Jump Over Left Menu
USA Visit 1970
16 August - 3 September, 1970
- 1. Computers in Education Conference, Illinois, August 17-21
- 2. Carnegie-Mellon University, August 24
- 3. DEC, August 25
- 4. MIT, August 26
- 5. Joseph Kaye and Co, August 26
- 6. Polytechnic Institute of Brooklyn, August 27
- 7. Bell Labs, Murray Hill, August 28
- 8. IBM Yorktown Heights, August 31
- 9. ACM National Conference, September 1-3
These notes are being written as I go along. English is therefore likely to be bad and it may get repetitive in places. I flew to Chicago initially to attend the Conference on Computers in Education at the Illinois Institute of Technology. IIT is situated in a poor area of Chicago. You are advised not to move off the Campus. A policeman was shot and killed a few blocks away on the day I arrived.
1. Computers in Education Conference
The first day was mainly devoted to describing systems which allowed the student to use the computer as though it was a piece of laboratory equipment. The emphasis was on teaching calculus and other areas of mathematics and physics by getting the student to produce practical solutions to problems. The examples which came up several times were least squares fitting, quadrature, Taylor's series, ordinary differential equations. The computer problems were designed to raise other questions which would then lead naturally into another piece of teaching. The amount of work being done on using the computer in this way is quite high and well advanced. Many establishments have computers devoted entirely to this purpose.
Dartmouth College and the language BASIC were much in evidence. For example, Dartmouth have 125 local teletypes and 60 remote ones (serving 46 outside institutions) all attached to their GE 635. About 90% of students at Dartmouth learn BASIC. They run 12,000 jobs a day. They have their own X,Y plotter connected with a teletype to form a user station. The plotter is not fast but it does give the student a good terminal with graphic capabilities to use. The introduction of the computer is changing the way many courses are being taught. I do not know of much work of this type being done in the UK.
One surprising fact that came out of this day was the large number of Universities and Colleges which had no computing facilities at all. You get the impression that everyone has 360/65's and above when you visit MIT and elsewhere. A number of people at the Conference had nothing. Others had 1620's and PDP8's. I began to think in the end that British Universities have, on the whole, very good computing facilities.
This started as a continuation of Monday's work. There was one very good demonstration by Burton Fried of the University of California's Mathematical On-Line System. A storage tube display and keyboard were happily working on an acoustic coupler with the main computer at Santa Barbara on the West Coast.
The day finished with two long papers on Computer Animation. The first of these by Judah Schwartz contained about 12 very elegant and provocative films on Quantum Mechanics. The second talk by Frank Sarno of Polytechnic Institute of Brooklyn described their Computer Animation course for students and showed a number of results produced by them. Some of the films were quite impressive considering how little time the students had to generate them. The PIB Voght camera still has jitter many times worse than our own. On the other hand, Schwartz's films were really excellent in quality. They are fortunate at the Education Research Centre at MIT in having a good photographic section with animation equipment for doing optical work and titling their films. Judah Schwartz has a set of 10 Calculus films being marketed by Harper and Row which have a very pleasant blue background which was added afterwards by optical methods.
Sarno's course on Computer Animation is quite impressive. The students get taught a number of animation packages including POLYGRAPHICS, SCORES, BEFLIX, IGS, CAPER and CAFE. On the hardware side, they get to know the SC4020, SC406O, FR80 and BETA90 as well as CALCOMP type plotters. Techniques such as hidden line removal, projections, holograms, contours digitising are all discussed. He indicated that the major pitfalls for the students were:
- Trap of triviality: producing film of, say, circles moving about but for no real purpose - unmotivated.
- Animating the textbook: reproducing a piece of teaching but not looking to see if the medium requires a different method of teaching.
- Death and texts: putting up too much text. They have now stopped the student adding text unless it is absolutely essential.
- Rediscovering the wheel: not looking for new fields to explore but repeating already filmed techniques.
- You ain't seen nothing yet: allowing sequences to go on much too long.
The morning session was devoted to simulation. The student was given the ability to change the parameters in a model of some physical system. For example one system allowed the student to define a variety of optical systems involving a number of thick lenses having different refractive indexes. The student, it is hoped, will get a feeling for what the parameters mean and how they change the characteristics of the lens system. The output in this case was on a storage scope. The use of these was large throughout the conference.
