Don't underestimate the power of Gimp...
This manual describes the many functions, plug-ins and options Gimp has to offer, but it doesn't describe how to create great digital art or designs. There are probably as many Gimp tricks and tips as there are Gimp-users, and even if we wanted to, we couldn't include them all in this book. We can't teach you how to be an artist, but we've included a few examples here that will hopefully inspire you to new ideas and help you on the way to getting the best out of Gimp. This is more a gallery than a tutorial, and the object of this chapter is not to give detailed instructions, but rather to demonstrate the great versatility and power of Gimp to beginners, and maybe give an insight on new ways of using Gimp to more experienced users.
so, let's unleash the power of Gimp and Unix!
You don't always need to import a photo, drawing or 3D-image of an
object. There are many ways of creating astonishingly convincing image objects
directly in
Gimp.
To create the emblem statue I started out with a b/w photo. The background was isolated with the bezier tool, removed and replaced with a gradient fill. The image was rotated to vertical position, and tinted yellow with Image/Hue-Saturation. I reopened the selection made for changing the background, inverted it and used it again to isolate the figure, which was copied to a white layer. The figure was further adapted to supply an interesting surface for the Distorts/Twist plugin to work on. Finally, the yellow layer was applied in Difference Mode on top of the twisted layer.
The gold emblem was created by running the Gradient Map filter on the twisted figure (using the custom Golden gradient). The pale outline of the non-twisted parts was painted with low opacity airbrush and blurred within a sharp-edged selection. The blue ghost emblem was made by three copies of the original yellow-tint image. The middle copy was twisted, trimmed and set to Saturation Mode, and the top layer was set in Difference Mode. To accentuate the water- or ghostlike appearance, the original twisted shape was added to a layer, desaturated and set to Overlay. The side parts needed to be more visible so they were pasted separately in Multiply Mode. Finally a pale fluorescent shape of the figure was added to enhance the shape of the woman inside the waterwheel.
The pen was made by filling selection shapes with different gradients.
In this case, I used Bilinear FG to BG with medium opacity, and
also a number of FG to transparent gradients on several cylinder
shaped selections. The metal pen tip was adjusted with
Brightness-Contrast to achieve the metallic look. The pen shadow,
as well as the emblem statue shadow, was made with the Perspective Shadow
Script-Fu, cleared from color and filled with a FG to transparent linear
gradient.
To make the multi-color splash, I started by drawing a simple black
sun-shape on a white background, using a medium pencil tip. I then applied
the Distort/Value Propagate plugin, choosing more white.
The result was blurred and bumpmapped and the background
was cut away. A copy of the splash was pasted to a transparent layer, filled
with a colorful pattern and set to Lighten Only
Mode.
The carved form pressed steel text was generated by using a deliberately jagged (not antialiased) font. The text was filled with a black/white shapeburst gradient, the tonal range was adjusted with Image/Levels and the whole thing was bumpmapped. Final touch with Image/Brightness-Contrast. A gray glow was added to emphasise the shape of the letters, by copying the text layer, filling it with gray and applying Gaussian blur (with Keep Trans. unchecked).
The leaf pattern was created in Render/IfsCompose, and Artistic/Apply canvas was added. The image was blended with the background with the help of a Layer Mask and a radial gradient.
Glass and water effects are usually one of the hardest things one
can try to set about, but with a little help from Gimp, you're halfway
there.
For the Wet Unix label I used a photo of waterdrops on a solid blue background. I desaturated it and made a duplicate to create a highlight and a shadow layer. This was achieved by adjusting the tonal range with Image/Levels as described in page x. To create the illusion of water, I needed to displace the background where the drops were, so I opened a new Channel and painted a mask for the drops. The channel selection was loaded on a black layer which I called Displace layer, and the drop shapes were filled with a b/w shapeburst gradient..
I used a nice, clean photo of a mountain top as background image, and added a little fog with a FG to Transparent gradient. To stylize and add some wetness to the image I created a rain layer. This layer was filled with the custom Rain pattern, darkened somewhat and set in Addition Mode. A text layer was also added. Then I ran the displace filter on the text and the mountain background, using the Displace layer with the drops.
To make the label fit the bottle shape, I made a new displacement map with the gradient editor (dark to the left and right and bright in the middle to make a round displacement). After displacing the label, the displacement map was used to add a metallic sheen to the label by setting it in Overlay Mode. The final adjustments were made with the Transform/Perspective tool.
