In recent years the use of transparency has become more and more common in media production. There are three different categories of image documents with transparency:
Raster image files (e.g. TIFF)
Native page layout application documents (e.g. Adobe InDesign)
PDF 1.4 and later documents
PDF 1.4 and later documents are supported by PDF HandShake. The support of ImageServer for native page layout application documents consists of the InDesign plug-in to generate previews of native InDesign documents (see 5.8.1 “The HELIOS Preview plug-in”) and of the QuarkXPress XTension to generate previews of native QuarkXPress documents (see 5.7.3 “HELIOS Preview XT”). The following text explains transparency in raster images only.
There are several different techniques to create transparency in raster images:
Clipping paths
Alpha and mask channels
Transparency channels
Multiple layers
The most common way of creating transparency in prepress production uses clipping paths. Clipping paths isolate foreground objects and make their background transparent. Clipping paths are vector-based line structures which create hard edges between foreground objects and transparent background.
Some image formats support alpha or mask channels.
In all raster images each pixel is represented by a set of numerical values for all colorants of the image, depending on its color space. If an image contains an alpha channel, each image pixel is represented additionally by a numerical value which specifies the opacity of that pixel. In alpha channels, 100% specifies an opaque pixel and 0% specifies a transparent pixel. Mask channels are exactly like alpha channels except for their polarity. In mask channels, 100% specifies a transparent pixel and 0% specifies an opaque pixel. Alpha and mask channels offer multilevel transparency for drop shadows and soft edges of objects, also known as feather effects. If an image format supports alpha and mask channels, any image of that format may contain zero, one or multiple alpha or mask channels. The extra channels can be identified by their unique names. If an image contains one or more alpha or mask channels, the user has the option to apply none of the extra channels or exactly one of them.
Transparency channels are much like alpha channels. Raster image formats supporting transparency channels can contain exactly one or no transparency channel. The user usually does not have the choice whether to apply the transparency channel or not. The application of a transparency channel is mandatory. Each image pixel is represented by a set of numerical values for the visible colorants and one additional numerical value containing the opacity information. 100% specifies an opaque pixel and 0% specifies a transparent pixel. Transparency channels offer multilevel transparency, too. A well-known example of an image format with transparency channels is PNG with RGBA color.
The most complex way of creating transparency currently is to combine multiple layers in Adobe Photoshop. Only TIFF and Photoshop native documents support multiple layers. TIFF images always contain a main image which is a combination of all Photoshop layers. But Photoshop native images contain a main image which is not necessarily a combination of all Photoshop layers. If ImageServer reads an image with multiple Photoshop layers, it uses the main image exclusively.
The following table lists the transparency capabilities of the most important image file formats:
Raster image format | Clipping paths | Alpha/Mask channels | Transparency channels | Multiple layers |
---|---|---|---|---|
BMP | – | – | * | – |
DCS 1 | – | – | – | |
DCS 2 | – | – | – | |
HELIOS raster PDF | – | – | – | – |
JBIG2 | – | – | – | – |
JPEG | – | – | – | |
JPEG 2000 | * | – | * | – |
Photoshop | ||||
PICT | – | – | * | – |
PNG | – | – | – | |
Scitex CT | – | – | – | – |
TIFF | ||||
Raster EPSF | – | – | – |
* | Image format standard or Photoshop support this feature |
but ImageServer does not. |
If an image contains a clipping path, the clipping path is displayed in the Photoshop “Paths” window with its name displayed in bold font.
If an image contains an alpha or mask channel, the extra channel is displayed in the “Channels” window. Every channel in the “Channels” window has an icon of the current image. If you double-click this icon, nothing happens for process colors, but a “Channel Options” window is opened for masked, alpha, and spot channels. This “Channel Options” window displays properties of extra channels.
If an image contains a transparency channel, it is NOT displayed in the “Channels” window. Instead, the “Layers” window contains exactly one layer entry (displayed in non-italic font) and no background layer entry.
If an image contains multiple layers, the “Layers” window contains more than one layer entry.
Adobe InDesign and QuarkXPress support OPI workflows using high-resolution images with clipping paths, but they do not support OPI workflows using high-resolution images with alpha, mask or transparency channels or with multiple layers. More precisely, if ImageServer generated layouts with alpha, mask or transparency channels or with multiple layers, and these layouts are placed in Adobe InDesign, then InDesign does not generate OPI information in PostScript print output or in exported PDF documents. The same applies to QuarkXPress. Therefore ImageServer does not generate layouts with alpha, mask or transparency channels or with multiple layers.
ImageServer supports transparency via alpha, mask and transparency channels, and via multiple layers for the file formats TIFF, Photoshop native, and PNG. However, support for multiple layers in ImageServer is limited; it cannot write images with multiple layers. If an image with multiple layers is converted to another image, the multiple layer information is converted to an alpha or transparency channel. If ImageServer reads an image with multiple layers, it uses the main image only, see D.2.4 “Multiple layers”. The main image of TIFF images is always the combined image of all layers. When converting a TIFF image to another image using ImageServer, only the main image is transferred. Viewing the converted image in Photoshop, it is visually equal to the original image viewed with all layers, but the individual layers of the original image can no longer be viewed separately in Photoshop. This conversion loses information only in the sense that further editing of converted images is limited. The combination of all layers is the default view of an image in Photoshop, and the Photshop default view is the very result of an image conversion users usually expect. On the contrary, the main image of Photoshop native images may not include all visible layers. For Photoshop native images, there may also be visible losses.
Another limitation is that transparency channels can be read from Photoshop native images, but only alpha channels can be written to Photoshop native images.
ImageServer tries to preserve transparency information in
image conversions if the IgnoreMasks
preference with
default FALSE
and the UseAlpha
preference
with default TRUE
remain unaltered. If the
IgnoreMasks
preference is changed to TRUE
,
clipping paths in input images are ignored. If the
UseAlpha
preference is changed to FALSE
,
alpha, mask, and transparency channels in input images are
ignored. If an input image contains transparency information that the
output image format does not support, ImageServer tries to
convert the transparency information to a type which the
output image format does support. ImageServer does
currently not support conversions between transparency
information types that are not induced by output image
format limitations. Here are some examples of raster image
transparency support in ImageServer:
If an input image with a clipping path is converted to the PNG image format, the clipping path is converted to a transparency channel since PNG does not support clipping paths.
If you want to convert a TIFF image with a clipping
path to a TIFF image with a transparency channel, the ApplyClippath
option (see 6.1.8 “Image conversion options”) can be used.
If a PNG image with transparency channel is converted to a Photoshop Native image, the transparency channel is converted to an alpha channel.
If a TIFF or Photoshop image with an alpha channel is converted to a PNG image, the resulting PNG image does not contain transparency information. ImageServer does not convert an optional alpha channel to a mandatory transparency channel. If the input image has more than one alpha channel, the resulting PNG image does not contain transparency information because only one transparency channel can be created, and there is no way currently to select one alpha channel to convert.