Jump over left menu
Report on a Visit to the USA
F R A Hopgood
15th June to 29th June, 1975
Atlas Computer Laboratory
- June 16: Simon Fraser University, Vancouver
- June 18: Los Alamos
- June 20: III, Los Angeles
- June 23: Livermore
- June 25-27: SIGGRAPH, Bowling Green
1. Simon Fraser University
The flight to Vancouver from London stopped at Winnipeg. The plane decided it would not make it any further. After about an hour's delay, another plane was found and it arrived in Vancouver about an hour and a half late.
The main purpose of the visit was to give a seminar on Computer Animation at the Atlas Computer Laboratory and to talk to people in the Kinesiology Department involved with computer animation. The seminar was well attended.
The major piece of work going on at Simon Fraser is the input and animation of Labanotation, a dance language widely used in the USA. The aim is to produce a system which will allow choreographers to input and see dance notation being performed while they are composing. A secondary aim is to produce films of existing dance and ballet scores. The work is similar to that proposed by the Royal College of Art using Benesh notation.
The Department has a GT40 with a floppy disc attached to a 370/155 in the Computer Centre over a 300 baud link. The Centre's machine appears to be lightly loaded. A 3-minute job has a turn-round of less than 5 minutes. It is, therefore, quite viable to do interaction using the 370/155.
The original work has been done by two dancers, Zella Wolofsky and Iris Garland. This was a batch system with the original dance notation being coded to allow input to a FORTRAN program. Output was initially by frames to a CALCOMP plotter. The centre has a spooling system which allows the CALCOMP output to be directed to a Tektronix.
The GT40 was purchased from DEC and the floppy disc drive from another manufacturer. They are currently trying to obtain DEC DOS software and manuals privately so that they can run it on their floppy disc. Handlers have been written to allow Tektronix input and output to be simulated on the GT40. The system at the moment can take the coded Labanotation, run the program on the 370/155, and then display the results on the GT40 in real time. The dancers are currently simple stick figures and the display is changed 12 frames per second. The time from input of data to animation display on the GT40 is about 4 minutes or less.
The next stage of the work is to provide means of inputting the Labanotation symbols directly using a menu of symbols and a lightpen. The last few time periods of the score being generated is available to the artist. He has the ability to scroll backwards and forwards through the score, displaying the actual dance figure at any point if he requires it. The long term aim is to move all the software to the PDP11 so that a low-cost viable system is produced which could be made available as a commercial system.
The main systems programmer in the Group is Gerry Barrenholtz who is interested in producing a general-purpose animation system which is similar to Ron Baecker's GENESYS. the major difference seemed to be the ability to define a number of p-curves for an object's path. A ball would be constrained to move with one point in contact with a curve while a second condition could indicate the motion of some other part of the ball. The user has the ability to specify whether the constraints imposed by the p-curves will cause the object to deform in shape to accommodate them or whether a best-fit solution is required. I think that most of his facilities could be achieved in SPROGS using Sequence Lists and Index Variables.
Barenholtz has recently visited Baecker in Toronto who has just taken delivery of a Vector General display and special-purpose video equipment. he thought that another visit to this group would be worthwhile in the future. He also mentioned a system being produced at Rank-Xerox, Palo Alto by Alan Kay called SMALLTALK, this looked as though it would be a high-class interactive system using special hardware.
My stay in Vancouver concluded with Western Airlines finding out that the Rutherford Laboratory Travel Section had managed to book me on to two flights later on that did not exist. Without any trouble they managed to book me on to equivalent flights. This was all done without even asking them. I was impressed by western Airlines computer system.
The Computer Centre at Simon Fraser are interested in getting hold of a comprehensive graphics package for local use. I agreed to send Ernie Dainow a copy of the SPROGS manual and a 360 version of the system if they require it.
2. Los Alamos
I flew out to Los Alamos from Albuqueque in a small prop plane capable of holding about 12 people.
Los Alamos have two FR80s and two SD4020s. One of the SD4020 systems is Serial No 3 and is the earliest SD4020 still running. They are currently considering the purchase of a third SD4020. Apparently they can get them for virtually nothing these days. Both SD4020s and FR80s run 24 hours a day, 7 days a week. they do not have an on-site engineer.
Both FR80s and one SD4020 have colour systems. Their FR80 cameras are slightly different from ours with the colour wheel looking more as though it was added on afterwards. Both FR80s have fiche cameras. Neither of them have a hardcopy camera. users are expected to get their own hardcopy output by directing it to a Tektronix and using the hardcopy unit attached to that. They have only recently installed the fiche cameras and do not have much more experience of these than ACL does. However, they generate vast amounts of 16mm and 35mm. The four systems between them generate about 2 million frames per month of quite complex plots.