The bottle was made from a photo of a decanter filled with dark red wine. This was a bit troublesome because I wanted the bottle to be empty, or at least filled with some transparent liquid. The wine glass which was visible behind the bottle was easily cloned away, but the rest was left alone, most of it would be covered by the label anyway.
The bottle was cut out, rotated and pasted to a transparent layer. A highlight and shadow bottle was created with Levels, (but not desaturated) and the highlight bottle was allowed to keep a quite large range of shades, otherwise the reflections in the glass would look too hard and unnatural.
To cover the flat and dull looking part of the bottle (where the wine was) the bottom part was recreated by cloning different parts of the highlight bottle and set this layer to Screen Mode. The shadow bottle was set to Multiply.
The background, a photo of a lake, was blurred to create the illusion of distance, and to keep the focus on the bottle. The highlight bottle was used twice as a displacement map on the background so that the rocks in the background would appear to be distorted through the curved glass.
Because the lake image was originally indexed, the sky looked banded and ugly. This was rectified by feather selecting the sky, and replacing it with a linear gradient. A few clouds and a sun reflex was also added.
The water reflection was made by flipping the image of the merged bottle to a copy of the lower part of the lake, blurring it, adding some ripples with Distort/Ripple and lowering the opacity level. The waterline (where bottle meets water) was a little trickier and had to be painted by hand in a couple of Overlay layers, and some water glitter was created with the sparkle filter.
A white haze was added to the foreground, and of course, a couple of Larry Ewing's adorable Linux penguins.
Because black is transparent in Screen mode (also Addition and Lighten
Only) black pencil strokes drawn on a white layer will reveal the image
underneath, just as if you had sketched all by hand. This is an easy way
of creating convincing pencil/ink drawings from a scanned photo.
There are several commercial plugins called things like charcoal, crayon or ink drawing that supposedly achieve these things. Certainly, these products can produce nice artistic outputs, but they never come close to the results you get using this method. Naturally, the quality of the final image depends a lot on the drawing in the screen layer, so this isn't an "instant artist" trick. A general advice is to reduce the number of shades in the background, otherwise too much of the underlaying image will show, and this will spoil the illusion. One way of improving coarse computer drawings is to use the Value Propagate filter and set it to more white. You can also create a crayon or charcoal look to an image by displacing or warping the pen strokes with a suitable displacement map, or just by using unusual brushes.
A very simple way of creating drawings from scanned photos is of course to use one of the Edge-Detect filters. Running Sobel on a duplicate results in a transparent layer with a black outline of the image object. Having done this, it's easy to paint the underlaying layer in large clean areas, and you'll get something very similar to a picture in a comic book.
To make the pencil drawing, a white layer was placed on a b/w photo. It was set to Screen mode, and the opacity was temporarily reduced, so that the background would be visible. The sketch was drawn with a small, sharp pencil tip, and made to follow the contours and shapes in the photo. The photo's tonal range was limited by using the Image/Posterize filter, and the sketch layer was displaced slightly with a canvas structure as map. The image was flattened and adjusted with Brightness-Contrast to get the right gray value of a pencil drawing.
The sepia ink drawing was created from the previous drawing (pencil). Color was adjusted with Hue-Saturation and Brightness-Contrast to a sepia-like quality, and a beige "sketch paper" layer was added in Multiply mode. The "white crayon" in the sketch paper layer was painted with the airbrush and several soft edged brushes.
For the crayon drawing I used coarser, rugged brushes. I also displaced the sketch layer twice to get the right scratchy crayon or charcoal look. The image was flattened and the contrast was increased with Levels.
I made a duplicate layer and used the Dark 1 gradient in the Gradient editor to map to the image (Colors/Gradient Map). I put the old layer on top of it, set it to Multiply and erased everything except the contours, and parts which I wanted to keep dark (like the baby's eye and ear).
In the top layer I placed the posterized photo as Darken Only, changed the color to violet, and applied some motion blur. This looks somewhat like watercolor, if it's only used in small areas of a composite image.
These special effects can provide that little extra which can make
a good image
great.
To create the Electric Horseman, I started out with a photo of a rodeo rider. The horse and rider were selected with the bezier select tool, and pasted to a transparent layer in a new image. The horse's halter, mane and tail was cloned away or erased, and the rider was selected separately (using a little feather) and saved as a copy in another layer. The horse and rider layer was duplicated twice, and those copies were desaturated and adjusted with Levels to create highlight/shadow layers.
