![W3C](w3c_home.html)
Use of CGM as a Scalable Graphics Format
W3C NOTE 18-June-1997 (Format and markup errors corrected 31
Jan 2007)
- This version:
- http://www.w3.org/pub/WWW/TR/NOTE-cgm-970618
- Editor:
- Chris Lilley, W3C
- Author:
- Roy Platon, Rutherford Appleton
Laboratory
Status of this document
This document is a NOTE made available by the W3 Consortium for
discussion only. This indicates no endorsement of its content, nor
that the Consortium has, is, or will be allocating any resources to
the issues addressed by the NOTE.
This document shows how the Computer Graphics Metafile (CGM)
[1] meets the requirements raised in [4] and proposes solutions to come of the
problems of use raised by the general nature of the CGM format.
On 31 January 2007 this note was converted to well formed and
valid XHTML (previously claimed HTML 2.0, but invalid) and to use the current CSS
styling for W3C NOTEs. The technical content is unchanged.
There has been a long-term requirement in the World Wide Web for
an additional type of graphics object to display scalable or vector
graphics. These are particularly useful for showing technical
drawings and maps, where much detail can be lost in a raster image
and it is useful to zoom in on details, but can also be useful for
the display of other schematic data (ie histograms).
A detailed comparison of CGM against the Scalable Graphics
Requirement is given in the next section.
The authors believe that CGM fulfills most of the requirements in
[4], but there is a danger that
unconstrained use of CGM would cause problems with portability and
interoperability. This paper is an attempt to address some of the
portability issues and to recommend how to use CGM on the World
Wide Web. This will include the use of parameters to the OBJECT
tag, how add links to CGM and the possible definition of a Web
profile.
This paper does not attempt to describe CGM, which is fully
described in the ISO Standard [1] and its two
amendments [2] and [3]. The standard defines 3 versions of CGM which
have different capabilities:
- Version 1 is the original 1992 standard,
- Version 2 added elements to support GKS, the most important was
SEGMENTS.
- Version 3 added many new drawing primitives, including curves,
and more colour models.
- Version 4 is defined in [3] and adds
support for APPLICATION STRUCTURES.
This section compares CGM against the requirements raised in
[4].
- Open Specification
- The specification is published in the ISO Standard [1].
- Ready Availability to casual implementor
- W3C is discussing with ISO the availability of a HTML version
of the standard. There is also a book on CGM written by the Authors
of the standard [7].
- Extensible to cope with changing requirements
- CGM has evolved over the last few years fairly rapidly, through
the normal ISO procedures.
- Widely implemented
- There are many packages which support the import or export of
CGM version 1 and it has been adopted as the standard means of
transferring pictures in some industries.
- Reference code desirable
- Most viewers and interpreters are commercial products, but it
is hoped to make some reference code available as part of
Amaya.
- Lack of subset problems
- There are still problems with interoperability of CGMs with
different vendors. This should be reduced by the acceptance of a
Web Profile (see below).
- Vector graphics elements
- CGM has a full set of Vector drawing elements, including
POLYLINE, POLYMARKER, POLYGON SET and RECTANGLE.
- Curved elements
- CGM version 1 has CIRCULAR and ELLIPTICAL ARCS, which may be
filled as either pies or sectors. CGM Version 3 adds PARABOLIC,
HYPERBOLIC, POLYBEZIER, Non-Uniform B-Splines and Non-Uniform
Rational B-Splines to give a rich set of curve drawing
options.
- Text and Font selection
- Text and Font selection is in accordance with ISO Standards,
including a provision for UniCode characters.
- Truecolour mode
- CGM version 1 supports both Indexed and RGB colours. CGM
Version 3 additionally supports CIELAB, CIELUV and CMYK colour
definitions as well as COLOUR CALIBRATION.
- Transparency/Alpha
- CGM does not support Transparency, as it must always have a
BACKGROUND colour associated with each picture. It is possible for
an Interpreter to ignore this element in order to allow
transparency or opaqueness (via an Alpha value). It is also
possible in CGM version 3 to specify TRANSPARENCY for individual
colours within a CELL ARRAY, TILE ARRAY or PATTERN TABLE, but not
Alpha values.
- Layering, stenciling/masking
- CGM version 2 and up support FIGURES and SEGMENTS, which may be
used for stenciling/masking and layering.
- Control of line termination and mitering
- This is supported in CGM version 3.
- Levels of detail
- This is not directly supported, but can be achieved by
selection within a CGM file using the APPLICATION STRUCTURE
elements.
- Include Raster data
- Raster data can be included internally as CELL ARRAYS, which
uses an internal run length encoding. With CGM version 3 the TILE
ARRAY element supports additional compression schemes including
CCITT T4, CCITT T6 and JPEG.
- Zoom and Pan
- This possible through the Interpreter/Viewer. Each picture is
defined in its own World Coordinate system, which has an Aspect
Ratio but not absolute dimensions.
- Pick single elements
- This is possible, either by associating a PICK IDENTIFIER with
SEGMENT or by using the APPLICATION STRUCTURE elements. This does
require that the viewer can handle this.