Los Alamos Colour Output: 1975
Los Alamos do not use any III equipment for film processing. they have two Houston Fearless processors for generating 16 and 35mm. Los Alamos think these give better quality output than the 5010 as well as being less costly to buy and to maintain. The maker's address is HP Photo Systems Div, 11801 West Olympic Boulevard, Los Angeles 90064.
For colour film processing they use a machine manufactured by Jamieson Film Co, 6911 Forest Park Rd, Dallas, Texas 75235. They find that the processing has to be forced by two stops to get adequate red output. the Ferranti tubes on their FR80s require several overstrikes to get adequate colour output. they gave me some samples which were hit as follows:
Red 5 Blue 8 Green 5
Note that their filter system has primary and secondary filters. Their colour output is excellent and I have a copy of a film showing the type of work being generated at Los Alamos. They have had no trouble with their filter system at all. However, just like ACL's, the clear filter is inserted after each ADVANCE FILM. I showed them some examples of ACL's colour output. they were impressed by the quality of the colour produced with only two hits and also the resolution achieved using the Litton tube. The Ferranti tube has been running for about two years in one case with no sign of wear yet.
The major problem that they have had with both machines has been the Kennedy tape decks. they have two 7-track decks on both machines and these often give trouble. The clipping levels have to be altered from time to time. They do not have the missing blocks problems experienced at ACL. This may well be due to the fact that the SD4020 simulator is used most of the time.
They have a very good quality fiche viewer capable of reading 24, 42 and 48 fiche. the viewer is easily switched between 42 and 48 while a lens change is required for 24X. The model is a Realist Vantage II, which costs $155 and is manufactured by Microform Products, Megal Drive, Menomee Falls, Wisconsin 53051. The image area is rectangular so that a full 132 character line can be seen. They are considerably better than the Kodak viewers at ACL.
Some minor hardware points noticed:
- They use rubber bands with a tag on them for holding film on reels. To get the band off you only need to pull the tag. these are made by Alliance Rubber Co, Alliance, Ohio. I have a couple of these.
- They have most of their card punches lifted up about 3 feet so that they are a comfortable height while standing up. It also ensures that a single user does not monopolise the machine.
I was quite disappointed with the amount of work done by Los Alamos on the software for the FR80. they had it written in to the contract that they would receive a complete set of source listings. even so, they seem to have modified very little. Unfortunately the contract insists that the software is not made available to a third party.
The main piece of software used is the SD4020 simulator still. This is partly to keep compatibility with the SD4020s and partly because it is faster than the Displayer on pictures involving large numbers of short vectors. They were interested in the subset of FR80 commands that we were using. they are currently negotiating with III for a simpler and faster version of the Displayer. I have agreed to send them a copy of our subset.
They have only had LoadGo for a short time and seem no further forward with it than us. A number of software points came up during the discussions. these are listed below in no particular order:
- They have also had trouble with numbering on Fiche Titles using the Displayer. their fiche used to get numbered 1, 3, 5 etc. By removing the STARTJOB command, numbering appears to work correctly.
- They have not used Graphics Art characters at all.
- They can run several tapes without getting blank frames between. They type CONTINUE at the end of each tape.
- They are currently urging III to produce a Displayer for colour output which uses double buffering.
- They have not attempted to list FR80 source on the PLS (Precision Light Source)
- They have no information on the format of the ADVANCE FILM IOT.
- They have not noticed any changes in abuttment when a filter is in place.
- they do not do any accounting on the FR80s.
- They check abutting numbers weekly.
One interesting project they have is to produce optical sound tracks directly on 16mm. One camera is being modified to accept single sprocketed film stock. They hope to generate complete soundtracks with speech by this method. the speech is not synthetic. Human speech is converted to the optical soundtrack using A to D converters. the main motivation behind the work appears to be to remove the long delay experienced when getting optical soundtracks done remotely (often takes a month or more).
Laser Film, Optical Sound: 1975
The main people in the Graphics section are:
- Ray Elliott: Head of Group
- Dave Buckner: Maintenance
- Joe Kleska: operations
- Jim George: System Software
- Ted Reed: Device Independent Software
Ted Reed and several people from Sandia are trying to define a machine independent Graphics language. they were very interested in SPROGS and I agreed to send them copies of ACL Graphics Manuals.
Some of my time was spent talking to other members of the Systems Group who were involved in Operating Systems and Language Development. The visit was cut short by high winds (45mph) causing my plane back to Albuqueque to be cancelled. I had to leave around 4.00pm to get a taxi to take me the 100 miles back!
The visit split into two main parts. the first consisted of talking to Bob Frost, Fred Senior and a number of the hardware support team about problems. The second consisted of discussions with Bob Roach and Tom McCarthy about software matters.
