Search
This issue contains the articles on the RAL GKS Validation and PV WAVE held over from the last issue. Thank you for your patience - authors and readers! Comments on PV WAVE are invited by the IUSC Graphics Working Party - please see the article and send your comments to Norman Wiseman. There is also a further article on the PV WAVE: NAG system. Other contributions in this issue include Graphical Tools available via ni-ftp, and reports on the Eurographics Working Group meeting on Visualisation in April, and the Eurographics UK Annual Conference. Thank you to all the contributors and those who supply information on events.
As some of you may be aware, our Coordinator, Anne Mumford, has been in hospital for an operation and is currently convalescing. She gave everybody a set of actions beforehand and has been controlling things by phone since she came out of hospital! Anne is recovering and should be back to work by August.
One of the major happenings has been the Visualisation Workshop and the participants have been burning the midnight oil to produce both an Introductory Guide and also a comprehensive Report of the meeting. These should be finished shortly.
The assessment of scanners has started and a shortlist of potential suppliers for both grey scale and colour versions has been sorted out. We have also been looking at video systems for displaying the contents of a workstation screen.
AGOCG is coming up to its second birthday and we have put together a report for SISC giving the set of completed tasks, those ongoing and what we are planning for the future. Quite a lot has been accomplished in the two years but we still seem to be adding items as fast as we finish others. We are also getting into the Forth Bridge problem. For example, the fast moving market means we should start looking at colour printers again quite soon.
This brief report describes the results of an evaluation of the PV Wave software package was recently carried out by the Graphics Working Party of the IUSC. The package was mounted on a DEC system, and accessed from a networked VAXstation 3100 and a PC running the Emu-Tek software connected to the central VAX 8550.
PV -Wave is a tool for data analysis, data manipulation and data display. It runs on a variety of Unix and VMS workstations from DEC, SUN, HP and SGI or can be accessed from a central multi-user system with Tektronix or Regis terminals. Output is available via Postscript, Computer Graphics Metafile (CGM) or Hewlett Packard Graphics Language (HPGL). It is driven by a series of high level commands, similar in form to those used in FORTRAN and C. Sequences of commands can be combined into procedures or macros, and whole applications can be built which can be used by non-computer literate scientists.
The command language is one of PV-Wave's strongest features, but would be difficult to use for anyone not reasonably familiar with either FORTRAN or C. It offers a comprehensive range of data input, manipulation and massage functions including curve and surface fitting, polynomial calculations and filtering.
Simple 2D display of data is basic but adequate, and includes
There is a rich image analysis and display functionality including
3D display is capable of taking advantage of any 3D firmware in the display device and includes
Output is available in industry-standard formats, although only Postscript was tested. It is generated by repeating PV -Wave commands to an output device, a window at a time. An application displaying several windows concurrently can only be rendered on a single page with a great deal of programming effort This shortcoming needs attention by the suppliers. Movie output can be obtained by storing multiple views and replaying them to a screen.
Features which are not supported are geographical projections, volume calculations and volume rendering and there is only a limited range of interpolation routines (but see future plans).
PV-Wave was particularly easy to use and a real application analysing a series of 2D images was designed and tested within a day. The same application, written from scratch in Fortran on an image processing system had taken nearly three months to reach the same level. The command language enabled data to be input, displayed, analysed, sampled and redisplayed in multiple windows very quickly. When all the sections had been designed satisfactorily it was comparatively simple to construct menus, windows and macros to enable the application to be used by a scientist with no previous experience of the software. There is a useful introductory guide, with examples which can be adapted easily, and a high overall standard of documentation. The similarity of the command language with Fortran and C will enable many scientists to get results very quickly. Non-programmers will find the version tested much more difficult to use (but see the next section).
Two new products have recently been announced but are untested. The first is a mouse driven Point and Click user interface which reduces the need to use the command language directly. The second is an interface to the NAG library which enables NAG routines to be called as if they were PV-Wave procedures. This is particularly valuable for any users who want to run the NAG routines on their data interactively and have easy to use, comprehensive display facilities to view the results.
PV -Wave is potentially a useful tool for the academic/research council community. It provides a wide range of functionality for scientists who have images or large 2D and 3D datasets and need to analyse them quickly and easily. It enables statistical analysis, extraction of subsets, samples and profiles, mathematical and matrix operations to be carried out relatively simply and the results viewed interactively using modem window based hardware. It is likely to be of use to many disciplines including scientists (physics, chemistry, environmental), engineers (mechanical, electrical, chemical) in addition to those requiring the functions normally associated with the package, image processing and computational fluid dynamics.