A second simulation was the popular car travelling near the speed of light and space paths in unusual gravitational fields. There was one rather odd paper. A person from Watson Research Centre, San Diego, had carefully simulated circuits involving resistors and capacitors on an interactive display. The user could add a simulated oscilloscope to view the voltage etc. at points in the circuit. The expense involved in the programming and machine time must have been enormous. In the end you obtained the equivalent of an inexpensive piece of laboratory equipment. It seemed completely pointless.
The afternoon was devoted to the use of analogue computers in education. This was rather uninteresting. Everyone seemed to be using them for solving Schrodinger's equation. I felt that a few good animation films would be just as good and a lot less expensive. Some of the equipment was inexpensive though. One analogue computer cost $300 and was capable of simulating, for example, water pollution. At the other end North Carolina State University had developed a time-shared analogue computer with remote terminals! A further paper described a digital simulation of an analogue computer solving, of course, Schrodinger's equation.
Wednesday evening there was a set of papers and films made by computer animation. This lasted five hours finishing just after midnight. There was so much film around that they decided to hold another film show on the next night. Some of the films were the old standards including the Bell Lab film (remember Daisy Daisy sung by a computer) and the Los Alamos film of John Shannon's showing a sluice gate opening, droplet falling into a fluid, oscillating bridge etc. Some of the other films of interest were:
- Disk Galaxies by Hohl. NASA Showed how a galaxy in the form of a disk would evolve under a variety of forces. The frames of films consisted of thousands of dots representing stars. Very effective. Looked just like photographs of the Milky Way.
- The Packet of an Uncertain Gaussian. Goldin and Bregman. Polytechnic Institute of Brooklyn An artistic film with quite a bit of optical work on it showing the Uncertainty Principle. Very elegant.
Computer Films in Calculus. Schwartz and Scheg. M.I.T. Education Research Centre
These are ten short films of about 2-3 minutes which have been made by ERC and will be marketed by
Harper Row. Cost was about Â£10,000 counting all the optical work. The films were defined by ERC
and generated by Joseph Kaye and Co. of Boston. Everyone speaks so highly of Kaye's that I am sure
that someone should visit them sometime. The quality of film they produce is really exceptional.
These ten films include:
- Mean Value
- Sin x/x
- Taylor's Series Expansion
- Newton's Method
The films are very simple with little in the way of titles, yet clearly put over the point to be made.
- Biological Systems Two short films produced by San Diego University in a variety of colours. Produced using the new Stromberg camera on the SD4060. The colour was produced by optical superimposition afterwards. The superimposition was rock steady and the results were very striking. The films showed quite complex molecules revolving in space with all the bondings added and the different atoms in separate colours.
In this session I showed the new version of the Chance and Thermal Equilibrium film. It was well received and I had a number of people enquiring about how they could obtain a copy.
There was a large number of other films at the session. Most were interesting but quality was about the same as our test runs and not worth obtaining or looking into. The number of different people producing film seems to have gone up dramatically recently. One or two companies now provide a commercial service.
Most of Thursday was devoted to Computer Supervised Instruction. There were one or two theoretical papers which did not seem to say too much. On the other hand there have been several projects involving completely automated teaching. For example, Florida State University have courses which are completely computer controlled. The computer system tells the student what to read, what films to go and see (educational) and questions him on his replies. A complete record of every question is kept with all his right and wrong responses and how long they took. The student is able to do his work when he likes. Films are always available to be seen whenever he requires them. The statistics imply that the students do better than another group doing the same course in a conventional way. The students seemed very pleased with the flexibility of the system. One student, for example, who was on the college football team, managed to do nothing for half of the term as he was training, and then completed and passed the course in a very short period. This included getting through 11 sessions of about an hour in one day!
Production of programmed courses in quite frequent; languages already exist for defining programmed courses. The teacher need only indicate the questions and both correct and incorrect answers and the system is produced for him. Automated course production is already on its way. The fact which I found surprising is that the children who take these courses appear to do better than by using more conventional methods. I do not think, at the moment, that anyone regards the results as conclusive. It may well be the novelty or alternatively the type of person who volunteers for such courses.
The PLATO system at ITT was probably the most complex system discussed. It has several hundred low cost terminals throughout Chicago and the surrounding district. The programmed learning is used to teach schoolchildren, undergraduates, nurses and several other classes. The terminal itself consisted of a CRT with keyboard. On top of the CRT display (plasma pannel) would be superimposed a microfilm image stored on a microfiche card. The particular scene or diagram could be selected automatically from a total of 120 per card. The user had a transparent plasma surface on which he could write to input information. This could either be held in front of the screen or used like a RAND tablet.