The refurbishing of the SD4020 camera was discussed in some detail. The major problem appears to be that III will not quote a realistic price until they have inspected the camera while ACL is unlikely to ship the camera to Los Angeles without some upper bound quotation. III suggested that it might be possible to add the hat which fits the camera to the PLS so that the SD4020 camera could be tried on the FR80. this would give III some indication of the quality of the camera so that a more realistic estimate could be made. The cost of the hat should be minimal. III agreed to send ACL a quotation.
III have not experienced the problem with abutting colour output and also abutting with no filters in position. They agreed that the filters would almost certainly cause some refraction so that it would be impossible to abut on both at the same time. Their only suggestion was that black and white output should be produced using the clear filter. Later, Bob Roach suggested that it might be possible to install some kind of switch so that abutting could be controlled in the same way as the hardcopy correction is applied. Apparently III have already produced two systems with the additional logic for hardcopy/roll camera switching that was first made available in our machine.
III assured me that their aim was to keep a complete set of spares in the UK. Monitor tubes rarely failed and they were looking into the possibility of having a second monitor in the UK as a spare. They were ready to ship one when the fault occurred. Unfortunately the new tube was found to be faulty.
There appears to be a number of different ways of solving the problem of programmers burning the PLS out by accident. III do not recommend turning the high voltage off using the switch behind the monitor panel. Their main suggestion is that the operator should turn the beam current down completely before leaving the machine. The phosphor on the monitor is the same for all customers and they did not know whether it would be possible to get a tube with a longer persistence phosphor. III suggested that the best solution was to hook a Tektronix 4010 in parallel with the monitor tube. This was a simple modification that Dave green could do and III had no objections. Bell Labs at Denver have a system set up in this way and find it a great advantage for both system debugging and during production runs. It is possible to instigate the clear screen command by hardware in the FR80. Dave Green can obtain details from III.
III had no simple solution to loading the hardcopy camera. they felt that some time could be saved if the camera was only just above the tube when it was moved into position initially. They did not recommend leaving the end panel of the cabinet off. The hardcopy camera is susceptible to light leaks and should be used in a dark compartment with all doors closed wherever possible. They were happy for Dave Green to add a fluorescent light in the bay if we felt that it would improve the situation.
It is possible to load the hardcopy camera without having power on. The rollers should be moved slightly forward so that they disengage from the drive wheel. Both rollers should be manipulated together to avoid tearing the paper.
There does not appear to be a problem with film unwinding when the camera is plugged in on any other site. III think that this may be due to the way in which the control bits have been OR'd together for the hardcopy and microfiche cameras. They will look into it.
III have had a number of complaints about the Kennedy tape decks which they have passed on to the manufacturer. There are two field changes which could be done to improve the performance in our environment. The deck has the ability to change its clipping levels on a retry. the standard logic will try different levels while it is retrying and then reverts to the base level when the record is read. This circuit has been found to be incorrect on some decks. III have another modification which, instead of reverting to the standard clipping level, attempts to read the next block at the same level as was successful for the last block. Fred senior will be made aware of all field changes that are possible before he returns to the UK.
There are hardware manuals for the Kennedy decks and these should be at ACL. There is no complete write-up for the controller. Details are given in an Appendix to the Plain 2 Maintenance Manual.
There appears to be no way of checking the address other than forcing specific addresses into them.
It is not possible for the image on the PLS and monitor to be different in any basic way.
The difference between the flags Transport Ready and Controller Ready is that the tape deck is still ready even when reading or writing to tape but not when rewinding. The Controller is ready when the tape is rewinding but is not ready if a read or write operation is taking place. No evidence could be found in the hardware logic diagrams to indicate that the EF flag ever got set. II would let us know.
The definition of the Advance Film IOT is given in the Circuit Operation Manual, Section 1, Tables 1-1 and 1-2. The major bits of interest are Retrace (move back to home position on fiche), Up and Down.
III had no real solution to the marks on film produced by the 5010 film processor. They are unsure whether it is a chemical change or which set of rollers are causing it. It was suggested that the film path through the processor should be loosened. the bowing of the film by the agitator is not a problem. The film is allowed to move across the rollers and, as long as it does not warp the film, it should not cause any problems. III suggested that samples of 35mm ad 16mm output should be sent. They wish to discover whether it is the top or bottom rollers at fault. The drying box should be between 142°F and 145°F. if this is too high it could cause marks on the film. There was no simple way of cleaning the rollers other than using a sponge. As long as this was not full of water, III did not think this would cause any trouble. They did not seem interested in providing rollers that could be taken out of the processor.
The deterioration of the background on processed film is almost certainly due to the age of the chemicals. III were unsure of the life of the chemicals being used by ACL. They suggested that we should contact the supplier. Unless the dark room is very hot (90°F and above) this should not affect the film produced.
