled
( → → ).
Fill it with the wanted gradient. This image will be your
Displacement map, with the dimensions of
original image.
-
Activate original image. Create a Text Layer
with your text. Set layer to image size: right-click on the layer
in layer dialog and, in the pop-menu, click on “Layer to
image size”. Note that letters in text layer lie on a
transparent background; now this filter doesn't displace
transparent pixels. Only letters will be displaced.
-
Activate the text layer. Open the Displace filter window. Set
parameters, particularly the displacement coefficient, according to
the result in Preview. Click OK.
This method also applies to standard layers:
|
Tip |
To get the wanted gradient, first draw a black to white gradient.
Then use the Curves tool to
modify the gradient curve.
|
12.3.5. Displacement Calculation
The following section will show you how to calculate the amount of
displacement, if you are interested in these details. If you don't want
to know it, you can safely omit this section.
The overview example showed the X displacement using a coefficient of
30.0: 19, 8, 4, or 15 pixels, depending on the grey level of the
displacement map's color.
Why just these amounts? That's easy:
If you check these equations, you will notice that the values they give
are not exactly the results we retained in the example
(using non-integers, that's not surprising). So, were the results
rounded to the nearest integer and then the pixels were
displaced by a whole-numbered amount? No. Every pixel is displaced
exactly by the calculated amount; a “displacement by a fractional
amount” is realized by interpolation. A closer look at the
example image will show it:
The displacement causes small (one pixel wide) areas of intermediate
colors at the edges of plain color areas. E.g., the black area (zoomed
in image) is caused by a displacement of -4.12, so the intermediate
color is 12% black and 88% gold.
So if you select a displacement coefficient of 30.01 instead of 30.00,
you will indeed get a different image, although you won't see the
difference, of course.
|
|
|