Thursday evening was another film show which lasted until 12.00. Several films from Computer Image Corp and Boeing were shown. One very good film on the GENESYS system at Lincoln Labs. This is a language for defining animation at a CRT and is rather similar to what we shall be doing. The author, Ron Baecker is now at the National Institute of Health, Washington. At this session I showed some of the excerpts from the BBC films.
This was probably the least interesting day. The major topics discussed were basically to do with University politics and not very interesting to me. The Preliminary Proceedings of the Conference are available if anyone wants to look at them.
2. Carnegie-Mellon University
There appears to have been a number of changes here recently. With the shortage of money needed to run the 1108 and 360/67 it seemed likely that one or other computer would have to go. However, another approach was finally adopted. A large number of Computer Centre employees were fired; others resigned in anticipation and Dave Nickerson, the head of the Computer Centre, resigned. The staff has been cut quite dramatically. The Support Group has disappeared and programming projects have been cut back. A number of peripherals on the 1108 have been removed and both the 1108 and 360/67 are frequently not run at night to save expense. With these economies it now looks as though both machines will stay. The Computer Science Department's PDP10 network is still only one machine. They cannot afford a second yet. However, they are hoping to pick up a second-hand machine. Apparently there is one or two on the market. All Computer Science Department work is being done on the PDP10. An APL compiler is being written and an ALGOL compiler is being produced under contract for DEC. In addition Newell is producing an extensible language system called L* which is very low level initially (IPL V like) but can be extended to a higher level. Perlis seems quite struck by APL and intends augmenting the language once they have written it.
Dave Nickerson, the head of the Computer Centre, has opened a Computer Shop in a shopping centre in Pittsburgh. You can buy books, magnetic tapes, learn to operate a terminal, buy time on a variety of machines etc. You really can walk in and say Can I have 30 minutes on a 6600 please? He hopes to attract small companies, students and freelance workers. There has been quite a large turnover in staff since I was at Carnegie. Virtually only Newell and Perlis remain of the faculty. One or two students are still attempting to get their Ph.D's from when I was there.
I visited Maynard to find out how our PDP15 and VT15 display were progressing. The PDP15 for us was on the assembly floor but had very little hardware attached to it. They have probably 200 machines being built and tested at anyone time. They deliver about 50 a month at the moment, so I would think it is unlikely that ours will be here before October. However, the date they gave me for the PDP15 was still September. It certainly will not get here until the end of September at the earliest.
Bob Hopgood with Hire Car on way to Maynard
Digital Equipment Company, Maynard
Digital Equipment Company, Maynard
The display is in an even more primitive state. They had only four or five in the factory. The REDAC display which should have been delivered in August was still there and looked as though it still required a lot of work on it. It does not look as though the display will be here before November. They have had a number of set-backs mainly, they say, because their own Products Testing Department did not like the standard of the display and they had to bring it up to scratch. This put them back about two months. Displays have now started to be delivered but it will be quite a while until they get rid of the backlog. The displays I saw looked OK except they were flickering at much less than 8000 inches of lines drawn and lining up was not too good. Two lines which should meet were quite a bit out. They did not seem too worried about it, saying that it had not been set up correctly yet.
The random vector option is still not available and it looks as though that might not be available in November even. The display software is, however, written so that it will make use of these orders if you have the random vector option. This will mean that a certain amount of work we expected to do will now be already in the display package. I managed to get copies of the Display hardware manual and listings of both the display handler and FORTRAN package. The manual for the FORTRAN routines was still not available. The size of the display handler was about 1K (octal) and the FORTRAN package should be less than 2K (octal). Most of this software was written and about 80% debugged. One trouble they had been having was with FORTRAN text which did not keep a count of text length nor did it put a unique character at the end.
The RAND tablet was progressing well and it looked as though this might be available in January. The price would be less than $5000. The light pen tracking routine would also allow tracking using the tablet. The display tracking cross is, in fact, an octagon. The arc generator will not be started on until after the random vector is finished. A software routine will be available eventually and this will use this option if you have it on your display.