III suggested that two sites with complex systems that it might be worth ACL visiting were the NASA Laboratory at Houston who have three FR80s and Data Dissemination Systems Inc of Los Angeles.
The response to our software problems was, on the whole, as we expected - disappointing.
The handling of tape errors was discussed in some detail. There may be a situation where rubbish in the inter-record gap could cause the next record to be ignored. Some customers have had problems in the past. III do have some code which ignores all records less than a specified length. They will add this to the next software release. The length is usually set around 10 words. Tom McCarthy confirmed that the next block after a filemark was not error checked. This had been done to allow ACL's mixed parity tapes to be read. They agreed to reinstall the error checking so that lineprinter simulation on the SD4020 simulator would not be allowed under LOADGO.
III agreed to look into the following changes:
- DATCOM in the first 100 words
- Replies to command errors under LOADGO
- Changing the default directory
- Character height = 0 with MANYUP
- Backspace with High Speed Page Print
- Incrementing Start Address
III will not do anything about the following as they require major changes:
- Double buffering for colour
- Space to expand monitor commands
III suggested that ACL should purchase a single copy of the Displayer with all the options required and create our own specific Displayer programs after making the changes required. There should be no problem in compiling this under the Version 3 Assembler.
III could not understand why we had problems in the following situations. they suggested that further evidence including tapes should be sent:
- Blank frames between parts of a multi-reel job. Using CONTINUE, this should not happen.
- III know of no programs which stop the real time clock. None of the Application programs do.
- 360 Print Program - line numbers out of place
- Fiche Displayer - frame reset (MANYUP)
A utility will soon be available for listing source files on the PLS. It should be possible to output to both hardcopy and microfiche.
Fiche title numbering should be correct in Version 3 Displayer.
III agreed that the Editor loses files. You just need to be careful in using EC and ER.
The ability to print lines greater than 120 characters and split them up so that part goes on the next line was incorporated in the 1900 print program on the 29 April. the command to vary lines to a page etc were also incorporated. Some of these facilities do get lost once the program is entered. it is not possible, therefore, to retain these facilities once the program has been redumped.
It is not possible to output the remains of the title for a fiche under LOADGO if a filemark is encountered without an ENDJOB command. It is possible that further data may be following. if the ENDJOB command is obeyed, the full title should be produced.
III confirmed that newly-defined characters will not get mirrored correctly. it is unlikely that they will do anything about it.
III provided me with a copy of their new manual describing Version 3 software.
III provided the patches to allow 40 code in the 1900 print program to either overprint or not.
4. Lawrence Radiation Laboratory, Livermore
The visit was disappointing partly because Livermore had not done very much but mainly because of the security there which made it virtually impossible to either talk to anybody or see anything.
Livermore have two STAR-100s - one is a full size machine while the other has only 512K words of storage. Both use a common media station.
There seems to have been a number of changes in the software development over the last two years. when I visited CDC in Palo Alto two years ago, CDC had just taken over the Livermore operating system and had decided to write their own compilers using the Livermore compiler LRLTRAN as a basis for the FORTRAN compiler. At the time, I thought that the CDC people were not aware of the problems involved.
Recently, CDC have returned the production of the Operating System for STAR-100 back to Livermore giving them CDC staff to work on the project. Sam Holland, the CDC person in charge of the project previously, has been fired as were a number of his staff.
Peter Dubois is in charge of the project and is now quite pleased with the state of the operating system. the mean time between software breaks is about 4 hours while hardware breaks are coming probably only one every 10 or 20 hours depending on whether Serial 1 or serial 2 is being considered. Most of the demand paging techniques used by CDC have been changed as has the file structure. It seems to be a lot similar to GEORGE now. Livermore have had virtually no memory faults and they are impressed by the 817 large paging discs which have also proved reliable.
They are getting about 55-65% of useful time to user programs and this can get as high as 90-98% on a good batch load.
After a number of problems with the transfer rate from backing store, this has gradually been increased from 1.6 to over 20 Mbits/sec.
Their estimate of the machine's performance is about a 6600 on scalar work and 3 to 3½ times a 7600 on array problems.
Hardware problems are often due to bad air conditioning. Neither machine likes the heat and, in the summer, the machines frequently break as soon as it gets hot in the afternoon.
The system is currently only being used by the local group. It is hoped to connect the second STAR-100 to the local Octupus network in July. I was not allowed to see either of the STAR-100s and the four 7600s only at a distance.
Livermore have two FR80s. Both are 8K machines and only one has a disc. Both machines have a microfiche camera while a combined camera and a 16mm camera are shared. The machines are completely loaded and they frequently have a back log of 24 hours of work. Almost all the output (80%) is microfiche.