The Graphics Working party would like to gauge the interest in a possible CHEST deal for this package and would welcome comments and suggestions from all parts of the community, including the applicability of the software to their needs, which machine ranges it would be run on and an indication of the number of licences required at individual sites.
Early in 1991, NAG and Precision Visuals Inc (PV) announced their intention to look at joint products in the area of data analysis.
The first fruits of this collaborative exercise are now available as the PV-WAVE:NAG system. PV-WAVE:NAG is an optional enhancement to the established PV-WAVE visual data analysis system that adds a range of integrated numerical and statistical techniques provided by the NAG Workstation Library to the existing data visualisation and analysis functionality.
The PV-WAVE product is a visual data analysis system originally developed for users such as NASA. The system provides integrated facilities for the analysis of scientific and technical data including a programming language, advanced plotting capabilities, image processing and display facilities, and data animation. The additional functionality of the NAG Workstation Library is made available through enhancements to the system command language allowing NAG routines to be invoked interactively from the command line or from within PV-WAVE procedures.
Designed for use by scientists and engineers, the PV-WAVE command language has a simple, easy to learn syntax that is suitable for both interactive data exploration and analysis, and the development of data analysis and display applications. Scientific data, in the form of scalars, vectors, matrices and arrays, can be manipulated as single entities with appropriate data types being available. Defined arrays and scalars can be passed as arguments to the integrated NAG routines and results are returned to additional PV-WAVE variables. Thus, users can take advantage of the flexible data handling and rapid graphing and visualisation capabilities of the PV-WAVE system fully integrated with the numerical power of NAG routines. Complex graphs may be easily produced, often with a single command. Graphs of varying types can be mixed and annotations are easily added.
The PV-WAVE:NAG system provides an integrated interactive environment for the processing and analysis of data and information. The system provides powerful capabilities which the user can apply flexibly to a range of activities. For the PV-WAVE user, the additional NAG routines provide tested and powerful numerical and statistical techniques. For the user of NAG Fortran routines the new system provides interactive access, powerful graphing and visualisation and easy data handling facilities.
PV-WAVE:NAG is equally suited to the development of large-scale programmed systems for data investigation and presentation.
Implementations of the PVWAVE:NAG system are available for a range of machines initially including Sun 3, Sun 4, DEC/Ultrix, VAX/ Ultrix, VMS, SGI and HP 9000. For more details of the PV-WAVE system and the PV-WAVE:NAG enhancement, please contact: Precision Visuals International
The team that brought you the Uniras Training Materials have met to discuss an update.
Various pieces of work including:
are being developed.
There will also be a series of corrections made to the original materials and second editions produced. These will be kept to a minimum and will not affect page layout of the examples used in most cases.
The comments made by people who returned the questionnaire were taken into account and I would like to thank you for sending them in.
One thing that was brought to my attention was the need to have new master copies of the booklets as they wear out with use. If you would like a new copy of any of the masters please send a list together with a large addressed envelope to Joanne Barradell, Computer Centre. Loughborough University.
I hope that the new materials will be available by the start of the academic year and order forms will be sent out.
It was good to see Uniras getting some TV exposure on Horizon in June.
A programme on the Big Bang and Cold Dark Matter discussed research work at institutions including Durham and Oxford Universities. An interview was conducted with a Uniras poster in the background and the points being made were illustrated by Uniras Plots.
Just shows what selling to the academic sector can do!
Martin Leese's article PostScript for Naive Users (Graphics newsletter, issue 17) contained a couple of minor errors which I would like to correct. Firstly, it is not true that a printer will distinguish between PostScript programs and straight text, and cope with both. The conversion from text to PostScript will on may systems be handled by the printer spooling software, which is able to recognise PostScript by the presence of the %! magic cookie. However, although all files conforming to Adobe's Document Structuring Conventions will start in this way this is not required by the language itself, and many valid PostScript programs will fail this test. What is more, many sites are driving laser printers without the benefit of clever spooling software, and users at such sites will need to explicitly convert text to PostScript themselves.
Secondly, there is a typing error in the error handler: for errordic read errordict. Martin briefly discusses the problems caused by using language operators not understood by older interpreters, and suggests one solution. At the risk of repeating material already covered by Peter Halls' article in issue 16 (which I don't have to hand) I would like to suggest another. The where operator tests for the existence of a given key, and can be used in a program like this:
{
% If 'set4d' is known, we come here with a dictionary
% on the operand stack
pop % discard the dictionary
set4d % and use the operator
}{
% if the printer doesn't understand 'set4d', we come here
% This arm of the ifelse could contain code to emulate
% 4-dimensional printing
} ifelse
% Rest of program follows
I look forward to seeing more PostScript articles in future issues!