The display keyboard does, in fact, behave as a normal teletype keyboard and there will be no need to use the standard teletype for input although the system responses will initially come back on the teletype I think. A teletype interface to the display is being constructed and eventually the display and keyboard will be able to act just like a teletype. The editor is available at the moment on the display. However, it works much the same as it does on the teletype. A full scope editor is being worked on but no delivery date is available.
They confirmed that the display software would not, at present, run under the foreground/background monitor system. Additional information:
- FORTRAN display package should be released in the middle of October.
- The Editor in the middle of December.
- 21" Tube should be available in about a year's time.
- Colour tubes were being looked at. Bits did exist in the parameter word in the display file to allow colours to be selected. Distant future.
- PDP15 listings should be available by the end of November. At the moment changes still being made. Current listing of resident Monitor, MACRO and EDIT should be sent to me by air-mail.
- The faults with file-oriented magnetic tapes should be corrected by now although I could not find anyone who really knew. The person concerned was in Chicago.
- Additional software envisaged:
- 3-D Line generator
- Build for Standard MEMO's
- CALCOMP routines
- I obtained a copy of the INTERFACE manual.
4. MIT, Wednesday August 25th
Bob Conrod showed me around the computing facilities at the Computation Centre. They have a 2-floor layout similar to Atlas except that the peripherals are on the lower of the two floors. They have a 360/67,75, 40 and 7094 in use together with a large array of disk packs and drums plus an IBM data cell which looks quite interesting.
The 360/67 is apparently working well now. A new time sharing system produced at the IBM Cambridge Laboratories allows a number of users to work in what looks like a virtual 360 of any size. The system shares core and drum space plus genuine peripherals between the virtual 360s. This top level system can, for example, run MVT and OS as subsystems under which there may be a number of other jobs running. For example there is a local time sharing system which runs under OS which runs under the master time sharing system. This apparently works in practice as well as theory.
Judah Schwartz, Education Research Centre
Judah Schwartz showed me around the facilities they have at ERC. This is basically a PDP7 with four DEC tapes, a drum and a 340 display. Unfortunately they only have 8k of store. The basic machine has many additional pieces on it which make it difficult to see the PDP7 even! These include quite a number of button boxes and two camera mounts for a POLAROID camera and also an Arriflex movie camera.
The Arriflex camera is designed for conventional animation. It costs about $2OOO plus approximately $600 for the animation motor. This allows you to move the film in both directions and also operate the shutter. It could easily be interfaced to our PDP15. The camera is probably not fast enough to use on the SC4020. It has a pin-registered movement and film that was taken using it looked completely jitter free. Judah Schwartz uses it for debug runs and low quality film. He only makes use of SC4020 for production runs.
The number of films produced by ERC is quite high. Probably about 40-50 three minute films (I have a list) and most are obtainable from distributors. They are not allowed to sell the films commercially and so have to do it through an optical house who makes the profit on the prints.
The PDP7 has a slow link, 1.2k baud, to the MIT 360 but it is not used too frequently as it is too slow. Their mode of working is to do debug runs on the PDP7 and final runs using the 360. They have two graphic languages which are both less sophisticated than our own software. The first is part of their BASIC compiler and only has about five orders. It is very simple to use. The second package is more equivalent to the standard software we have on the PDP15 from DEC. I have a manual which describes most of the work that has been done at ERC.
They have not made any attempt recently to get colour films direct from the SC4020 or PDP7. In both cases there is insufficient light in the red range. While Judah Schwartz was at Livermore, they had a very good photographic department and he made a number of colour films by optical methods. There were a number of problems in the superimposition when lines crossed and special care was needed to ensure this did not happen. They tend to favour the Joseph Kaye method of adding a textured background only. They feel this is worth doing as it covers up film blemishes and is more restful than white on black.
I was able to see most of the software in action and to use some of the educational programs they have including the telegraph poles and road one.
They seemed to have had a lot of trouble with DEC tapes (several were out of use while I was there). Also they found that the DEC software had many bugs in it when they got it initially.
5. Joseph Kaye and Co. Ltd.
Sherrill Martin of Joseph Kaye met me at ERC and took me out to see their facilities. Before doing this we visited the Computation Centre again to see the camera mount used by Kaye's when they produce film on the SC4020 at MIT. They required a much more flexible camera mount than the one in the SC4020 as they wished to mount both their old Bell and Howell camera and their new Flight Dynamics camera very precisely. These are both 35mm cameras and the film produced by Joseph Kaye is easily the best computer animated film produced so far.