There is a small group of about 8 people involved in Graphics programming. They seem to have a fairly open assignment to provide useful routines for users. The group is headed by Bob lee who was previously at Utah and before that with Bendix Corporation. Others in the Group are Mike Archuletta (3-D hidden line), Steve Levine (film making), Roy Keir, Carol Hunt and Nancy Storch.
Livermore seem to have had even less luck than us in getting software out of III. By some queer government regulation they are not allowed to sign the standard III software terms and they seem to get the binaries out of III's kindness. they use the FR80 Displayer quite extensively in a completely unmodified form via the tape LOADGO system. III have not sent them the disc version yet even though they have requested it several times. It appears as though they soon got disillusioned with III's software and decided not to touch it at all. Since then the machines have got very heavily loaded and there is little chance of ding any software development. They are currently generating about a million frames a month. Quite a lot of this is film (10%).
Both FR80s have three 7-tracks drives. They started off with Pertec but have had so much trouble that they have added Ampex drives. they do not seem to have any real problems with the tape software. Most of the errors cause an error message to come out. They do not get undiscovered errors.
Both systems have blue phosphor tubes with colour recording being done optically in San Francisco. One tube has lasted about three years while the other has already gone. They thought this was due to a faulty tube.
They felt it was important that the tube should be used uniformly. If a great deal of 16mm output is being produced, that area of the tube face gets burned out considerably more than the rest. This is not a problem with the 16mm output as the tube area there is degrading uniformly. However, 35mm output has a burned area with lower light output in the middle. this has become quite noticeable on one FR80. They now make a practice of drawing a complete frame of vectors across the whole raster at frequent intervals in an attempt to average out the wear on the tube more.
They are quite interested in sound track generation and do
this directly on to 35mm film. I thought the quality
was a great deal better than those done on the SD4020
here. Bob Lee is sending me a copy of the programs and also some films with computer generated sound tracks. One
of these is an educational film showing Laser Separation of
U235 and U238 while a second
generates sound depending
on the X,Y values of the point currently being output.
Livermore Laser Separation Film, Computer Generated Sound: 1975 He is quite interested in our Synthesiser output and I agreed to send him a copy of the Hash film in return.
They have a standard software package TV80LIB available on all their computers which generates output for a variety of devices via a spooling system similar to our own. It does not do anything much more than allowing a user to define his own coordinate system and store basic picture information away.
They are currently producing 42X fiche as standard and have just convinced a user meeting that it would be sensible to go over to 48X fiche. Their standard test for the fiche camera is to draw boxes on all the fiche frames, row by row and then go back and insert a plus sign across each. It can get quite bad in alignment if it is not watched.
The only non-standard piece of software they use is a Point Plot program which takes a stream of 6-bit fields which indicate the intensity of points on a raster. This runs much faster than the FR80 Displayer. the program was written at Los Alamos and they only have the binary.
After much discussion I was finally allowed to see the FR80s but not the film processors. the latter are similar to those used at Los Alamos except for one 5010 which they use for 16mm. they felt that there were a number of superior processors on the US market at much the same cost.
Both FR80s looked very old and battered. they do have a light installed in the camera bay and also a mirror on the back door so that the back of the camera can be seen from the front. They had not experienced the same problems as we had with film unwinding. They did indicate that it was important that the doors were closed before the camera was plugged in. Also, the clips that stop the take-up from unloading do not operate correctly when the camera is getting old. They have added a piece of sponge on the back of the camera door to ensure that these are well pushed in.
They had had problems with the logic boards behind the lights shorting when they got worn. They now have sheets of cardboard between each row to stop this happening.
They have a rotating pool of operators which ensures that many of the operators are uninterested in the machine and do not follow the correct procedures all the time. To simplify things, they have changed F-stops and film used until all cameras run using the same intensity settings for the tube. They always used the fastest vector drawing speed as well. the film stock used for 16mm film was Daco E.
They are currently looking at the possibility of modifying an old DD80 recorder for colour work. They have been looking at a variety of phosphors for the tube. Their main comment was that the phosphor on the Litton tube was deficient in the red but good resolution. They are currently experimenting with a P45 phosphor with an increased Red component. this should give good colour results but the resolution may be down.
Maintenance for the FR80s is provided by their own engineer. Apart from tape decks, he felt the machine was well engineered and gave little trouble. neither monitor tube had gone since they had the machines.
4.3 Molecule Program
Steve Levine has a molecule program that he obtained initially from Ken Knowlton. It displays molecules in colour with the viewport and perspective defined by the user. The molecules are displayed as solid balls with thick bonds. He is happy to let us have a copy if we write to him. It generates FR80 output via the plotting system. he has to get Knowlton's permission to release it first but this should be a formality. He has a very good film of Tetraglycerol (?). It averages at about 10,000 lines per frame.