RAL GKS validation has been successfully completed. Following the on-site testing in March and the GKS Testing Services Control Board decision favouring RAL's view in a query concerning some of the validation suite's metafile tests, National Computing Centre Ltd (NCC) have issued RAL with a Certificate of Conformance and a Test Report. Valid till May 1992, the certificate states that in RAL-GKS 1.34 no deviation has been detected at GKS Level 2b.
GKS Validation consists of three parts: pre-validation, on-site testing and processing of the results.
Pre-validation is performed by the client who is requested to run all the tests and let NCC have the results at least a fortnight before the on-site testing date. These are then analysed and any deviations discussed with the client.
Providing the pre-validation has been successful, the on-site testing is carried out. A testing specialist arrives with a tape containing the validation suite sources and supervises the entire process of mounting the tape, compiling, linking and running all the tests. As in pre-validation, the client acts as test operator.
Processing of the results depends on the outcome of the on-site testing. If, as was the case in RAL GKS validation, the test suite reports any . errors, they must be attributed either to the implementation or to the test suite. In difficult cases, where either party's interpretation of the standard is questioned, disputes are referred to the GKS Test Services Control Board, a body of international GKS/Validation experts.
Validation RAL GKS prevalidation results were all good, except for the Metafile State List checks, where RAL challenged the test suite's interpretation of the standard and announced their intention to launch an official query. NCC acknowledged RAL's position.
The on-site testing took place in the week beginning March 18 1991. David Bamber and Jane Pink of NCC put RAL GKS through no less than 110 tests covering the following areas:
The configuration of RAL GKS offered for validation included drivers for Sun workstations running Sun view , PostScript printers, Tektronix 4010/4014 terminals and GKS Metafiles. These comprised some 120,000 lines of code which compares to approximately 100,000 lines of code of the validation suite. One could almost say this was a clash of the giants, lasting for three whole days and causing a lot of excitement.
RAL GKS performed admirably. The only errors appeared, as expected, in some of the metafile tests. Backed by UK GKS experts, who considered these not to be errors of RAL GKS, but rather of the validation suite itself, RAL initiated an official - dispute query. This was sent to the GKS Technical Authority in Germany, who in turn passed it on to the GKS Testing Services Control Board. Within the prescribed 30 days, the Control Board decided in RAL's favour, which enabled NCC to proclaim the validation without errors and issue RAL with a Conformance Certificate and a Test Report.
The dispute itself is an indication of the benefits of the whole exercise. This being the first GKS validation in the UK, it not only ensured RAL GKS adheres to the standard, but also tested the validation process itself - one of the reasons why AGOCG initiated and supported this project throughout.
Staff at RAL, who had put a considerable amount of effort in preparing for the validation, were much relieved at the outcome of the testing. So much so in fact that, minutes after the NCC team headed for home, they promptly disappeared into a local pub.
The celebrations did not stop there. Once in possession of a certificate, RAL organized a wet-that-document gathering which brought together a number of original developers, including those from ICL. Some interesting facts emerged: the first RAL GKS planning meeting took place late in 1981 and since then over 50 people have directly contributed to the code. Not wishing to comment on what these figures do to the software life-cycle theories, one of the party noted: We started out by saying - it's only a graphics library! Ah, those famous last words!
Netlib, the mathematical and scientific software server via e-mail is now offering an ni-ftp service. This service will provide a more efficient means of accessing netlib' s large files and those containing non-printable characters. Indeed, files larger than 0.5 Mb will only be accessible via ni-ftp, although all currently supported files via e-mail will remain. Tapes are also available for large packages. Niftp access also permits the direct transfer of binary files.
The ni-ftp database and the e-mail database overlap but are not identical. Existing small netlib files will remain as e-mail accessible only. Larger existing netlib files will be available via both access methods. New large files will be ni-ftp accessible only. This includes a directory of unix tools which is frequently updated.
All UK based users with access to the JANET network can access the netlib ni-ftp server. Unfortunately users outside the UK cannot due to the incompatible ni-ftp protocol specific to the JANET network. To transfer files, use ni-ftp
<NETLlB>path-of-file from uk.ac.ukc.harrler
The ni-ftp database is arranged as a tree structure. The top (root) level directory contains the file 00DIRECTORY (the first two characters are zeros) which holds a recursive listing of all the accessible directories and files. This file is updated daily. Every directory contains the two files 00FILES and 00README. 00FILES contains the path to that directory, a date stamp and a listing of the current directory showing all the files available at that level and their last modified date. 00README provides more detailed information about each file such as its function, author, version number etc.