- Switchable layers
- This is possible using SEGMENTS and/or APPLICATION STRUCTURES.
Again it needs to cooperate with the viewing software.
- Element grouping into semantic structure
- The APPLICATION STRUCTURE elements were designed to add
Metadata to a group of primitives.
- Active menus on Pick
- This is a function of the viewing software, but is is possible
to use the APPLICATION STRUCTURE PARAMETER in CGM version 4 to add
menu options to a group of primitives.
- Link to other views
- See below for details on how both internal
and external linking can be achieved.
- Link to external media
- See below for details on how both internal
and external linking can be achieved.
- Linkable from external media(#)
- See below for details on how both internal
and external linking can be achieved.
- Extractable Metadata
- There are many elements which can be used for Metadata.
CGM was registered as an Internet Media type in November 1995
[5]. Only the Binary encoding is registered
and the registration includes two required parameters,
version=n and ProfileId=name. The addition of
these 2 parameters may cause problems to some servers, as the
proper mime type for most CGMs available today should be
'image/cgm;version=1;ProfileId=None'. If the
ProfileId does not appear in the MFDESC element,as
required, it is assumed to have the value ProfileId=None.
The only standard way to add in-line CGMs to a HTML documents is
through the proposed OBJECT tag, using the DATA attribute for the
CGM file and the TYPE attribute to specify the full Mime Type.
[6]. So that the minimal tag for adding CGM
into a document would be:
<OBJECT DATA="xxx.cgm" TYPE="image/cgm;Version=1;ProfileId=None"
WIDTH=200 HEIGHT=100>
Other attributes which may be used in the OBJECT tag are ID,
DECLARE, ALIGN, BORDER, HSPACE, VSPACE, USEMAP, SHAPES. The use of
CLASSID and CODEBASE are not recommended, as these are system
dependent attributes for accessing single implementations The
OBJECT tag also has an additional PARAM element, which allows the
HTML to pass additional data to the OBJECT. To use CGM the names of
these PARAM name/value pairs need to be specified The question of
which PARAMs CGM should use is open for discussion. The following
are proposed:
- SCALABLE Yes|No
- States whether the CGM is fixed or can be looked at in detail.
This is useful to divide icons, logos etc from browseable
diagrams.
- TRANSPARENT On|Off
- States whether the CGM should be drawn on the existing
Background.
- OPAQUENESS Alpha value
- Give an Alpha value for the opaqueness of the picture in the
range 0.0 to 1.0.
- VIEWPORT topx topy botx boty
- Gives a viewport of the CGM (a part of the picture) to display.
The values are the top-left and bottom-right corners of the
sub-picture either in the World Coordinates of the CGM or as a
percentage of the picture, if the value is followed by a percent
sign (%). This facility will allow for parts of a CGM Picture to be
displayed at different scales in different places in the
document.
- HALIGN top|middle|bottom
- VALIGN left|middle|right
- Each CGM picture has a fixed aspect ratio, determined by the
VDCEXTENT element, which may not agree with the HEIGHT and WIDTH
specified by the OBJECT tag. These parameter can be used to specify
where to place the CGM within the window specified by the OBJECT
tag, ie. the gravity of the window This is different to the ALIGN
attribute to OBJECT, which is used to specify where the OBJECT is
placed in the document. The default value should be MIDDLE &
MIDDLE. This could be expressed using the PARAM tag as:
<PARAM name="halign" data value="middle">
<PARAM name="valign" data value="middle">
Note: the data attribute is optional and could be omitted.
As a CGM can contain many pictures, it is desirable to be able
to refer to individual pictures within a CGM in a URL
specification. It is proposed that the same mechanism is used as
within HTML files. therefore picture.cgm#picture%202 or
picture.cgm#2 should both be allowed to refer to the
second picture in file picture.cgm. use This will the
first in the following priority order:
- The Application Structure ID, if one exists (CGM version
4).
- The BEGPIC identifier string.
- The absolute picture number, if a number is specified.
The default value shall be #1 ie. the first picture in the
file. Note that spaces in strings need to URL encoded.
One of the advantages of CGM version 4 is that it is possible to
enclose a set of graphics primitives within an APPLICATION
STRUCTURE and attach metadata to this structure. In the Web context
this makes it easy to associate a URL with a part of a drawing,
identified as a set of primitives rather than an area on the
screen. This should be done in a standard way, so that any CGM
interpreter can handle the link. It is therefore proposed to add
the following application structure attribute types:
- LinkURL
- The data is a string containing the uncanonical URL, which may
be a full URL eg.
"http://www.w3.org/pub/graphics/cgm/example.cgm", a
relative URL eg. "test/test2.cgm", or even an application
structure within the current CGM eg. "#fillercap"
This would be in addition to the current client-side (using the
SHAPES attribute in OBJECT) and server-side image maps (using the
USEMAP attribute in OBJECT), which should still be allowed. These
have the advantage that the URL is in the HTML Document, so that if
it changes it is easier to edit than editing the CGM file. Their
disadvantage is that the shapes can only refer to areas on the
screen, not to sets of primitives. It may be difficult to describe
areas using pixels so it is suggested that area are described as
NDC (Normalized Device Coordinates) in the range 0.0 to 1.0 or
percentages of the visible area. The advantage of using APPLICATION
STRUCTUREs is that a set of CGM primitives can be grouped together
and used to define the link.