5. SIGGRAPH 75
Bowling Green, Ohio looks as though it is everybody's idea of a small American town. It has a population of about 8000 but a University which has 18000 students. Virtually no industries exist in the town. Once the term ends, that is now, the town just about closes down. Half the shops do not even bother to open. Even then, the shopping centre, when fully open, is considerably worse than that of Wantage. As the Dean of the College said when he opened the Conference it must be a working conference as you would never have come to Bowling Green. The countryside is flat in all directions for as far as the eye can see apart from one small mountain, the 38 foot high man-made ski slope.
There were about 300 people at the Conference including most of the names in Graphics - Knowlton, Csuri, Wein etc. The overall impression is that the Conference is worth supporting. The quality of the papers was, on the whole, good and the SIGGRAPH committee members are hoping to make it an annual event of international significance. The next conference will be in Philadelphia on 6 July 1976.
Unlike last year, the Conference Proceedings were available at the start of the Conference. This has been produced as a SIGGRAPH quarterly so that the library will get a copy automatically. I have a copy which is available for borrowing. Rather than give a complete account here, I will just add my comments to the published papers.
5.1 Potpourri Session
- F R Stocker: this paper described a number of programs available on the Adage System to help a user get familiar with graphics concepts such as 3D rotation, display files etc. He showed a number of films which were very good. For example, a display tube rotating in space with a 3-dimensional object also rotating and its 2-dimensional view on the front of the tube being shown. All of course in two dimensions.
- Nelson Max: would you believe that Nelson is still trying to get that film of the sphere inverting correct? He had his first attempt in a paper he gave at the San Diego UAIDE conference five years ago, He seems to have tried every known graphics device. For example, he had wire frame models animated meshes made up of triangles and polygons with and without hidden line removal, random dots on the surface in black and white as well as colour, Utah colour output with shading and a light source etc. He flicked backwards and forwards between these representations until everybody was thoroughly confused. Even so, in five years time Nelson will really have a beautiful film of the sphere inverting if he can get somebody to fund it.
- Robbins: this paper described a number of techniques for wave form analysis. the output was produced on an Anagraph colour graphics display which is a video display which they have hooked up to a projector capable of displaying 4ft by 6ft large pictures. Very good colour output.
- Hessdorfer: this paper described a system for interactively displaying thematic maps of Germany. Poor system and pretty ordinary output.
- Meads: described a graphics operating system comparing it with and making analogies with a standard operating system. Well presented with some good ideas.
- Trambacz: an unintelligable German speaker.
5.2 Graphics Languages Session
- O'Brien: another attempt like Ira Cotton's at defining a graphics operating system. Not a lot here that is not in PIGS. It has its own language which is preprocessed by a STAGE2 program into Fortran.
- Gilloi: would you believe structured graphics now? He did not convince me that he had retained the structured concept once he introduced sub-pictures as the display file was no longer a tree then.
- Shapiro: a graphics language which is an extension of SNOBOL. Some good ideas. major interest was the pattern matching facilities such as find all the snowmen in the picture. Incredibly slow though and do people want to do picture matching of this type?
5.3 Computer Animation
- Wein: there was a lot of films we had seen already when Nestor Burtnyk came to the Laboratory. They have some good hidden line elimination using cel levels now. Apparently La Faim got nominated for an Academy Award!
- Levine: this was another showing of most of the films I had seen at Livermore together with some ideas of how colour film production might go in the future.
- Tressel: a paper by a conventional animator stating that most animation is just movements on an animation stand and not changing art work. Consequently he has fully automated an animation stand using a PDP11. he showed some so-called bad examples of computer animation which were quite good followed by some very boring examples of his own animation. He was not very convincing.
- Csuri: he had his 30-minute film which he talked over the top of - it seems to be his standard presentation technique. They have got interested in real-time production of shaded output. Starting with Archuletta's package they decided it was too slow and Al Mayer now has a new algorithm which is much faster - 13 hours reduced to 15 seconds on the same picture if you believe Csuri. Mayer's algorithm generates half-tone images in a time proportional to the number of edges rather than the square. He is writing it up at the moment. It together with the FORTRAN code should be available by the autumn. Showed some good animation of an apple and a violin.
5.4 Argonne National Laboratory
In between sessions, I met Bob Clarke of Argonne. they have an FR80 with only 4K of store. It is also unique in not having a character generator. Apparently they could not afford it when they bought the machine a few years ago.
They use the Displayer all the time and generate characters on the main machine. they are just about to upgrade their system by purchasing more core and a colour tube. They will have the second Litton tube. He agreed to let me have the spectral curves on their tube when it arrived so that we could check the differences between the two tubes.
They are interested in the work we have done on colour and accounting. It appears as though they have been thinking on similar lines for the colour - trying to define some formulae - and would like to add accounting.