The camera mount fits just inside the door of the SC4020 cabinet. Holes have been added both sides and the mount is bolted to these so that the tube in its test forward position fits into a holder for it. The tube is now in a much more accessible position and there is no optical system between the tube and camera. Martin thinks that this does cause a certain amount of distortion which is removed by filming in this fashion. The mount has retaining screws which allow the distance from the tube and the position relative to the tube to be easily altered for precise alignment. A test pattern is produced on the tube face and the camera is carefully aligned over it. To make sure the academy image will be correctly positioned on the film stock. A slide is inserted run the camera and illuminated to cause an image on the tube face which can be aligned to. The mount is simple to make and I have a copy of the blue prints. It would be quite useful for increasing the illumination level on films with virtually no alteration to the SC4020.
I had a very useful discussion with Sherrill Martin concerning how they use the SC4020. Their software is basically an extension of the SCORS package. They have a number of subroutines which sit on this and, in general, seemed quite sophisticated. Their curve drawing routines all attempted to draw as few lines as possible. They start with a certain step length and attempt to double it each time (rather like a Simpson's quadrature algorithm). The condition is that the curve slope must not deviate from the line drawn by more than half a raster position. They tend to overstrike all lines once to avoid discontinuities. The film they use is KODAK 5498 RAR. There is an equivalent 16mm. They used Ektachrome in the past but found that KODAK was about a stop faster and does not helate so much. They tend to run at between F2.8 and F4 (with the tube swung out). They check the positioning and brightness of the tube each line. They use a Western Ranger 9 meter with a black cloth around it. I think they spend a lot of time on this partly because they use a number of SC4020s and other microfilm recorders in the Boston area.
Joseph Kaye's do not have any microfilm recorders themselves. They have their own cameras and take these to the various recorders to use. All their processing is done at local film laboratories. They have stopped using their old Bell and Howell camera and now use a Flight Dynamics 35mm camera, FR207. This costs about $3695. It has a single fixed pin guiding against one film edge. The movement is quite complicated to load with many sprocket wheels. It requires a modification to the aperture plate ($600) to ensure the correct positioning of the picture image. The various microfilm recorders they use all have slightly different pulse mechanisms for the camera. They find, however, that it is not too much trouble to make up connecting cables for all the leading makes. The pulse heights and widths vary but do not seem to cause any trouble. Some recorders pulse every sprocket hole while others just generate one per frame. This can apparently be sorted out by just leaving some of the pins blank so that only one pulse out of six say is used. Sherrill will let us have details of these cables if we require them. We discussed other cameras. The Stromberg camera he thinks is quite good but expensive. Another possibility is the Red Lake Laboratories LOCAM (also make HICAM) 16mm. camera.
They tend to use local film laboratories for their processing, using the High Contrast Processing Cycle at the laboratory. They would like to get their own processor (Kodak Recordak Processor, $7000). The film they use is compatible with this and can use the high drying capability.
They tend to produce only microfilm even in debug runs. This is partly economic and also because they feel that you do not get a good impression of the final film using hardcopy (MIT charge them 75 cents per frame for hardcopy). Initially they produce every 50th frame. This they project using a film strip projector. Then they produce every 8th or 9th frame and project at 2 frames/second. This allows them to see a jerky version at the correct speed. I saw examples of this and it was quite effective. Of course you need a projector capable of running at this speed. They use a KODAK ANALYST PROJECTOR which is capable of running from 2 to 24 frames per second. It is possible to buy similar projectors from La Fayette and also a Dutch company.
The new recorders recommended by them are:
- BETA COM 600 produced at Newton, Mass. This has 3Ã—3in screen and 4096Ã—4096 raster. Costs about $120,000.
- CALCOMP 890. 4096Ã—4096 with 32 brightness levels.
- FR80. Good if you can afford it. Possibly better than you need for movie work.
- SCANGRAPHICS 12000 Ã— 16000 grid.
- CALCOMP 1670.
The BETACOM and FR80 do not have hard copy output.
6. Polytechnic Institute of Brooklyn
I visited Frank Sarno at the Polytechnic Institute of Brooklyn. They have an SC4020 and 360/50. They find that they have insufficient work to run their SC4020, full time and only have one operator who comes in about three or four times a week (sometimes in the evening) to run it. Unfortunately Thursday was not one of his days so I was unable to get any details of how they operate it. They have the same 16mm camera as us but get worse results in general. All their high quality output is done on the 35mm camera which is rock steady. They used to have an external mount when they had an older 35mm camera. They now have one which fits into the SC4020. The old one was too large to get in the cabinet. They have done a whole series of film loops on Geometry with Frank Sinden of Bell Labs. These are very elegant. Theorems on angles of polyhedra, triangles in circles, the inscribed and circumscribed triangle etc.