A typical user making her first ni-ftp access to netlib might take the 00DIRECTORY to see all the files contained in the database. Using ni-ftp
NETLIB>00DIRECTORY from uk.ac.ukc.harrler
After receiving this file she may decide to examine a particular directory, raltex say. Using ni-ftp with the file path
<NETLIB>unlx/raltex/00FILES from uk.ac.ukc.harrler
would give her a list of files in that directory, and
<NETLIB>unix/raltex/00README from uk.ac.ukc.harrler
would give detailed information about each file in the raltex directory. After receiving this file she may decide that tgrind.tar.Z is the file she requires. Using ni-ftp
<NETLIB>unlx/raltex/tgrind.tar.Z from uk.ac.ukc.harrler
transfers the required file.
A frequent user of netlib may well know exactly which file she wants and request that file without any preliminary information or she may request the 00FILES for that directory to see if any of the files have been updated since she last obtained them.
Directories currently available via ni-ftp include the following graphical tools:
This was the third meeting and second workshop, of the Eurographics Working Group on Scientific Visualisation in Computing. The workshop aims being to provide an exchange platform for people or groups working the field of visualisation and associated application domains.
The meeting was well attended (approximately 50 people, representing nine countries) and included a good mix of industrial, research and academic institutions. Application areas covered by the delegates was very wide and included fluid dynamics, architecture, general support. mathematics, space research, and pure research (e.g DEC, IBM). The format of the Workshop was a good mix of plenary and parallel sessions with ample time for discussion.
Invited speakers were James Clark of Silicon Graphics Inc and Arie Kaufman of New York State University. Dr Clark gave a fascinating insight into the development of hardware systems for Scientific Visualisation. This was obviously based on his experiences with the development of SGI hardware but was not a Silicon Graphics sales pitch. Professor Kaufman gave an equally interesting talk and superb, but somewhat disturbing, video illustrating amongst other things the peeling of a human head (in real time)! A number of other excellent papers were presented by workshop attendees, these included animation and rendering techniques, application orientated work and visualisation environments.
A number of working groups have been set up with the aim of furthering initial ideas beyond discussions at the Workshop itself. These include animation, sample data collection! distribution, Flow Visualisation and Reference Model work. A further workshop is to be planned for next Spring. Anyone wishing to join these working Groups or requiring more information about the Working Group itself is welcome to contact me.
This year's EG UK Conference was hosted by the University of Sheffield on the 10th, 11th and 12th of April. The conference format followed that adopted last year at Bath with the first day of the conference being devoted to tutorial sessions and the major part of the second and third days being run in parallel sessions. As in previous years, an important feature of the parallel sessions were the many speakers put forward by the seven participating Special Interest Groups.
This year the Conference themes were Computer Imaging and Image Processing and Computer Graphics and Sports Science. The latter theme linking with the World Student Games which are being held in Sheffield in July. A total of 113 delegates attended the Conference and to quote a phase a good time was had by all. The Conference organisation proved to be excellent and ran smoothly throughout (above the surface anyway!).
Three tutorials were offered for the first day, the sporting theme was taken up by Fred Yeadon, Andrew Harrison, Roger Bartlett, Carl Payton and Derek Littler with Computer Graphics Applications in Sports Biomechanics in Britain, the second conference theme was taken up by Simon Thorpe with Image Processing by the Human Visual System and Henri Gouraud and Mike Stapleton gave a tutorial on PHIGS PLUS aimed at those already familiar with PHIGS.
The main conference included three plenary sessions. Thursday opened with Computer Simulation of Twisting Somersaults by Fred Yeadon and concluded with Henri Gouraud talking on Mapsketch: A New Drawing Paradigm Based on Planar Maps. Friday (and the conference) concluded with John Woodwark on Have we got all the shape representations we need for graphics?.
The parallel sessions covered a variety of topics, mostly but not exclusively within the themes of the conference. Subjects ranged from foot pressure measurements to neutron scattering data, from image processing standards to the automatic inspection of socks and from image processing in abstract art to the recovery of geometry from images. Colour models and data compression also featured, as did reports from the AGOCG workshop on Scientific Visualisation and SIGGRAPH 1990.
As has become the norm, two social events were on offer. On Wednesday evening our buffet was served, and eaten, among the exhibits of the Kelham Island Industrial Museum. On Thursday evening an excellent conference dinner was served to us in Earnshaw Hall which was also our residence for the conference.
Next year the conference will be held in Edinburgh. The format will be changed and the duration reduced to just two days (14th-15th April) as a consequence of the main Eurographics event being held in the UK (Cambridge). The themes for Edinburgh will be Parallelism for Real-Time and Distributed Graphics and Scientific Visualisation. Despite the reduced length, the social programme still includes two events a visit to the Scotch Whisky Heritage Centre and a civic reception in the City Chambers accompanied by a Ghost and Ghouls tour beneath the High Street. I look forward to seeing you there.