They have just finished a cross-assembler in PL/I for a 360. I was a bit unclear whether this was working yet. He said he would let me have details. They have a Tektronix in place of the teletype which is on a 110 baud line to the 360 also. They load programs from the 360 via the pseudo paper tape input on the Tektronix. there is a PDP15 interface available from Tektronix.
They have bought a listing of the Displayer and are considering modifying it to allow raster scan data to be input to it. He suggested that ACL, Argonne and Los Alamos should work towards a new Displayer format. He felt that the basic Displayer commands were alright and could not necessarily be improved upon. It is only the later commands that are not so clean.
They have a 48X fiche but generate it from a different recorder. They have standardised on 48X fiche and use interchangeable 24X and 48X viewers. Bell and Howell plus one or two other manufacturers produce suitable viewers. Kodak are just bringing out a dual Ektaline viewer.
He introduced me to Richard Bertrand who would be doing most of their FR80 software. I promised to send him any relevant papers we had.
His address is Robert K Clarke, Argonne Mathematics Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439.
They were interested in any software available on a 360/195 (that is also their computer). I agreed to send them copies of manuals and possibly systems at a later date.
They have decided that they do not like the colour system defined in the Displayer of backing up the tape between different colours being output. They would rather have a system where the responsibility is on the user and change filters as required. They will also remove the clear filter insertion.
Their colour system is due for delivery in August.
5.5 Hardware Session
- Crawford: complex system for checking the design of integrated circuits.
Judice: interesting paper in the Bell labs tradition of displaying grey level output on a storage tube only capable of black and white. Basic principle is a dither matrix which is added to the signal at any particular position to randomise it. For example:
0 128 32 160 192 64 224 96 48 176 16 144 240 112 208 80
This shows the matrix to be added to the signal at 16 points on a 256 scale. If signal + value is greater than 256 then output 1 else output 0. thus 0 to 16 points are output depending on intensity.
Had some nice algorithms for pseudo changing contrast.
- Staudhammer: this was a beautiful video colour display generating images every 8 seconds. he showed some film of a person doing graphics design. he also showed some interesting film where colour was used as a fourth dimension (the phase angle). Colour tends to be good if the extra dimension is circular.
5.6 Large System Section
- Jarvis: large complex system aiming to give you quick response to an interactive request to view a complex picture, 10000 rectangles. His examples were very complex circuit layouts. Tended to block the main picture into sections with quite a lot of overlap so that most requests always fell in only one partition and therefore less disc accesses.
- Stephenson: the topic of this paper was the display of FE calculation results. he had examples of Utah coloured objects made up of a number of different materials which you could split and see the internal formats. You could tell whether it was hollow or not and the metal distribution inside. Very good output. A 5 minute film had taken 30 hours of computation on the Utah system.
- Schneider: awful social sciences paper - dull and mundane.
5.7 General Applications Session
- Cardman: an interesting paper talking about keeping the timing right in a real-time display. the person works with Csuri. Problem is that you don't know how long it is going to take to generate the next frame but you need to know the time instance you want before you start.
- Smollar: random music generating random squares with bugs!
- Gillenson: given a photograph the system allowed anybody to produce an identi-kit picture. Apparently it didn't work if you were remembering the face so never saw the point of the paper but interesting.
- Mohilner: this was a student system which simulated a display processor on a microfilm recorder. Her display processor code looked almost like FR80 Displayer.
5.8 Device Independent graphics
- Davis: never arrived
- George: discussed some work whereby, if you have a function on your display which gives you back the beam position, you can find out all the display's characteristics. Does it clip etc. Interesting.
- Wright: a plotter package with a preprocessor producing code for a virtual device. the more papers I hear like this the more I think we ought to publicise the ideas in SPROGS more widely.
the next conference is likely to have quite a few of the papers printed in a special edition of Communications of the ACM.
5.9 Film Show
The conference finished with the usual film show. Details are given on the attached sheet. Probably the film that had the most impact was Wessler's human figure simulation with voice-lip synchronisation. The film I liked best was Knowlton's Crystal Growth Simulation which was excellent in both content and quality of film. I think I am biased but, on the whole, the artistic films were awful. Highly repetitive, over long and just a load of gimmicks. Lillian Schwartz's Metamorphisis was shown at the end and this still stands up well with the so-called second generation of computer artists.
The computer video show was dominated by the systems of Eta and De Fanti. These were, on the whole, much more successful. Eta's system is commercially available and is, basically, a video synthesiser with probably more functions than the one demonstrated at the Laboratory. His most interesting film was one of his wife wired up so that the manipulations and distortions were controlled by her heart beat and similar biological parameters. The piece is called The Kiss and shows the effect when she is kissed. Interesting!