I was able to obtain a manual of the Polygraphics system and should have a magnetic tape containing the 360 object code of the system. (Already for the 360/195!) I was unable to get a copy of the source deck as they have unfortunately lost it at the moment due to someone overwriting the disk. They only have out-of-date copies and it will be a while before they manage to sort it out. Frank Sarno does not mind us using the system but requests we do not make too many copies of the manual. He is currently turning it into a book with John Wiley and they do not want it to have too much exposure before the book comes out.
Apparently no additional work is being done on Polygraphics. The students who implemented it have left and it is difficult to get staff for a sufficient length of time to really get familiar with the system. I had an interesting discussion with Frank Sarno. We talked about our future plans and I will try to keep him in touch with these.
7. Bell Telephone Laboratories. Murray Hill
One interesting point about this trip was that I found a new way to get there. Since they stopped providing a limousine from their down-town office, it has been a bit of a problem. You can take a subway train to the Hudson terminal at the South End of Manhattan and then catch the PATH tube (Port Authority Trans Hudson) to Hoboken in New Jersey. From there you can catch a train on the Erie-Lackawanna railway to Murray Hill. The train is about 1880 variety with wicker seats. Murray Hill is really just a dropping off point and you are then walking distance from Bell Labs (just; about half a mile). It takes about 1.5 hours if you don't have to wait too long for trains. I have timetables if anyone is going in the near future.
There is a lot of construction work going on at Bell Labs. It is acquiring at least two very large new buildings. The main people I went to see were Kenneth Knowlton, David Deutsch and Frank Sinden. Frank Sinden, who made the Force, Mass and Motion film, does not do any animation work now apart from the Geometry series which he does in his spare time. We spent most of the morning having a film show at which they showed me their latest films and vice versa. Knowlton had just finished a rather way-out Modern Art film with an artist (Schwartz). It is a mixture of BEFLIX animation with cuts to conventional animation. The sound track was also computer generated (also way-out). The BEFLIX system in FORTRAN IV looks as though it is in a fairly static condition and well documented. It has been copyrighted by Bell Labs and can be leased from them. It is possible in the USA to obtain it free for educational purposes. Knowlton was not sure how we stood. I wrote him a formal letter requesting details of what it would cost us and he has passed it on to their copyright department. If we can get it free or at a nominal cost, I would like to get it working on one of our computers (probably the 1906A). It is capable of doing much that is not possible to do by other packages. I wonder how much it will cost in computer time. For example one frame of output he showed me required nearly a full IBM tape and took 15 minutes to plot on the SC4060.
The SC4060 is only used to produce microfilm and this is usually 35mm. They have a Stromberg pin registered camera. They have their own developer which looks rather like our own. Hard copy output is either obtained by an off-line microfilm to paper Xerox copier or alternatively by an on-line Xerox printer. The latter is very fast. We generated some frames on a 344 display and by the time we had walked up a flight of stairs the output was in Deutsch's pigeon hole. There are two of these on-line to the GE635.
The computing facilities seem a lot worse than they were the last time I was there. The fast queue turn-round time on batch jobs was up to seven hours. Teletype jobs, of course, do have a better turn round. The GRAPHIC 2 system consisting of a PDP9 and 344 display connected to the GE635 over a telephone line looks a good system. Response from the GE635 was not always good. It looked as though the scheduler was not getting in to look at our job which was supposed to have extra high priority. Also the job got in a core jam at one point and was moved out to a drum. The system is interesting in as far as all work in the PDP9 is duplicated in the GE635. All light pen hits, for example, which are serviced in the PDP9 are also sent to the GE635. Consequently the GE635 file is always up-to-date and display file is only passed one way from 635 to PDP9. They have an interesting economic modelling system which is a research tool at the moment but looks as though it could eventually be a useful management tool. The basic software package is called GRIN and I have a manual of the system.
One other system under development is one involving a PDP8 with 4k of store, a 611 storage tube and a teletype. The system looks like a teletype to the GE635 and is quite flexible. The output from the 635 is either directed to the teletype or display. The software package allows quite complex pictures with sub-pictures to be displayed. It is quite impressive considering the size of the system. It looks rather similar to the Imperial project we are funding.