It would be preferable to use both methods by referring to the
area using the APPLICATION STRUCTURE ID in the document. This is
not possible with the existing definition of the OBJECT tag, which
only allows SHAPES to define areas.
There is a requirement within documents to draw a picture on top
of the existing background. This can be achieved with PNG by
allocating a transparent colour or using the Alpha values. The CGM
standard specifies a background colour for each picture, which is
inconsistent with this requirement. As specified above, it is proposed that a Parameter to the OBJECT
tag be used to say whether the background is taken from the CGM or
the document.
In each CGM there should be a FONTLIST element, which lists the
names of fonts used within the CGM. The CGM Interpreter has to map
these fonts to ones used internally. It is possible as part of
defining a Web Profile, that a set of
permitted fonts are defined in the Profiles definition. The Model
Profile defines the following permitted fonts [8]:
- Times-Roman
- Times-Bold
- Times-Italic
- Times-BoldItalic
- Helvetica
- Helvetica-Bold
- Helvetica-Oblique
- Helvetica-BoldOblique
- Courier
- Courier-Bold
- Courier-Oblique
- Courier-BoldOblique
- Symbol
and the Advanced Presentation and Visualization Profile allows the
following in addition:
- AvantGarde-Book
- AvantGarde-BookOblique
- AvantGarde-Demi
- AvantGarde-DemiOblique
- Bookman-Demi
- Bookman-DemiOblique
- Bookman-Light
- Bookman-LightItalic
- Helvetica-Narrow
- Helvetica-Narrow-Bold
- Helvetica-Narrow-BoldOblique
- Helvetica-Narrow-Oblique
- NewCenturySchlbk-Bold
- NewCenturySchlbk-BoldItalic
- NewCenturySchlbk-Italic
- NewCenturySchlbk-Roman
- Palatino-Bold
- Palatino-BoldItalic
- Palatino-Italic
- Palatino-Roman
- ZapfChancery-MediumItalic
- ZapfDingbats
It is prefered that CGMs use these fonts, but this is not always
possible, as they are determined by the generator. In CGM version 3
an element FONTPROPERTIES was introduced to allow a CGM to provide
a list of properties of the Fonts used in the CGM. This element
should be used as it gives a hint to the interpreter about the
relative importance of each property.
It would be preferable if the CGM Interpreter used the same
mechanism for font specification and negotiation as the Browser.
W3C is currently preparing a proposal for use of fonts on the Web,
which will include a technique for matching fonts and how to access
resources for downloading fonts when needed. This technique should
also be followed in a CGM Interpreter.
In a CGM there are about 18 attributes ie. Line type and colour,
which may be either BUNDLED or INDIVIDUAL. For INDIVIDUAL
attributes the value is explicit within the CGM, but when BUNDLED
the values depend on the media. It is proposed that BUNDLED
attributes will be determined by the current style, either set in a
style sheet or by value determined by the cascading.
Profiles were introduced in CGM version 3 to aid the
interoperability of CGM within a fixed community. It was originally
proposed that the Web used the Model Profile defined in [2], but it is intended to add APPLICATION
STRUCTURE PARAMETERS, which are not in the model profile.
CGM version 3 introduced the POLYSYMBOL primitive and a SYMBOL
LIBRARY element. These could be used to define a set of symbols for
web use. No symbols are yet proposed, but this facility could be
useful on the Web by defining a set of commonly used graphics
objects. The registration of a SYMBOL LIBRARY would become part of
a Web Profile.
It will therefore be necessary to submit a proposal for a Web
Profile. Are the any Restrictions or Additions required for Web
use?
- 1. The CGM
Standard
- Metafile for the storage and transfer of picture description
information. ISO/IEC 8632-1/4:1992
- 2. Rules for
Profiles
- CGM Amendment 1 Rules for Profiles. ISO Standard ISO/IEC
8632-1:1992/Amd.1:1994.
- 3. Application
Structures
- CGM Amendment 2 Application structure extensions. ISO Standard
ISO/IEC 8632-1:1992/Amd.2:1995.
- 4. Scalable
Graphics Requirements
- "W3C Scalable
Graphics Requirements" Chris Lilley, May 1996.
- 5. CGM Mime
type
-
CGM Mime type registration
- 6. Objects in
HTML
- "Inserting
objects into HTML" Dave Raggett, Feb 1997.
- 7. The CGM
Handbook
- "The CGM Handbook" Lofton R Henderson & Anne M Mumford,
Academic Press, 1993, ISBN 0-12-510560-0.
- 8. International
Standardized Profiles
- "International Standardized Profile for CGM" ISO Standard
ISO/IEC 12071-1:1996.
Chris Lilley
(editor)
Webmaster
Date: 1997/06/18 04:27:44
Revised $Date: 2018/10/09 13:31:53 $