De Fanti's work was more abstract. he showed a number of pieces which were made in real time with a sound track which was influencing the video display being produced. Some of these were quite inventive and I enjoyed them.
I was persuaded to show PIGS, Hash tables with Sound and Antics. They were well received especially the sound track for the Hash Film which always gets a good laugh.
A late entry was a long GE Film showing satellite operations in the 1980s. the system has improved greatly since the early GE efforts such as the Hancock Airport film. this one was quite impressive in both the quality of the animation and the complexity. Some sequences had two quite complex space ships with solar shields and the whole thing had excellent shading and highlights from the light source.
As always the show run over by a couple of hours with a large audience remaining to the end.
On the whole the trip was disappointing on the FR80 side in that little has been done by other sites and any work will have to be done by ourselves. It will be worth somebody going to Argonne later on once they have got some of their software developed and the colour FR80 installed. They seemed to be a more lively group than either of those at Los Alamos and Livermore. This is mainly because their machine is the only one that is not heavily overloaded. It was quite obvious that the large microfiche loads were the main reason for the overload. Argonne do not have this only because they have a CALCOMP on-line to the 360/195 purely for generating 48X fiche.
It is likely that 48X fiche will become a standard in the future that is widely used in the USA. All three establishments were changing or had changed. All three had combined 24X/48X fiche viewers liberally spread around.
Computer Film and Video Program - 2nd National Computer, Graphics Conference
Technical Computer Films
- The Tainted Sky Dr.Kent Wilson and the Senses Bureau, Univ.of Calif, San Diego
- An entertaining film showing how pollution data for the Los Angeles basin is computer-animated. 8 min.
- The Formation of the Solar System, Dr.M.L,Meeks, MIT Haystack Observatory 1974
- A short film, totally computer-animated, showing one theory of the evolution of our sun and planets. Made for the Smithsonian Institute. Color was added by optical printing. 2 min.
- Geographics Dr.Jerry Schneider, Univ.of Washington 1975 8 min.
- Population trends over the next 20 years for the Seattle region are displayed via computer graphics.
- Smalltalk by Kids at Xerox, Kids , Allan Kaye, Dick Shoup ,Xerox Palo Alto Research Center 1975
- Shows some results of children working with the Smalltalk language being developed at Xerox. 10 min.
- Not Just Reality Barry Wessler,Univ.of Utah 1973 6 min.
- A computer-animated narrator shows how he was created, some of his capabilities, and explains the goals of human-figure simulation by machine.
- Crystal Growth Simulation Dr,George Gilmer and Dr. Kenneth Knowlton, Bell Telephone Laboratories 1975
- The growth of artificial crystals was simulated and displayed to test for optimum conditions. 8 min.
Artistic Computer Films
- Hunger Peter Foldes, Marcelli Wein, Nestor Burtnyk, National Research Counci1 of Canada 1974
- The latest film made on the one-of-a-kind metamorphosis system at the NRC. An outstanding work of art, winner of a special prize at the Cannes Film Festival and nominated for an Academy Award. 10 min.
- Closed Mondays Will Vinton, Portland, Ore. 1974 6 min.
- Animation of the figures in the film was done in modelling clay (and isn't computer-animated). Film was selected to show the prevailing view of the computer in the film making world. Academy Award winner, 1975.
- Rainbow Pass Gary Demos, Los Angeles, Cal. 1975 7 min.
- The fourth film by a member of the '2nd generation' of computer filmmakers, artists who do their own programming. Abstract imagery includes some hand drawings. Synthesizer soundtrack.
- Blokpix Sampler Ed Manning , Leon Harmon 1975 5 min.
- Demonstration of a startling new technique for producing block, 'averaged'-type pictures from standard or abstract footage. The technique is mechanical, but closely duplicates earlier groundbreaking work done digitally. Selected to show that things need not always be done the hard way.
- Mosaic Norman McClaren, Canada 1965 6 min.
- The final non-computer film in the program, done by hand by the famous Canadian animator. Shows abstract effects that closely approximate much of the work that's being done in computer-art films.
- Space, Judson Rosebush, Syracuse University 1974 5 min.
- Another example of the work being done by the '2nd generation' of computer filmmakers. Abstract imagery includes sections from many different programs. Programmed in Fortran and colored by optical printing.
- First Fig, Larry Cuba, Los Angeles 1975 8 min.
- A first film by a third member of the '2nd generation'. The filmmaker wrote the programs for John Whitney, Sr. and the film resembles some of the latter's work. Colored by optical printing.
- Hash Tables, F R A Hopgood, ACL
- PIGS, Wade Shaw
- Antices Show Reel, Alan Kitching
Film Program assembled by Richard Speer, Computer Services, Evergreen State College.
Video Program assembled by Dan Sandin, Art Dept., Univ.of Illinois, Chicago Circle.