The SC4060 seemed to have a number of problems. The intensity of lines varies depending on where in the buffer the order sits. Thus if you shade an area from one side to another, the lines start dark and get lighter; then dark again and so on. It really looks awful. They should be getting an FR80 possibly to replace the SC4060.
My impression is that there is not as much work going on in computer animation at Bell Labs as I expected.
8. IBM Yorktown Heights
This can be reached by taking a train from Grand Central to Croton-Harmon and then getting a taxi. It takes about 90 minutes.
The graphics work at Yorktown Heights is being curtailed considerably. The people involved have been put on to producing a 1130 disc based operating system. Camille Volence showed me her animation system for generating the big display board animations that you see at many American baseball stadiums. The system is quite simple to use and allows the animator four levels of picture which can be moved relative to each other. It is possible to do shading in much the same way as you would shade a picture. There is quite a course mesh so that any kind of motion across an area will give the desired result. It does not do hidden line elimination. This tends not to be too much of a problem as the animation is simple and any elimination required can be done manually. There is a rubout facility. The work has now been stopped and the extensions envisaged by Camille Volence do not look as though they will be implemented.
Franklin Gracer has an animation system on the 2250 for doing inbetweening. The user may draw the start and finish picture. He then indicates either the path between the two or the turning or both. The intermediate positions are calculated by first ensuring that the two pictures have equal numbers of line segments. Unlike our own system, if the number of lines are different, the smaller number of lines is extended by re-defining a completely new set of lines at equal intervals. The method assumes that all lines are of equal length and works only with a large number of lines per picture. Pictures are normally input using a tablet and the input routine ensures that equal length lines are stored. Again this work has been stopped.
The third major piece of work is that of Art Appel. He is working on general three dimensional shading algorithms and has some very good pictures of a human head made up of a large number of connected flat faces. The output is on the SC4020 using the plotting dot. Their machine has a modification allowing characters to be plotted at 16 different intensity levels. Of these, about 14 are useful. Appel was interested in Paul Nelson's still camera as this might allow him to get the same number of intensity levels using vectors rather than dots. This way he thinks he could save a considerable amount of computer time. At the moment he only gets one frame per tape approximately. The generation of this takes 5 to 10 minutes on the 360/91.
The SC4020 is, after four months, up to 3 shift working. About 50% of this time is cine work. They have two cameras the same as ours plus a pin-registered 16mm camera supplied by Stromberg. This gives good jitter free output. They have a Stromberg processor allowing both black on white and white on black developing. They had a few problems with their new camera. It is larger than the old one and a hole had to be cut in the top of the SC4020 cabinet! They have three engineers of their own that Stromberg trained for them. Having had the SC4020 only a short time, they were not too familiar with it and I was able to give them more information than I got back.
9. ACM National Conference
This was titled the unconventional convention and was supposed to be a forward look into the 1970s. It was unconventional firstly because the number of people was much smaller than usual. The economic recession in the USA has really hit the American computer industry. Most research projects using Defence money have been cut back a great deal and government and industry funding of computer projects is much less now than it was six months ago. There is no money available to send people to Conferences. A large number of people are out of work. Several people came up to me asking if it was possible to get a post in the UK for a year or two. The SHARE conference in Canada recently had only 50% of the delegates that were expected.
On the whole there were too many general sessions. There has been a conscious effort to make at least part of the conference understandable to a layman and special sessions have been added for this purpose. The Graphics session I was a panelist on was well attended and the questioning was good. Even so no earth shattering facts came to light.
I met Kah Liang of the National Film Board of Canada and spent a couple of hours talking to him. They have been attempting to get an SC4020 to enhance their graphic facilities but without success. The Film Board does not seem to recognise the merit of the work there even though it is some of the best in computer animation from the artistic and educational side. He is looking around for a junk SC4020 that they can get to work on. Liang is chairing a complete session on Canadian animation at UAIDE. He would like either Paul Nelson or myself to visit Montreal sometime to see their facilities and suggested that perhaps Paul could do this after UAIDE. He would prefer closer co-operation with us if it could be achieved and was wondering if it would be possible to run jobs on our SC4020. The Film Board has recently acquired a new Chairman and Liang is wondering whether or not the educational work done by the Film Board will be curtailed.