To create the leopard skin, two more copies were made. In the first
copy, the entire horse shape was filled with the Leopard pattern from the
Patterns dialog box. Since this pattern isn't entirely seamless,
it was adjusted by painting yellow and black spots in the visible joints.
The leopard layer was set in Darken Only mode over the second copy,
which was filled with a yellow-orange gradient to add color depth to the
leopard. To add some extra glow to the leopard horse, one more texture layer
was added. This time I used the black and white cow pattern, and set it to
Overlay. Note that this does not make the horse look like a cow
(!). In Overlay the large dark spots on white background rather gives the
illusion of powerful muscles under a shiny coat. Now I turned to the layer
with the rider, and changed saturation, brightness and contrast to make it
fit the new
"horse"
The glow layer was made by filling a feathered horse and rider selection with red, yellow and white, each time with lower feather values, and set the layer to Screen mode.
The illusion of movement was more complicated, because just adding some motion blur wouldn't suffice to create the subtle effect I wanted. I had to create two different motion layers - one dark and one light. For the Dark movement layer Blur/Motion blur was applied to a copy of the cow glow layer, then I did the same for a copy of the orange gradient layer, but this time I inverted the selection, and only the blurred parts outside of the horse was used. These layers were merged after adjusting the opacity. The dark movement was still a bit to strong in some parts, so a layer mask was used to tone down or remove motion glow where it wasn't wanted. This layer was set to Darken only. For the Light movement layer I blurred the shadow layer and the leopard layer (as with the orange gradient layer before) merged it and adjusted to darker. To protect the rider figure from too much blur, a layer mask was used here as well. This layer was set to Screen mode.
The background was made of a photo of a blue sky, a city panorama by night and a yellow evening sky. The blue sky image was transformed to dark clouds with Image/Hue-Saturation, and the yellow sky was merged to the city panorama. The flash of lightning was a bit harder, because the only lightning image I had was the Lightning pattern in the Pattern dialog, and that was too repetitive or intertwined with lightning to be used directly. The problem was solved by scaling an image with lightning pattern and then use the Map/Fractal Trace plugin. From that image I could feather select a suitable part, and adjust it with Transform/Perspective and Screen mode to look the way I wanted.
There are many ways of blending images, but for advanced montages,
the most versatile method is to use different layer
masks.
For the Chevelle montage, I used a drawing from the 1966 Chevrolet Chassis Service Manual for background. To make it less dominant, it was inverted, blurred and noise was added. Then color and brightness was changed to a soft "old looking" sepia tone.
The main collage element was a holiday snapshot of me and our Chevelle. This image was blended to the background with the help of a layer mask. I started by making a round vignette shape with a radial gradient. This masked out soft and nice, but it also meant that the gradual transparency was evenly distributed. To put an emphasis on the person rather than the car (I like to keep it that way) some of the mask was painted white to protect face and especially the arms from transparency.
The vignette was still a little bit weak, and too smooth at the edges, so
I made a duplicate where the layer mask was brightened up with
Levels, and noise was added to the mask (not the image). This made
the top layer blend well with the spotted background, as well as with the
smooth copy
underneath.
The next element in the montage was the fuel pump assembly. Here,
noise was added toboth mask and image. The Glow
Mask was made with a feathered lasso selection which was filled with black
and heavily blurred before applying the noise. Finally, the image was tinted
with the same ochre or sepia tone as the
background.
The steering wheel was treated in a similar manner, only this image was taken from an old magazine, so no noise was added (it was already quite noisy).
The picture of the car (from the 1966 Chevrolet dealer album) was desaturated
and tinted. To create the illusion of speed, linear motion blur
was applied to an inverted selection of the car, then a simple layer
mask was added to blend the car with the background. To make the car
stand out more in the composition, contrast and brightness
value was
increased.
Two photographs from the service manual (removal of drive plate from generator and disassembly of crankshaft gear for the camshaft transmission) were pasted to a layer with low opacity.
To make the Chevelle Malibu SS text, I used three different layers. The
"Chevelle" layer is supposed to look sharp and close to the observer, while
the "Malibu" layer lies deeper, or further away from focus. This was accomplished
by placing the Malibu layer very close to (almost under) the Chevelle layer,
tinting and blurring it and placing a very soft shadow on top of
it.